PD-L1 and IDO1 expression and tumor-infiltrating lymphocytes in osteosarcoma patients: comparative study of primary and metastatic lesions

Abstract

Purpose

Programmed death ligand 1 (PD-L1) and indoleamine 2,3-dioxygenase 1 (IDO1) are immunosuppressive proteins known to be associated with poor prognosis in various cancers. However, their expression and clinical relevance in osteosarcoma remain unknown. In this study, the relationships of PD-L1 and IDO1 expression with clinicopathological features and prognosis were explored.

Methods

The expression of PD-L1, IDO1, CD3, CD4, and CD8 in 112 formalin-fixed, paraffin-embedded tumor tissues collected by biopsy or surgical resection from 56 osteosarcoma patients was evaluated immunohistochemically. Moreover, four osteosarcoma cell lines were evaluated for the effects of IFNγ on PD-L1 and IDO1 mRNA expression by real-time reverse-transcription polymerase chain reaction.

Results

In pre-neoadjuvant chemotherapy (NAC) primary specimens, 10 cases (17%) showed PD-L1 expression and 12 (21%) showed IDO1 expression. Six of ten cases (60%) with PD-L1 positivity co-expressed IDO1. In post-NAC metastatic lesions, the frequency of immunoexpression of PD-L1 and IDO1 was increased compared with that in pre-NAC specimens. PD-L1 and/or IDO1 expression was not associated with poor prognosis. PD-L1 immunoexpression was significantly associated with the infiltration of CD3⁺ T cells, CD4⁺ T cells, and CD8⁺ T cells; while, IDO1 immunoexpression was significantly associated with the infiltration of CD3⁺ T cells and CD4⁺ T cells. In all osteosarcoma cell lines, PD-L1 and IDO1 expression was upregulated by stimulation with IFNγ.

Conclusion

Our results suggest that the PD-L1 and IDO1 immune checkpoint inhibitors may provide clinical benefit in osteosarcoma patients with metastatic lesions after conventional chemotherapy.

Electronic supplementary material

The online version of this article (10.1007/s00432-020-03242-6) contains supplementary material, which is available to authorized users.

Introduction

Osteosarcoma is the most common malignant bone tumor in adolescents and young adults. After the establishment of anti-cancer drugs in the 1970s, patients with osteosarcoma were treated by wide resection and chemotherapy (Harrison and Schwartz 2017; Liao et al. 2017; Palmerini et al. 2017; Shimizu et al. 2017; Sundara et al. 2017). As a result, in Japan, the 5-year overall survival rate improved to 78% (Iwamoto et al. 2009). However, even now, the survival rates in patients with recurrence or metastasis to the lung or to other bones remain poor (Harrison and Schwartz 2017; Liao et al. 2017; Palmerini et al. 2017; Shimizu et al. 2017; Sundara et al. 2017). No new anti-cancer drugs have been developed for osteosarcoma over the last 30 years. Therefore, there is a need to develop novel therapeutic agents against advanced osteosarcoma.

Programmed death ligand 1 (PD-L1) is the ligand of programmed cell death 1 (PD-1), and their interaction promotes T-cell immune tolerance. PD-L1 expression has been shown to enable malignant tumors to avoid immune surveillance (Deng et al. 2017; Gu et al. 2017; Jung et al. 2017; Liao et al. 2017; Lin et al. 2015; Miyoshi et al. 2016; Muenst et al. 2014; Nakanishi et al. 2007; Okita et al. 2017; Troiano et al. 2018; Yagi et al. 2019; Yamaki et al. 2017; Zhou et al. 2017). PD-1 inhibitors have been approved for the treatment of various cancers and shown to confer survival benefits in recent years (Meng et al. 2018; Torabi et al. 2017; Wang et al. 2017). It has also been reported that the rate of immunopositivity for PD-L1 in osteosarcoma ranges from 14 to 75% (Koirala et al. 2016; Palmerini 2017; Shen et al. 2014; Sundara et al. 2017; Torabi et al. 2017) and that about 80% of osteosarcoma patients have high PD-L1 mRNA expression (Shen et al. 2014). Furthermore, it was shown that PD-L1 is more highly expressed in osteosarcoma patients at an advanced stage (Liao et al. 2017; Sundara et al. 2017). However, no consensus has been reached about whether PD-L1 can be a novel therapeutic target.

Indoleamine 2,3-dioxygenase 1 (IDO1) is an enzyme of tryptophan metabolism. Tryptophan is necessary for the activation and proliferation of T cells, and is metabolized by the kynurenine pathway. In the tumor microenvironment, the expression of IDO1 and kynurenine, which suppresses effector T cells and activates suppressor T cells, is increased. This enables tumor cells to escape from immune responses (Uyttenhove et al. 2003). IDO1 immunoexpression is known as a factor associated with a poor prognosis in various malignancies (Kiyozumi et al. 2019; Kozuma et al. 2018; Ma et al. 2018; Rosenbaum et al. 2018; Schalper et al. 2017; Zhai et al. 2017). However, the clinical significance of IDO1 expression in osteosarcoma has not been reported.

An interest in immunotherapy for osteosarcoma has developed in recent years. PD-1/PD-L1 pathway inhibitors have been tested for the treatment of bone and soft-tissue sarcomas in clinical trials, but unfortunately little response has been seen in advanced osteosarcoma patients (Tawbi et al. 2017). Therefore, these patients may require a combination of immune checkpoint inhibitors and conventional cytotoxic agents (Harrison and Schwartz 2017).

In cancers such as lung (Fournel et al. 2019), ovarian (Kim et al. 2018; Lo et al. 2017; Mesnage et al. 2017), breast (Pelekanou et al. 2017), and head/neck (Jie et al. 2017) cancers, focus has been placed on changes in the immune microenvironment in patients between before and after neoadjuvant chemotherapy (NAC). These studies revealed that the expression of PD-L1 and the level of tumor-infiltrating lymphocytes (TILs) were increased after NAC compared with those before it (Fournel et al. 2019; Gao et al. 2018; Jie et al. 2017; Kim et al. 2018; Lo et al. 2017; Meng et al. 2018; Mesnage et al. 2017; Pelekanou et al. 2017). Furthermore, some reports described that, in osteosarcoma patients, metastatic tumors showed improved immunogenicity, the induction of PD-L1 expression, and more tumor-infiltrating lymphocytes compared with the primary tumors (Negri et al. 2019; Wang et al. 2019).

We, thus, investigated the associations of PD-L1 and IDO1 expression with clinicopathological characteristics and their prognostic value in patients with primary or metastatic osteosarcoma. We evaluated PD-L1 and IDO1 immunoexpression in the tumor immune microenvironment by comparing pretreated conventional osteosarcoma primary tumor biopsies and post-treated specimens or metastatic specimens.

Materials and methods

Patients and tissue samples

We used the samples of osteosarcoma patients registered from 1997 to 2017 in the database of the Department of Anatomic Pathology, Graduate School of Medical Sciences, Kyushu University, Fukuoka, Japan. All cases were reviewed based on imaging findings and histological examination using hematoxylin and eosin (H&E) staining.

A total of 112 formalin-fixed, paraffin-embedded (FFPE) samples from 56 patients were obtained for immunohistochemical study (Supplementary. Fig. S1). These samples had been collected from biopsy specimens or surgically resected specimens in primary, metastatic, and recurrent sites. All biopsy specimens from primary sites were obtained before chemotherapy. Samples resected after chemotherapy were also included in the immunohistochemical analysis. All patients received preoperative chemotherapy including methotrexate (MTX), cisplatin (CDDP), and adriamycin (ADR), or CDDP + ADR (Table 1). The histological response was assessed by the largest cut surface stained by H&E of the operatively resected post-NAC specimen. Three pathologists (YT, KK, and YY) measured the percentage of viable cells (PVC) and classified this as follows: ≥ 50% or < 50%. Survival data were available for all cases.

Table 1.

Clinicopathological parameters

Parameter (n = 56)	No. of patients	%
Median age (years): 16.5 (7–63)
≤ 20	42	75.0
20 < , ≤ 49	8	14.3
50 ≤	6	10.7
Sex
Female	15	26.8
Male	41	73.2
Histology variant
Osteoblastic	34	60.7
Chondroblastic	9	16.0
Fibroblastic	5	8.9
Giant cell rich	5	8.9
Telangiectatic	2	3.6
Malignant fibrous histiocytoma-like	1	1.8
Site
Femur	30	53.6
Tibia	14	25.0
Rib	3	5.4
Humerus	2	3.6
Pelvis	2	3.6
Fibula	2	3.6
Radius	2	3.6
Nasal	1	1.8
Neoadjuvant chemotherapy
Not Given	1	1.8
Given	55	98.2
Methotrexate, doxorubicin, cisplatin (MAP)	51	91.0
Doxorubicin, cisplatin (AP)	4	7.1
Metastasis
Absent	30	53.6
Present	26	46.4
Initial metastasis	3	5.4
Recurrence	1	1.8
Tumor death
Absent	41	73.2
Present	15	26.8

Immunohistochemistry

Formalin-fixed, paraffin-embedded tissue was sliced into sections at a thickness of 4 μm. Antigen retrieval was performed by heating the slides in 10-mM sodium citrate (pH 6.0), ethylenediaminetetraacetic acid, or Target Retrieval Solution (Dako, Carpinteria, CA). The following rabbit and mouse monoclonal antibodies were used as the primary antibodies: anti-PD-L1, anti-CD3, anti-CD4, anti-CD8, anti-IDO1, and anti-IFNγ (Supplementary. Table S1).

The cut-off for defining positivity regarding the percentage of PD-L1- and IDO1-positive cells was 1%, as reported in the literature (Kozuma et al. 2018). PD-L1 exhibits cytoplasmic membrane staining. IDO1 exhibits cytoplasmic and membrane staining. The numbers of CD3⁺, CD4⁺, and CD8⁺ cells were counted per microscopic field at 400× magnification in five independent fields (Perea et al. 2017). To examine whether CD3⁺, CD4⁺, and CD8⁺ cells secrete IFNγ, dual-color immunochemical staining was performed for IFNγ and CD3, CD4, and CD8 for 10 cases, using EnVision G/2 Doublestain System (Dako Inc.). Three pathologists (YT, KK, and YY) independently evaluated the immunohistochemical staining results for each sample.

Cell cultures

Human OS cell lines, U2OS, SaOS2, MNNG, and MG63, were provided by the American Type Culture Collection (Rockville, MD). These cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM; Invitrogen Corp., Carlsbad, CA) supplemented with 10% fetal bovine serum plus penicillin and streptomycin. Cells were grown at 37℃ in a humidified 5% CO₂ incubator and were treated with recombinant IFNγ (200 ng/mL; PeproTech, Inc., Rocky Hill, NJ) for 24 h in DMEM.

Real-time RT-PCR

Total RNA was extracted with QIAzol reagent (Qiagen, Valencia, CA) and cDNA was synthesized from total RNA using ReverTra Ace reverse-transcription kit (TOYOBO Co., Osaka, Japan). SYBR/ROX Master Mix was used. The primers used are shown in Supplementary. Table S2. Finally, the ratio of cDNA relative to the EEF1a1 endogenous control was calculated using the 2^−ΔΔCt method. Real-time PCR was performed in triplicate and repeated at least three times in separate experiments.

Statistical analysis

The cases of conventional osteosarcoma were evaluated by statistical analysis. Fisher’s exact test was used to analyze correlations between two dichotomous variables. The statistical analysis of PD-L1 and IDO1 expression and TILs among primary biopsy specimens and metastatic specimens or wide-resection specimens was performed by Wilcoxon’s signed-rank test. The correlations of PD-L1 and IDO1 with TILs were calculated by Wilcoxon’s test. The correlations among mRNA levels were evaluated by t-test. Survival curves were calculated using the Kaplan–Meier method and the differences were evaluated by the log-rank test. Statistical significance was defined as p < 0.05. Data analysis was performed using JMP statistical software package (version JMP version 13.0.0; SAS Institute Inc., Cary, NC).

Results

Clinicopathological and histopathological results

The clinicopathological parameters of the subjects are shown in Table 1. The age at tumor onset ranged from 7 to 63 years (mean 16.5 years old). There were 41 males and 15 females among the OS patients. There were 26 cases of distant metastasis and one case of local recurrence. Three cases had proven distant metastasis at diagnosis. These tumors were located mainly in the femur or tibia (Table 1). The histological variants are summarized in Table 1. The most common such variant in this cohort was osteoblastic osteosarcoma (n = 34/56, 60.7%).

Immunohistochemistry

PD-L1 and IDO1 expression in osteosarcoma patients

Representative results of PD-1 and IDO1 immunohistochemical studies are shown in Fig. 1. Among 56 biopsy specimens before chemotherapy, 10 cases (17%) exhibited PD-L1 expression and 12 (21%) exhibited IDO1 expression. Six of ten cases (60%) with PD-L1 positivity also co-expressed IDO1, which was a significant association (p = 0.0038). Our findings revealed that positivity for IDO1 expression in biopsy specimens could be a predictor of chemotherapy efficacy (p = 0.0382) (Table 2).

Fig. 1.

Representative images of the immunoexpression of PD-L1 (a) and IDO1 (b) in primary biopsy specimens from patients with osteosarcoma. PD-L1 membrane staining, IDO1 cytoplasmic and membrane staining. Scale bars: 50 µm. PD-L1 programmed death ligand 1, IDO1 indoleamine 2,3-dioxygenase 1

Table 2.

Correlations of clinicopathological characteristics with PD-L1 and IDO1 immunoexpression

	N	PD-L1 + (%) (biopsy specimen)	p value	IDO1 + (%) (biopsy specimen)	p value	Chemotherapy response (viable cell ≤ 50%)	p value	Metastasis	p value	Initial metastasis	p value
All	56	10 (18)		12 (21)		29 (52)		26 (46)		3 (4)
Age, years
≤ 20	42	6 (14)	0.3546	8 (17)	0.6435	23 (55)	1.000	19 (45)	0.5487	0 (0)	0.0071*
20 < , ≤ 50	8	2 (25)		2 (25)		3 (38)		3 (38)		1 (13)
< 50	6	2 (33)		2 (33)		3 (50)		4 (67)		2 (33)
Gender
Male	41	5 (12)	0.111	7 (17)	0.2701	23 (56)	0.27	19 (46)	1.000	2 (5)	1.000
Female	15	5 (33)		5 (33)		6 (40)		7 (47)		1 (7)
Site
Extremity	50	9 (18)	1.000	11 (22)	1.000	28 (56)	1.000	22 (44)	0.4006	2 (4)	0.2929
Trunk	6	1 (17)		1 (17)		1 (17)		4 (67)		1 (17)
Chemotherapy
No	1	0 (0)		0 (0)		–		–		–
Yes	55	10 (18)		12 (22)		–		–		–
Viable cellsa
> 50%	13	2 (15)	1.000	0 (0)	0.0382**	–		–		–
≤ 50%	29	5 (17)		9 (31)		–		–		–
N/A	14	3 (21)		3 (21)		–		–		–
Metastais
Present	26	5 (19)	1.000	6 (23)	1.000	14 (54)	0.7393	–		–
Absent	30	5 (17)		6 (20)		15 (50)		–		–
Initial metastasis
Present	3	1 (33)	0.4524	0 (0)	1.000	1 (33)	0.5285	–		–
Absent	53	9 (17)		12 (23)		28 (53)		–		–
IDO1
Positive	12	6 (50)	0.0038*	–		–		–		–
Negative	43	6 (14)		–		–		–		–

PD-L1 programmed death ligand-1, IDO-1 indoleamine 2,3-dioxygenase-1

^aViable cells in the primary site after chemotherapy

^*p < 0.1, **p < 0.05

Effects of NAC in primary site related to PD-L1 and IDO1 expression

The expression levels of PD-L1 and IDO1 of specimens obtained from biopsy (pre-NAC) or wide resection (post-NAC) were directly compared in the same patients. The frequency of PD-L1 immunoexpression did not differ significantly between biopsy specimens and post-NAC resected specimens. The frequency of immunoexpression for IDO1 in post-NAC specimens was significantly higher than that in pre-chemotherapy specimens (p = 0.0032) (Fig. 2a).

Fig. 2.

The changes of immunoexpression of PD-L1 and IDO1 between before and after neoadjuvant chemotherapy (a: resected specimens, b: metastatic specimens) in matched samples

Differences in PD-L1 and IDO1 expression between pre-NAC biopsy specimens of primary sites and post-NAC specimens of metastatic sites

We evaluated the expression of PD-L1 and IDO1 in primary and metastatic lesions in the same patients. Two representative cases are shown in Figs. 3 and 4. The frequency of immunoexpression for PD-L1 in post-NAC specimens of metastatic sites (60%, n = 8/14) was significantly higher than that in pre-NAC biopsy specimens of primary sites (28%, n = 4/14) (p = 0.0304). Similarly, the frequency of IDO1 immunoexpression was higher in metastatic lesions (78%, n = 11/14) than in primary sites (42%, n = 5/14) (p = 0.0113) (Fig. 2b). In our cohort, most metastatic lesions were present in lung (n = 11/14: 78%), but they were also present in bony tissue (n = 3/14) such as tibia, femur, and ilium (Table 3). In bony metastatic osteosarcomas, PD-L1 immunoexpression was higher than in primary lesions (n = 2/3: 67%).

Fig. 3.

Representative images that show the changes of PD-L1 expression and infiltration of CD4⁺ and CD8⁺ lymphocytes between primary and metastatic tumors. Scale bars: 50 µm. In metastatic lesions, PD-L1 expression and the number of infiltrated CD4⁺ lymphocytes were increased compared with those in the primary lesions

Fig. 4.

Representative photographs revealing the changes of IDO1 expression and infiltration of CD4⁺ and CD8⁺ positive lymphocytes between primary and metastatic tumors. Scale bars: 50 µm. In metastatic lesions, IDO1 expression and the number of infiltrated CD4⁺ lymphocytes were increased compared with those in the primary lesions. Moreover, the number of CD8⁺ lymphocytes was moderately increased compared with that in the primary tumor

Table 3.

Clinical features and PD-L1 and IDO1 expression in matched primary and metastatic samples

Primary site	Metastatic site	NAC	AC*	Duration (months)**	Outcome	PD-L1 (TPS: %)		IDO1 (TPS: %)
						Primary	Meta	Primary	Meta
Rib	Pleura	AP	AP***	38	AWD	1	0	0	0
Tibia	Lung	MAP	MAP → ICE	6	DOD	0	3	1	1
Tibia	Lung	MAP	MAP + I	2	AWD	0	1	0	15
Femur	Lung	MAP	MAP + I	4	DAD	1	3	0	0
Radius	Lung	MAP	MAP + I	17	NED	0	1	0	3
Tibia	Lung	MAP	MAP + I	11	DAD	0	1	0	7
Ilium	Tibia	MAP	MAP + I	15	DAD	0	0	0	0
Thigh	Lung	MAP	none†	2	DAD	0	0	1	1
Femur	Lung	MAP	MAP + I	2	DAD	1	0	1	10
Femur	Lung	MAP	MAP + I	14	AWD	1	1	0	1
Rt.femur	Lt.femur	MAP	MAP + IE	50	DAD	0	4	1	1
Femur	ilium	MAP	MAP + I	42	DAD	0	5	1	1
Femur	Lung	MAP	MAP + I	22	NED	1	1	0	5
Femur	Lung	MAP	MAP + I	13	NED	0	0	0	3

NAC neo adjuvant chemotherapy, AC adjuvant chemotherapy, MAP Methotrexate + Adriamycin (ADR) + Cisplatin (CDDP), AP ADR + CDDP, ICE Ifosfamide (IFO) + CDDP + Etoposide (VP-16), I IFO, IE IFO + VP-16, AWD alive with disease, DOD died of disease, NED no evidence of disease

*Chemotherapy resimen perfomed between surgical operaion and metastaic biopsy or resection

**Time duration between surgical operaion and metastaic biopsy or resection

***This case has not undergone surgical operation

^†This case has undergone surgical operation of metastatic lesions immediately following surgical operation of primary tumors

Tumor-infiltrating lymphocytes (TILs)

Associations of PD-L1 and IDO1 expression with TILs

We evaluated the correlations of PD-L1 and IDO1 expression with TILs. The infiltration rates of CD3⁺ T cells, CD4⁺ T cells, and CD8⁺ T cells in pre-NAC biopsy specimens of primary sites with PD-L1 positivity were significantly higher than those in equivalent PD-L1-negative cases (p = 0.0142, 0.0282, and 0.0088, respectively) (Fig. 5a). There was significantly more lymphocytic infiltration (CD3⁺ and CD4⁺) in primary biopsy specimens of IDO1-positive patients than in IDO1-negative patients (p = 0.0049 and 0.0394, respectively) (Fig. 5b).

Fig. 5.

The correlations of tumor-infiltrating lymphocytes with the immunoexpression of PD-L1 (a) and IDO1 (b)

Effect of NAC on TILs in primary tumors

Next, we investigated whether there was a difference in T-cell infiltration between pre- and post-NAC. The infiltration rates of CD3⁺ and CD4⁺ T cells in resected specimens (post-NAC) were significantly increased compared with those in biopsy specimens (pre-NAC) (p = 0.0408 and 0.0127, respectively) (Fig. 6a).

Fig. 6.

The changes of TILs between before and after neoadjuvant chemotherapy (a: resected specimens, b: metastatic specimens). TILs: tumor-infiltrating lymphocytes

Differences in TILs between pre-NAC biopsy specimens in primary sites and post-NAC metastatic specimens

We compared TIL counts on pre-NAC biopsy specimens between primary sites and post-NAC metastatic specimens. We found that lymphocytic infiltration (CD3⁺, CD4⁺, and CD8⁺) in metastatic lesions was significantly increased compared with that in primary lesions (p = 0.0010, p = 0.0046, and p = 0.0063, respectively) (Fig. 6b).

Prognostic significance of PD-L1, IDO1, and TILs in primary tumors

We assessed the prognostic significance of clinicopathological factors, namely, PD-L1 and IDO1 expression and TILs, using Kaplan–Meier survival analysis (Fig. 7a, b). There were no significant correlations for overall survival or event-free survival. The level of TILs in the pre-NAC primary tumors was not a significant prognostic marker. However, a high number of TILs in primary lesions tended to be associated with better overall survival (Fig. 7a). Distant metastasis was associated with poor overall survival (Fig. 7a).

Fig. 7.

Kaplan–Meier survival curves of PD-L1 and/or IDO1 expression, TILs, and clinicopathological characteristics illustrate overall survival (a) (log-rank test). Kaplan–Meier survival curves for event-free survival are also presented according to PD-L1 and/or IDO1 expression, TILs, and clinicopathological characteristics (b) (log-rank test)

Co-expression of IFNγ with CD3, CD4, and CD8

We performed a double immunohistochemical staining study (n = 10) to evaluate whether TILs secreted IFNγ. As shown in Supplementary. Fig. S2, each TIL (CD3⁺, CD4⁺, and CD8⁺) was positive for IFNγ in all examined cases.

IFNγ stimulation induces expression of PD-L1 and IDO1 in human osteosarcoma cell lines

We performed real-time RT-PCR for the osteosarcoma cell lines U2OS, SaOS2, MNNG, and MG63 to determine the mRNA levels of PD-L1 and IDO1 in these cell lines stimulated with IFNγ. The results revealed that the mRNA levels were significantly increased by IFNγ in all osteosarcoma cell lines (Fig. 8).

Fig. 8.

Interferon-gamma-induced mRNA expression of PD-L1 and IDO1 in human osteosarcoma cell lines

Discussion

Combination treatment of pre- and postoperative chemotherapy and surgical resection improved the outcomes for patients with non-metastatic osteosarcoma (Iwamoto et al. 2009; Palmerini et al. 2017). However, in metastatic patients, additional chemotherapy after neoadjuvant chemotherapy does not improve the poor outcome (Harrison and Schwartz 2017; Liao et al. 2017; Palmerini et al. 2017; Shimizu et al. 2017; Sundara et al. 2017). The cause of death in osteosarcoma patients is usually metastatic lung disease or recurrence of lung metastasis (Harrison and Schwartz 2017). Therefore, to improve the outcome of osteosarcoma patients, there is a need to manage and prevent distant metastasis. We established this study to focus on patients with progressive metastasis after resection and NAC of primary and metastatic tumors.

Our results showed that the frequency of immunoexpression of PD-L1 and IDO1 in post-NAC metastatic tumors was significantly higher than that in pre-NAC primary tumors. Therefore, PD-L1 and IDO1 immune checkpoint inhibitors may provide clinical benefit in osteosarcoma patients with post-NAC metastatic lesions.

Takamori et al. showed that, in lung metastatic specimens, the frequency of PD-L1 immunoexpression on tumor cells did not differ significantly between metastatic and primary lesions (Takamori et al. 2018). This suggested that, in metastatic lesions, PD-L1 immunoexpression was not induced by “metastasis.” On the other hand, it was also reported that NAC including adriamycin or cisplatin enhanced PD-L1 expression in cell lines and cohorts of patients with lung cancer (Fournel et al. 2019), ovarian cancer (Lo et al. 2017; Mesnage et al. 2017), breast cancer (Pelekanou et al. 2017), and head and neck (Jie et al. 2017) cancer. The mechanisms behind this were considered to involve tumor cells killed by chemotherapy emitting immune cell death signals that activate cytotoxic T cells and CD8⁺ T cells infiltrated into tumor tissue (Kim et al. 2018). Methotrexate, adriamycin, and cisplatin are the current standard regimens worldwide for patients with osteosarcoma. In our study, in osteosarcoma patients with metastasis, the frequency of immunoexpression of PD-L1 and IDO1 in post-NAC metastatic lesions was significantly increased compared with that in pre-chemotherapy primary specimens. In addition, the infiltration rates of CD3⁺, CD4⁺, and CD8⁺ T cells were increased in metastatic lesions. Seeber et al. investigated IDO1 and PD-L1 immunoexpression in clear cell renal cell carcinoma of patients undergoing treatment with anti-PD-1 inhibitor after chemotherapy and concluded that IDO1 expression was observed more frequently in responders than in non-responders. There was no difference in the PD-L1 expression between responders and non-responders (Seeber et al. 2018). However, Mesnage et al. revealed that patients with the induction of tumor-infiltrating lymphocytes after chemotherapy were significantly more likely to be PD-L1-positive. Moreover, it was expected that high co-expression of IFNγ and PD-L1 immunoexpression would be a prognostic marker of response to immunotherapy (Mandai et al. 2016). These results suggested that the alteration of tumor immunity by anti-PD-1 inhibitor might be beneficial as maintenance after primary treatment (Mesnage et al. 2017). On the other hand, some studies performed whole-exome and whole-genome sequencing on paired primary and pulmonary metastatic osteosarcomas. They reported that the metastatic tumors had significantly higher mutational burden and genomic instability compared with the primary tumors (Wang et al. 2019). It was also described that three genes, TP53, RB1, and SETBP1, were recurrently altered in matching primary-metastatic pairs (Negri et al. 2019). These findings indicated the therapeutic potential of immunotherapy for metastatic osteosarcomas. In osteosarcoma patients, it is possible that PD-L1 expression is induced by NAC and/or by metastasis.

We also expected that these lesions would be sensitive to anti-PD1 antibody. It is possible that anti-PD1 immunotherapy might improve the outcome of osteosarcoma patients with metastatic lesions appearing after NAC.

IFNγ is a multifunctional cytokine with dual roles in tumor immunity, namely, antitumor and protumor activities (Castro et al. 2018), as well as having important roles in innate and adaptive immunity. Upon the development of antigen-specific immunity, IFNγ is secreted by activated effector T cells (Dunn et al. 2006), so IFNγ was expected to work as an antitumor agent. However, it is becoming clear that IFNγ actually has effects to support tumorigenesis in the tumor microenvironment, reflecting negative feedback pathways that limit cytotoxic T-cell activation. Some reports have shown that IFNγ induces PD-L1 expression in tumor cells and upregulates IDO1 expression. Abiko et al. reported that IFNγ in the ovarian cancer microenvironment was significantly correlated with the number of infiltrating CD8⁺ and CD4⁺ lymphocytes, that IFNγ was derived from lymphocytes, and that an IFNγ-rich microenvironment is strongly correlated with a lymphocyte-rich microenvironment (Abiko et al. 2015).

In the present study, IFNγ stimulation increased PD-L1 and IDO1 expression in all osteosarcoma cell lines. Many reports have suggested that IFNγ from immune cells induced the expression of PD-L1 (Mandai et al. 2016). Our double immunohistochemical staining showed that CD3⁺, CD4⁺, and CD8⁺ T cells secreted IFNγ. In the literature, it is asserted that a case with high co-expression of IFNγ and PD-L1 immunoexpression would be expected to be sensitive to anti-PD1/PD-L1 inhibitors (Mandai et al. 2016). It has also been reported that, in vivo, high IFNγ stimulation such as via vaccination would induce PD-L1 expression and increase tumor immunity in lung cancer (Kozuma et al. 2018) and ovarian cancer (Abiko et al. 2013). It was also shown that IFNγ upregulated IDO1 expression in lung cancer (Kozuma et al. 2018).

In the present study, at metastatic sites, PD-L1 and IDO1 expression was associated with the infiltration of CD8⁺ T lymphocytes. Therefore, it was considered that PD-1/PD-L1 inhibitors might be useful for treating metastatic lesions in osteosarcoma patients. Moreover, it was suggested that combination therapy targeting PD-L1 and IDO1 would improve the poor outcome of advanced osteosarcoma patients. Some clinical trials also revealed that combination therapy of PD-1 inhibitor and IDO1 inhibitor might be a treatment option in patients with non-small-cell lung cancer (Kozuma et al. 2018) and glioblastoma (Zhai et al. 2017). Seeber et al. showed that IDO1 immunoexpression is related to response to immunotherapy in metastatic renal cell carcinoma (Seeber et al. 2019). In addition, some studies examining the possibility of using immune check point inhibitor to treat patients with other metastatic malignancies such as melanoma (Angela et al. 2019), urothelial carcinoma (Sharma et al. 2017), and renal cell carcinoma (Gill et al. 2016) have been reported. Therefore, it seems that patients with metastatic osteosarcomas would benefit from immune checkpoint inhibitors.

Our study revealed lower frequencies of immunoexpression of PD-L1 (17%) and IDO1 (21%) in pre-NAC primary lesions. However, PD-L1 and IDO1 co-immunoexpression was significantly present in pre-NAC primary lesions (60%) and tended to be associated with post-NAC metastatic lesions (57%). Our results showed that the rate of co-immunoexpression of PD-L1 and IDO1 was higher than in previous reports on lung cancer [from 7.1% (Schalper et al. 2017) to 28.8% (Kozuma et al. 2018)].

Our results also showed that neither PD-L1 nor IDO1 expression was associated with shorter overall survival (OS) or event-free survival (EFS). However, if PD-L1 or IDO1 immunoexpression was observed, OS and EFS tended to be shorter than in the cases immunoexpressing neither PD-L1 nor IDO1. In other malignancies, PD-L1 expression was found to be correlated with worse prognosis (Deng et al. 2017; Gu et al. 2017; Jung et al. 2017; Lin et al. 2015; Miyoshi et al. 2016; Muenst et al. 2014; Nakanishi et al. 2007; Okita et al. 2017; Troiano et al. 2018; Yagi and Baba 2019; Yamaki et al. 2017; Zhou et al. 2017), as was the case for IDO1 expression (Kiyozumi et al. 2019; Kozuma et al. 2018; Ma et al. 2018; Rosenbaum et al. 2018; Schalper et al. 2017; Zhai et al. 2017). Similarly, in the literature, it was claimed that PD-L1 expression in osteosarcoma is related to a worse prognosis (Koirala et al. 2016; Liao et al. 2017; Palmerini et al. 2017). However, in contrast, Sundara et al. claimed that there are different immunomarkers associated with poor prognosis in primary osteosarcoma patients (Sundara et al. 2017). It was also indicated that PD-L1 expression was not significantly associated with overall survival in other malignancies (Pelekanou et al. 2017). CD8⁺ T-cell infiltration was also reported to be correlated with better prognosis in osteosarcoma patients (Palmerini et al. 2017). Against this background, further study related to the tumor microenvironment of osteosarcoma should be conducted. Findings suggest that metastasis after the completion of treatment (operation and chemotherapy) might be related to poor prognosis in osteosarcoma patients. It is, thus, important to reduce metastatic potential and maintain a progression-free status. The present study emphasizes the importance not of radical treatment in advanced osteosarcoma patients but rather of the suppression of disease progression using immune checkpoint inhibitors. Further studies using a larger number of cases will be required to confirm the usefulness of these therapeutic tools. From the immunohistochemical results in this study, it can be concluded that the PD-L1 and IDO1 immune checkpoint inhibitors may provide clinical benefit in osteosarcoma patients with metastatic lesions after classical chemotherapy.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

PD-L1

Programmed death ligand 1

IDO1

Indoleamine 2,3-dioxygenase 1

NAC

Neoadjuvant chemotherapy

TIL

Tumor-infiltrating lymphocyte

MTX

Methotrexate

CDDP

Cisplatin

ADR

Adriamycin

OS

Overall survival

EFS

Event-free survival

Compliance with ethical standards

Conflict of interest

All authors declare that they have no conflicts of interest.

Ethical approval

The institutional review board at Kyushu University approved this study (approval codes: 29-625, 29-429).

Informed consent

Informed consent was obtained from all participants included in this study.