Immune microenvironment of intimal sarcomas: Adaptive immune resistance with potential therapeutic implications

Abstract

Objectives

Intimal sarcomas are rare, aggressive neoplasms that arise from large blood vessels. Characterization of the tumor immune microenvironment may suggest new treatment strategies.

Methods

Seventeen specimens from 7 patients were labeled by immunohistochemistry for programmed cell death 1 ligand 1 (PD-L1), CD45, CD8, CD4, FOXP3, CD20, CD68, and LAG3. The immune cell density was scored as a percentage of the tumor area (1+ [<5%], 2+ [5%-50%], 3+ [>50%]); PD-L1 expression was scored on tumor cells and on intratumoral immune cells. Immune marker density was quantified using image analysis software.

Results

All intimal sarcomas showed immune cell infiltration (41% were 1+, 53% were 2+, 6% were 3+). Tumor and immune cell PD-L1 labeling was seen in 35% and 76% of cases, respectively; PD-L1+ intimal sarcomas had higher CD45+, CD8+, FOXP3+, CD68+, and leukocyte activation gene 3 (LAG3)+ cell densities (P ≤ .01). Similarly, PD-L1 expression on immune cells correlated with higher densities of CD8+ and FOXP3+ cells (P < .04). Higher LAG3+ cell density correlated with higher CD68+ cell density and necrosis (P < .05). One patient with prolonged survival had the highest immune cell density and PD-L1 expression.

Conclusions

These data show that intimal sarcomas have an active tumor microenvironment with an adaptive pattern of PD-L1 expression. Our results suggest that immunotherapy may be an effective treatment option.

Key Points

• Intimal sarcomas have an active tumor immune microenvironment.

• Programmed cell death 1 ligand 1 shows an adaptive pattern of expression.

• Immunomodulatory therapy may be a treatment option for this aggressive neoplasm.

INTRODUCTION

Immunomodulatory therapy has revolutionized the field of oncology and become a cornerstone of treatment for some solid tumors, including melanoma. Mechanistically, this treatment modality involves blocking 1 or more molecules involved in the immune checkpoint, thereby allowing the patient’s own immune system to recognize tumor cells. Two targetable components of the immune checkpoint are cytotoxic T-lymphocyte–associated protein 4 (CTLA-4) and the programmed cell death 1 protein (PD-1)/programmed cell death 1 ligand 1 (PD-L1) axis.¹ Future therapies may involve targeting other immune checkpoint molecules, such as leukocyte activation gene 3 (LAG3), which plays a role in T-cell exhaustion.^2,3

Immunomodulatory therapy has shown some efficacy in treating sarcomas. The SARC028 study, a phase 2 trial of pembrolizumab (an anti–PD-1), showed an 18% objective response rate in soft tissue sarcomas.⁴ An expansion cohort showed an objective response rate of 23% in undifferentiated pleomorphic sarcoma (UPS) and 10% in dedifferentiated liposarcoma, with median progression-free survival of 3 months and 2 months, respectively.⁵ The Alliance trial, a phase 2 trial of nivolumab (an anti–PD-1) with or without ipilimumab (an anti–CTLA-4) for treatment of sarcomas, showed a 5% response rate for nivolumab alone and a 16% response rate with combination treatment, with responses seen in UPS, leiomyosarcoma, myxofibrosarcoma, and angiosarcoma.⁶ Additional clinical trials are underway to evaluate combinatorial immunotherapeutic approaches.⁷

Features of the immune tumor microenvironment (TME), including CD8+ T-cell density and PD-L1 expression, can serve as biomarkers for response to immunomodulatory therapy.⁸ In the previously referenced SARC028 trial, response to pembrolizumab was correlated with higher density of cytotoxic and regulatory T cells and increased expression of PD-L1 on tumor cells and tumor-associated macrophages (TAMs).⁹ Genomically complex sarcomas, including UPS, have greater lymphocyte infiltration than translocation-associated sarcomas.¹⁰ Further, this higher immune cell density is associated with expression of PD-1 on lymphocytes and of PD-L1 on tumor cells.¹¹ Expression of the recently described checkpoint molecule LAG3 has also been described in sarcomas, suggesting a component of T-cell exhaustion in these tumors.¹² Tumor-associated macrophages have recently been described as a prominent component of the TME in prototypical sarcomas and are thought to contribute to the immunosuppressive milieu.¹³

Although significant progress has been made on understanding the TME of sarcomas as a group, less common histologic types remain to be studied. Intimal sarcomas are rare, malignant soft tissue tumors that can resemble UPS histologically but characteristically arise from the intima of large blood vessels and show nuclear expression of mouse double minute 2 homolog (MDM2, positive in up to 70% of cases).¹⁴ On a molecular basis, intimal sarcomas are characterized by amplifications of MDM2, EGFR, PDGFRA, and KIT receptor tyrosine kinase.^15,16 Secondary to their origin from the intima of large vessels, intimal sarcomas can present with signs of vascular obstruction. The primary treatment modality is surgery, although complete resection is often not possible. Chemotherapy and radiation have questionable efficacy, and imatinib (a tyrosine kinase inhibitor) is not effective.¹⁶

Given the lack of effective treatments for intimal sarcoma, there is a need for development of other therapeutic modalities. Although our knowledge of the TME and role of immunomodulatory therapy in sarcomas is expanding, no studies have addressed the TME in intimal sarcomas, a rare subtype in need of additional treatment options. Here, we profile the composition of immune cells along with various immune checkpoint proteins in intimal sarcomas to assess for the possibility of immune-based therapeutic approaches in these tumors.

METHODS

Case Selection and Review

This study was approved by our institutional review board (IRB00190241). A search of the electronic health record was performed for all cases of intimal sarcoma diagnosed at our institution between 1985 and 2019. Seventeen specimens with available tissue blocks were identified from 7 patients with pulmonary artery intimal sarcoma. H&E-stained slides for each case were reviewed by 2 pathologists, noting histologic features such as tumor morphology, mitotic count, necrosis, and lymphoid aggregates. Representative blocks were identified for further immunohistochemical studies. Clinical information, including demographics, treatment, and follow-up data, was obtained from the electronic medical record for each patient.

Immunohistochemistry

All specimens were stained by immunohistochemistry (IHC) on an automated Bond RX platform (Leia Biosystems) for CD45 (clone 2B11+PD7/26, Agilent/Dako Technologies), CD8 (clone 4B11, Leica Biosystems), CD4 (clone EP204, Millipore Sigma), FOXP3 (clone 236A/E7, Abcam), CD20 (clone L26, Leica Biosystems), and CD68 (clone KP1, Agilent/Dako Technologies) to identify immune cell subsets.

We performed PD-L1 IHC using a manual IHC assay (clone SP142 [0.096 µg/mL], Spring Bioscience), as previously described.¹⁷ LAG3 IHC was also performed using a manual IHC assay (clone 17B4 [0.1 µg/mL], LifeSpan BioSciences), as previously described.¹⁸ Additional immunohistochemical protocol information is available in Supplemental Table 1 (all supplementary material is available at American Journal of Clinical Pathology online).

Scoring of Immune Cell Infiltration and PD-L1 Expression

Immune cell density was scored qualitatively on H&E-stained slides, based on the percentage of the tumor area lymphocytes occupied,¹⁹ and was categorized as rare (1+ [<5% of tumor area]), moderate (2+ [5%-50% of tumor area]), or brisk (3+ [>50% of tumor area]). We scored PD-L1 expression based on membranous staining, with more than 1% labeling on tumor cells considered positive. We scored PD-L1 expression on tumor-infiltrating immune cells based on the percentage of immune cells, with positive staining noted as none (0%), focal (<5%), moderate (5%-50%), or diffuse (>50%).

Immune Cell Quantification and Image Analysis

Immunostained slides were scanned at ×20 (0.49 µm/pixel) on a Hamamatsu NanoZoomer XR digital slide scanner. Slides were annotated for representative tumor areas using HALO (Indica Labs), excluding areas of necrosis. Where specimens showed areas of interface between tumor and adjacent normal tissue, the interface was defined with a line, and additional areas within 200 µm on either side of the interface line were annotated as the tumor-normal interface. Cells expressing CD45, CD8, CD4, FOXP3, CD20, CD68, and LAG3 were quantified with the HALO Immune Cell Module using best-fit parameters. Density of cells expressing each marker was calculated as cells/mm².

Statistical Analysis

Statistical analysis was performed using IBM SPSS statistics software, version 25. Median values were used when combining data for several specimens from the same patient. Fisher exact test was used for categorical variables, the Mann-Whitney U test and Kruskal-Wallis H test were used for unpaired continuous variables, and the Wilcoxon signed rank test for paired continuous variables. P ≤ .05 values were considered statistically significant.

RESULTS

Clinicopathologic Characteristics of Intimal Sarcomas

The clinicopathologic features of the 7-patient cohort are summarized in Table 1. The median patient age was 62 years (mean, 64 years [range, 46-83 years]). The cohort was predominantly female (71%) and White (71%). All tumors were located in the pulmonary artery, with a median size of 11.0 cm in greatest dimension (mean, 10.3 cm [range, 5.5-15.0 cm]). All patients were treated surgically for the primary tumor, with initial procedures ranging from pulmonary artery thromboendarterectomy for presumed pulmonary embolus to pneumonectomy. Three patients (43%) required additional surgical procedures for recurrent tumor, including pulmonary artery excision in 2 patients and a pneumonectomy in 1 patient who had initially undergone a more limited surgical procedure. Two patients (29%) received adjuvant chemotherapy and radiation, and 1 patient (14%) received adjuvant chemotherapy without radiation. Three patients (43%) developed metastatic disease during the follow-up period, with metastases occurring in the brain, femur, contralateral lung, and elsewhere in the thorax. Four patients (57%) died of their disease after a median follow-up interval of 12 months.

TABLE 1

Demographics, tumor size, treatment, and clinical outcome

Characteristic	Value
Age, mean (range), y	62 (46-83)
Sex, No. (%)
Female	5 (71)
Male	2 (29)
Race, No. (%)
Black	2 (29)
White	5 (71)
Tumor size, median (range), cm	11.0 (5.5-15.0)
Initial surgery, No. (%)
Pneumonectomy	3 (43)
Lung lobectomy	1 (14)
Lung wedge resection	1 (14)
Pulmonary artery excision	1 (14)
Pulmonary artery endarterectomy	1 (14)
Adjuvant treatment, No. (%)
None	4 (57)
Chemotherapy alone	1 (14)
Chemotherapy and radiation	2 (29)
Recurrence, No. (%)
No	4 (57)
Yes	3 (43)
Metastases, No. (%)
No	4 (57)
Yes	3 (43)
Death during follow-up, No. (%)
No	3 (43)
Yes	4 (57)

Seventeen evaluable surgical specimens were included in our study, with 1 to 5 samples obtained from each patient in the previously described cohort. Additional details, including specimen numbers, sites, and the timeline for each patient, are available in Supplemental Table 2. Fifteen samples were from pulmonary artery tumors (10 resections, 4 biopsies, 1 autopsy specimen), and 2 were from tumors that had metastasized to the brain (1 biopsy, 1 autopsy specimen). Histologically, the tumors consisted of spindled to epithelioid cells, with varying degrees of pleomorphism and mitotic activity, and frequently showed areas of necrosis. Predominantly spindled cell morphology was seen in 9 cases (53%), predominantly epithelioid morphology in 1 case (6%), and a combination of the 2 morphologies in 7 cases (41%). Significant pleomorphism was seen in 8 cases (47%) and was seen more frequently in tumors with epithelioid or mixed morphology. One primary tumor showed prominent pseudovascular spaces. The mitotic rate was 0-9/10 high-power fields (HPFs) in 13 cases (76%), 10-19/10 HPFs in 1 case (6%), and more than 19/10 HPFs in 3 cases (18%). Necrosis was noted in the majority of specimens (14 cases [82%]).

Four cases had previously undergone immunohistochemical testing for MDM2, and all were positive. Additional positive immunostains included SMA (2 cases), focal S100 (2 cases), focal CD31 (2 cases), Cam5.2 (1 case), and patchy staining for PR (1 case). Molecular testing had been performed previously in 1 case and showed EGFR duplication and MDM2 amplification, characteristic of intimal sarcoma.

Characterization of Tumor-Infiltrating Immune Cells

Infiltrating immune cells were present within the tumor to some degree in all cases and at the interface between tumor and normal tissue where evaluable. Infiltrating immune cell density was classified as rare (1+) in 7 cases (41%), moderate (2+) in 9 cases (53%), and brisk (3+) in 1 case (6%) Table 2 and Figure 1. The majority of cases showed higher densities of CD45+ cells and all individual immune cell subsets at the interface between tumor and normal than within the tumor, although these differences were only statistically significant for CD45, CD20, and FOXP3 Table 3. Higher infiltrating immune cell density was associated with an increased intratumoral density of all lymphocyte subsets, including CD8+ T cells (P = .002), CD4+ T cells (P = .008), FOXP3+ T regulatory cells (P = .017), and CD20+ B cells (P = .005) Table 2. There were no associations, however, between overall infiltrating immune cell density and the density of specific immune cell subsets at the interface between tumor and normal tissue Table 2. Lymphoid aggregates without germinal centers were seen in a minority of the cases (2/17 [12%]) Table 2 and Figure 1. The presence of lymphoid aggregates was associated with significantly higher densities of intratumoral CD20+ cells (P = .029) and FOXP3+ cells (P = .029) and a higher CD8:CD68 ratio (P = .015) Table 2. Despite increased FOXP3+ cells, there were concomitant increases in CD8+ cells, with stable CD8:FOXP3 ratios between cases with and without lymphoid aggregates. Conversely, cases with necrosis had a higher density of intratumoral CD68+ cells (P = .003) and a lower CD8:CD68 ratio (P = .047) Table 2. Dominant morphologic pattern and mitotic count were not associated with overall immune infiltration or with any individual immune cell subset density (Supplemental Table 3). No statistically different patterns in immune cell infiltration were identified between tumors that were primary or metastatic in the entire group (Supplemental Table 3) or in individual patients. Although the degree of immune cell infiltration varied over time in patients who had multiple samples, no clear trend was identified (Supplemental Figure 1).

TABLE 2

Median correlations between immune cell density, histologic features, and checkpoint protein expression

	Immune cell density			Lymphoid aggregates		Necrosis		Tumor PD-L1		Immune cell PD-L1				LAG3 density
	1+	2+	3+	‒	+	‒	+	‒	+	None	Focal	Moderate	Diffuse	Low	High
Total specimens, No.	7	9	1	15	2	3	14	11	6	4	6	5	2	9	8
Immune cell density, No. (%)
1+				7 (47)	0 (0)	2 (67)	5 (36)	7 (64)	0 (0)	3 (75)	3 (50)	1 (20)	0 (0)	5 (56)	2 (25)
2+				8 (53)	1 (50)	1 (33)	8 (57)	4 (36)	5 (83)	1 (25)	3 (50)	4 (80)	1 (50)	4 (44)	5 (63)
3+				0 (0)	1 (50)	0 (0)	1 (7)	0 (0)	1 (17)a	0 (0)	0 (0)	0 (0)	1 (50)	0 (0)	1 (13)
Intratumoral density, cells/mm2
CD45	36	606	2399a	167	1568	32	309	108	723a	34	131	737	1396	108	500
CD8	26	213	1201a	63	724	50	129	39	241a	34	74	213	1057a	50	180
CD4	70	481	2013a	195	1286	195	225	163	516	116	116	481	1362	163	430
FOXP3	9	57	328a	24	237a	57	27	16	70a	13	15	62	279a	21	59
CD20	4	11	247a	7	180a	6	9	4	15	5	5	19	129	7	10
CD68	111	326	242a	210	136	20	234a	146	434a	66	177	398	504	142	320a
LAG3	15	73	63	17	37	10	31a	14	154a	15	15	126	123	–	–
CD8:FOXP3	2.92	3.80	3.66	3.01	2.68	1.70	3.33	2.92	3.73	4.10	3.41	2.60	3.82	2.92	3.82
CD8:CD68	0.18	0.61	4.97	0.37	6.56a	4.93	0.33a	0.29	0.71	0.49	0.45	0.37	3.08	0.78	0.47
Specimens with interface, No.	2	4	1	5	2	1	6	4	3	1	3	2	1	5	2
Interface density, cells/mm2
CD45	280	1050	2854	324	2150	1446	671	280	1082	89	324	1264	2854	324	1968
CD8	154	283	1176	158	803	430	186	154	407	44	158	419	1176	158	792
CD4	263	498	2126	285	1657	1187	353	263	573	43	285	805	2126	285	1274
FOXP3	106	168	438	109	388	339	136	106	226	4	109	282	438	109	332
CD20	57	36	273	30	271	270	39	57	48	1	48	147	273	48	148
CD68	236	359	231	330	228	225	312	259	388	330	294	645	231	294	648
LAG3	38	29	68	44	41	14	44	23	68	6	44	68	68	33	95
CD8:FOXP3	1.6	1.63	2.69	1.8	1.98	1.27	1.88	1.64	1.80	10.40	1.45	1.54	2.69	1.45	2.25
CD8:CD68	0.76	0.40	5.09	0.38	3.50	1.91	0.40	0.76	0.41	0.13	0.41	1.15	5.09	0.41	2.74

LAG3, leukocyte activation gene 3; PD-L1, programmed cell death 1 ligand 1.

^aP ≤ .05.

TABLE 3

Comparison of immune marker density in the tumor and at the interface between tumor and normal tissue

Immune marker	Density, median, cells/mm2
	Intratumoral	Interface
CD45	224	1017a
CD8	119	213
CD4	214	422
FOXP3	31	163a
CD20	8	48a
CD68	210	294
LAG3	17	44
CD8:FOXP3	3	1.8
CD8:CD68	0.6	0.4

LAG3, leukocyte activation gene 3.

^aP ≤ .05.

FIGURE 1

Immune marker expression is shown in 3 representative patients. Case 1 is positive for PD-L1 expression on tumor and immune cells and has a high density of CD8+, CD68+, and LAG3+ cells, with rare CD20+ cells. This specimen was obtained from a patient with an unusually prolonged survival interval. Case 2 shows moderate PD-L1 expression on immune cells only, with an associated lymphoid aggregate and a low density of LAG3+ cells. Case 3 shows rare PD-L1 expression on immune cells only and has a lower density of all immune markers (H&E stains and immunostains; ×10). LAG3, leukocyte activation gene 3; PD-L1, programmed cell death 1 ligand 1.

Expression of Immune Checkpoint Proteins

Six cases (35%) showed tumor cell labeling for PD-L1 Table 2 and Figure 1 (range, 1%-10%; mean, 5%). Similarly, PD-L1 expression on tumor cells was associated with increased infiltrating immune cell density (P = .019) and increased density of intratumoral CD45+ cells (P = .010), CD8+ cells (P = .010), FOXP3+ cells (P = .003), and CD68+ cells (P = .010) Table 2 (Supplemental Figure 1). Thirteen cases (76%) expressed PD-L1 on immune cells, which was classified as focal in 6 cases (35%), moderate in 5 cases (29%), and diffuse in 2 cases (12%) Table 2 and Figure 1. The percentage of immune cells labeled for PD-L1 ranged from 1% to 60%, with a mean of 14%. PD-L1 expression on immune cells did not show a statistically significant association with infiltrating immune cell density, but higher PD-L1 expression on immune cells was correlated with an increased density of intratumoral CD8+ cells (P = .036) and FOXP3+ cells (P = .013), with similar CD8:FOXP3 ratios Table 2 (Supplemental Figure 1). PD-L1 expression on both tumor cells and immune cells was distributed throughout tumors and was not localized preferentially to the interface. Further, PD-L1 expression by tumor cells and immune cells showed no significant association with density of individual immune cell subsets at the tumor interface Table 2. There was no significant difference in PD-L1 expression between primary tumors and metastases (Supplemental Table 3).

LAG3 expression on immune cells was detected in all cases Table 2 and Figure 1. An increased density of intratumoral LAG3+ cells was seen in cases with tumor cell PD-L1 labeling (P = .003) and necrosis (P = .047) Table 2. Cases with high LAG3 density (>50th percentile) had an increased density of intratumoral CD68+ cells (P = .015) Table 2. LAG3 density was not significantly associated with the density of individual immune cell subsets at the tumor interface Table 2 and did not differ significantly between primary tumors and metastases (Supplemental Table 3).

Correlation With Clinical Variables

Infiltrating immune cell density and density of all immune cell subsets and checkpoint markers were not associated with recurrence, development of metastases, or survival during the follow-up period. The study was not sufficiently powered to assess differences in length of survival, particularly given the limited clinical follow-up for many patients. Anecdotally, 1 patient who had an unusually long survival interval of 9.7 years showed the highest immune cell infiltration and expression of PD-L1 on tumor and immune cells Figure 1. Specimens from this patient showed a significantly higher density of cells expressing CD45 (P = .047), CD8 (P = .003), CD4 (P = .003), and CD20 (P = .032), a higher CD8:CD68 ratio (P = .028), and a higher percentage of immune cells expressing PD-L1 (P = .028) than specimens from other patients. Although limited clinical information was available for the last 5 years of this patient’s care, per available records, this patient received surgical treatment only for the primary tumor and 1 local recurrence 5 years later, with no chemotherapy or radiation (Supplemental Table 2: patient 7). At the last available clinical follow-up, 1 year before the patient’s death, there was no evidence of recurrence or metastasis. No specific information about cause of death was available.

DISCUSSION

Our data demonstrate several key features of the immune TME of intimal sarcomas. First, we observed a rich immune cell infiltrate in all cases that consisted of a mix of CD8+ cytotoxic T cells, CD4+ helper T cells, FOXP3+ regulatory T cells, CD20+ B cells, and CD68+ macrophages. In a minority of cases, CD20+ B cells were seen in association with lymphoid aggregates. Second, this evidence of an active immune response is accompanied and balanced by expression of immune checkpoint proteins PD-L1 (on tumor cells and immune cells) and LAG3. Intimal sarcomas with the highest qualitative and quantitative immune cell densities are most likely to express PD-L1. Further, although higher densities of immune cells were present at the interface between tumor and normal tissue than within the tumor itself, it is the intratumoral densities of immune cells that correlate with PD-L1 expression. In tumors with the highest densities of LAG3 expression, an accumulation of CD68+ macrophages was also seen. Our study cohort, although limited in size, contained both primary and metastatic tumors and did not reveal any significant differences between these sites of disease.

Most existing studies of the immune milieu in sarcomas are broad, incorporating many tumor types and providing minimal insight into rare subtypes. Our current study both highlights specific features of 1 such rare tumor, intimal sarcoma, and helps place it in the broader context of sarcomas in general. Although tumor mutational burden (TMB) is often lower in sarcomas relative to many solid tumors with a robust immune response,²⁰ immune cell density does increase in genomically complex sarcomas compared with translocation-associated sarcomas.¹⁰ The same association between PD-L1 expression and tumor infiltrating lymphocyte (TIL) density that we see in intimal sarcomas has been described in other soft tissue sarcomas.¹¹ Similar to data in multiple other sarcoma types, our findings demonstrate evidence of an active immune response within intimal sarcomas, despite the overall low TMB in these tumors. The robust expression of PD-L1 we observed on both tumor and immune cells may be considered a surrogate for this response.

PD-L1 upregulation in the TME can occur through adaptive or constitutive pathways. In the former, expression is driven by interferon-γ (IFN-γ) and toll-like receptor signaling generated by activated immune cells.^21-24 The PD-L1/PD-1 pathway serves as 1 of multiple mechanisms of immune regulation to restrain immune responses and regulate their propagation. The induction of PD-L1 expression within tumors has been termed adaptive immune resistance, reflecting the ability of tumors to escape the active antitumor immune response through upregulation of the PD-1 immune checkpoint pathway in response to inflammatory stimuli.²⁵ Alternatively, PD-L1 can be constitutively expressed by tumor cells as a result of oncogenic signaling. Adaptive immune resistance has been described in an array of tumors, including melanoma, Merkel cell carcinoma, non-small cell lung carcinoma, and breast carcinoma.^25-28 The association we observed between PD-L1 and immune cell density suggests that the dominant pattern of PD-L1 expression in intimal sarcomas is adaptive. This finding further supports the existence of an active immune TME within these sarcomas, despite their relatively low TMB.

Despite evidence of an active immune response with adaptive expression of PD-L1, our data demonstrate the existence of balancing immunosuppressive forces within intimal sarcomas. LAG3, an immune checkpoint protein with constraining effects on CD8+ T cells, is expressed in all cases in our series. Further, increasing expression of LAG3 is associated with an increase in intratumoral CD68+ macrophages and, although not statistically significant, a trend toward a decrease in the CD8:CD68 ratio. In contrast, increases in regulatory T-cell marker FOXP3 are balanced by parallel increases in CD8+ T cells. LAG3 is a known phenotypic indicator of T-cell exhaustion.³ The widespread expression of LAG3 suggests that despite indications of initial T-cell activation, as evidenced by IFN-γ–driven adaptive PD-L1 expression, these intratumoral T cells may eventually become constrained by an immunosuppressive microenvironment. The concomitant increase in CD68+ macrophages associated with high LAG3 expression suggests the accumulation of a potentially suppressive population of TAMs within intimal sarcomas at a rate disproportional to CD8+ T cells. A population of such intratumoral macrophages has previously been reported to help sustain immunosuppression in a cohort of rhabdomyosarcomas and undifferentiated pleomorphic sarcomas.¹³ More investigation is required to better characterize the phenotype of these macrophages in intimal sarcomas and the potential role of other myeloid-derived suppressor cells.

Although it would be interesting to understand the relationships between the immune TME in intimal sarcomas and recurrence and survival as well as the progression over time within patients, our cohort was not sufficiently powered to fully evaluate these relationships. Across the cohort, we did not see significant differences in the immune components and checkpoint protein expression between primary tumors and metastases, but the cohort was too small to fully explore this idea within individual patients. Despite these limitations, we did have 1 interesting patient with a prolonged survival of more than 9 years. This patient also showed the highest TIL density and percentage of PD-L1 expression on both tumor and immune cells. Although anecdotal, the association is intriguing. Increased TIL density has been associated with improved survival in sarcomas.²⁹ An adaptive pattern of PD-L1 expression in combination with infiltrating immune cells has also been associated with improved survival in patients with melanoma and other malignancies, likely because it reflects an initial active immune response.⁸ In contrast, PD-L1 expression in sarcomas has been associated with improved survival in some studies but decreased survival in others.^30-32 Although anecdotal, the prolonged survival in our patient with a robust adaptive pattern of PD-L1 expression is intriguing and suggests the possibility that this immune response within intimal sarcomas could confer clinical benefit.

The results of our study support further investigation into therapeutic augmentation and support of the immune response in intimal sarcomas. Prior studies have shown that sarcomas broadly and undifferentiated pleomorphic sarcomas specifically can show high-density inflammatory infiltrates that predict response to immunotherapy.^5,9 It is difficult to directly compare degree of inflammation across studies because of differences in methodology, but our study suggests that intimal sarcomas may harbor the type of inflammation seen in other sarcomas that was predictive of response to immune-based therapies. Specifically, in the SARC028 trial, high T-cell density and PD-L1 expression predicted treatment response to pembrolizumab,⁹ features we see in a subset of intimal sarcomas. Another factor to consider is whether the molecular changes described in intimal sarcomas, including amplifications of MDM2, EGFR, PDGFRA, and KIT,^15,16 could have any bearing on response to immunomodulatory therapy. Dedifferentiated liposarcomas, which frequently show MDM2 amplification,²⁰ showed a response to pembrolizumab in the SARC028 trial.^4,5 Hyperprogression of tumors following anti–PD-1 therapy has been described, however, albeit rarely, in metastatic carcinomas with amplification of MDM2 or alterations in EGFR.³³ It would be important to know whether these rare hyperprogressions occur in other tumor types and whether they are seen only with blockade of the PD-L1/PD-1 pathway or are also seen with other forms of immune manipulation. Given the expression of LAG3 and the increase in TAMs seen in intimal sarcomas, there are multiple potential immunomodulatory pathways to explore beyond the PD-1/PD-L1 axis.

To our knowledge, our study is the first to characterize the immune TME of intimal sarcomas. We demonstrate a robust immune response with an adaptive pattern of PD-L1 expression. Because of our small sample size and retrospective study design, we are unable to draw conclusions as to the prognostic significance of this immune response, although the prolonged survival in 1 patient with a high TIL density and PD-L1 expression is provocative. Further studies are needed to evaluate the potential role of immunomodulatory therapy in treating patients with this aggressive tumor, which does not respond well to other adjuvant or neoadjuvant treatment modalities. Our findings would support further investigation of immune-based therapeutic interventions in intimal sarcomas as well as further studies of the TME of individual soft tissue sarcomas to allow for development of more guided and precise therapy.

Funding

This work was supported in part by funding through the Bloomberg-Kimmel Institute for Cancer Immunotherapy.

Acknowledgments

The authors thank Norman Barker, MS, MA, RBP, for his assistance with preparing figures for this article.

Conflict of interest disclosure

The authors have nothing to disclose.