PD-1/PD-L1 expression in a series of intracranial germinoma and its association with Foxp3+ and CD8+ infiltrating lymphocytes

One histopathological characteristic of intracranial germinoma is abundant tumor-infiltrating lymphocytes (TILs) showing a two-cell pattern with large undifferentiated tumor cells. The programmed cell death 1 (PD-1)/programmed cell death 1 ligand (PD-L) axis has recently been recognized as an anti-tumor immune system. To evaluate intratumor immune status in intracranial germinoma, we examined expressions of PD-1 and PD-L1 (clone 28–8) and subtypes of TILs. Expressions of PD-1 and PD-L1 were detected immunohistochemically in 25 formalin-fixed, paraffin-embedded tumor specimens from 24 patients with intracranial germinoma consisting of 22 primary and 3 recurrent tumors. To evaluate subtypes of TILs, quantification of lymphocytes with CD3, CD8, CD4, and Foxp3 was performed. Statistical analyses were performed among PD-1, PD-L1 and subtypes of TILs. In 25 tumor tissue, expressions of PD-1 in TILs and PD-L1 in tumor cells were identified in 96% (24/25) and 92% (23/25), respectively. Expression of PD-1 was associated with CD3+ TIL density. Expression of PD-1 correlated with Foxp3+ TIL density and CD8+ TIL density, but not with CD4+ TIL density. Furthermore, expression of PD-1 correlated strongly with Foxp3+/CD4+ ratio. Taken together, increase of PD-1+ expression is associated with accumulation of Foxp3+ and CD8+ TILs. These findings intimate that PD-1/PD-L1 axis might shape the immune infiltration suggesting a modulation of the immune response and subsequent tumor growth in intracranial germinoma. Anti-PD-1 and anti-PD-L1 are potential immune therapeutic strategies in intracranial germinoma.


Introduction
Central nervous system (CNS) germ cell tumors (GCTs) are common in Asia, where they account for 8-15% of all CNS tumors of childhood, compared to 3-4% in the United States [1][2][3]. Intracranial germinoma is the most common type of CNS GCT, accounting for up to a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 two-thirds of all intracranial GCTs [2,4]. Intracranial germinomas develop mostly in children, with a strong predilection for the young adult population [1][2][3]. Surgery followed by radiotherapy plus chemotherapy has achieved excellent survival outcomes for patients with intracranial germinomas [5,6]. In about 10-20% of patients, however, the tumor recurs 10 years after firstline treatment [6]. Furthermore, since the majority of patients with this disease are children and adolescents, a large irradiated volume or high radiation dose results in late adverse effects such as growth disturbances or brain dysfunction [7]. Thus, with this highly curable disease, alternative therapeutic strategies for treating refractory tumors, preventing avoidable morbidity and maintaining quality of life have become the main goals of current pediatric oncological efforts.
Programmed cell death 1 (PD-1) is a receptor in the CD28 family, and plays an important role in immune tolerance and immune escape for a variety of tumor cells [8,9]. One major ligand of PD-1 is programmed death ligand 1 (PD-L1). The PD-1/PD-L1 axis attenuates antitumor immune system as immune checkpoint [10]. The effector functions of T lymphocytes expressing PD-1 in the tumor microenvironment could be down-regulated upon activation by PD-L1, which is frequently expressed on tumor cells [11]. Immune checkpoint inhibitors have recently constituted a novel class of treatment that target ligands and receptors [12].
Large tumor cells of intracranial germinoma are frequently accompanied by an abundance of tumor-infiltrating lymphocytes (TILs), which demonstrate a characteristic histology known as the "two-cell pattern" [13]. It is no exaggeration to say that the genesis, development and survival of tumor cells depend on effective immune escape mechanisms. Recent research has revealed that immune escape is a complex process that includes tissue isolation, molecular simulation and immune suppression [11,14]. One of the major components of this system is an immune checkpoint signal, the PD-1/PD-L1 axis [11]. However, expressions of PD-1 and PD-L1 as well as anti-tumor immune reactions remain unclear in intracranial germinoma.
This study examined expressions of PD-1 and PD-L1 in intracranial germinoma and characterized the subtypes of TILs showing surface antigens such as CD3+, CD8+, CD4+, and Foxp3+ using a quantitative evaluation method. Statistical analyses were performed to evaluate correlations among expressions of PD-1/PD-L1 axis and subtypes of TILs in intracranial germinoma.

Patient selection and sample collection
We evaluated formalin-fixed, paraffin-embedded (FFPE) tumor specimens from patients with intracranial germinoma who were treated in Kyoto University Hospital between 2002 and 2016. We excluded two cases in which specimens were accompanied with large areas of necrosis and only small numbers of tumor cells were left. A total of 24 patients of surgically resected intracranial germinoma consisting of 22 primary and 3 recurrent tumors were recruited for this analysis. Among these, a pair of primary and recurrent tumors were from the same patient experienced recurrence, who underwent surgical resection twice in our institute. Clinical and pathological information was retrieved from the medical records and telephone interviews, including complete medical history, physical examination, magnetic resonance imaging (MRI) of the brain, and survival. After surgery, all patients received radiotherapy and chemotherapy. Three patients experienced recurrence, and one of those patients died. Demographic data and tumor characteristics are indicated in Table 1. Progression-free survival time was defined as the duration between initial treatment and identification of recurrence. Overall survival time was defined as the duration between initial treatment and date of death.
This retrospective study was conducted in accordance with the Declaration of Helsinki, with approval from the ethics committee at Kyoto University Hospital. Written consent to use stored specimens was obtained from all living patients or their legal surrogates.

Immunohistochemistry and immunohistochemical assessment
Five-micrometer thickness of FFPE tissue sections were baked at 60˚C for 30 min, deparaffinized in xylene, and rehydrated in graded concentrations of ethanol. Heat-induced antigen retrieval was carried out in Tris-ethylenediaminetetraacetic acid buffer (pH 9.0) by microwaves, then endogenous peroxidase activity was quenched by incubating in 3% hydrogen peroxide at 37˚C for 30 min. Monoclonal primary antibodies (anti-PD-1, 1/100, #329911, Biolegend, San Diego, CA; anti-PD-L1 [8-28], 1/100, ab205921, Abcam, Cambridge, UK; anti-CD3, 1/200, ab16669, Abcam; anti-CD4, 1/100, ab133616, Abcam; anti-CD8, 1/100, ab17147, Abcam; anti-Foxp3, 1/50, ab22510, Abcam; anti-OCT4, 1/100, ab181557, Abcam) were applied at 4˚C overnight, then rewarming was performed at 37˚C for 30 min, followed by incubation Twenty-five consecutive intracranial germinoma patients were reviewed, including 22 primary cases and 3 recurrent cases. These included a pair of primary and with biotinylated secondary antibodies (EnVision™+ Dual Link System-HRP; Dako, Santa Clara, CA) at 37˚C for 30 min. To observe specific antibody localization, 3,3'-diaminobenzidine (DAB) (ImmPACT DAB Peroxidase Vector1; Vector Laboratories, Burlingame, CA) was used as the substrate and hematoxylin was used as a nuclear counterstain. Human placental tissue served as a positive control for PD-1 and PD-L1 antibodies, while human tonsil tissues served as positive control for CD3, CD4, CD8, and Foxp3 antibodies. For co-staining of PD-L1 and OCT4, 2 different peroxidase substrates were used (DAB for OCT4 and VEC-TOR1 SG [Vector laboratories, Burlingame, CA] for PD-L1) and no counterstain was performed. For negative controls, slides were incubated with PBS in replacement of primary antibodies. Sections were examined and scored by two independent observers who were blinded to the clinicopathological background of patient samples. Sections for which evaluations differed were re-evaluated simultaneously by observers using a double-headed microscope and a consensus opinion was reached. All samples were stained at the same time.

Quantitative valuation of TILs
Quantification of TILs was undertaken by counting immunopositive cells in serial sections stained for lymphocyte differentiation antigens with the use of photomicrographs. CD3+, CD4+, CD8+, and Foxp3+ TILs were evaluated: 5 tumor areas with abundant TILs in a 400× magnification (high-power fields [HPF]) were pictured with the use of cellSens Standard 1.9 (Olympus, Tokyo, Japan). Enumeration of TIL were carried out using Qupath software, following the instruction on the official website (https://qupath.github.io/) [18]. TIL density was calculated as the mean number of counted cell numbers divided by field area (350 μm × 260 μm).

Statistical analysis
Spearman correlation was applied to analyze correlations between two ordinal parameters. Unpaired Welch's t test was used to compare group differences, as appropriate. A linear regression model was fitted for trend analysis. A two-tailed significance level P < 0.05 was applied. All statistical analyses were performed with Statistical Package for the Social Sciences (SPSS) version 20.0 software (SPSS, Chicago, IL).

Highly frequent expressions of PD-1 and PD-L1 in intracranial germinomas
We found prominent PD-1 and PD-L1 expressions to variable extents in primary intracranial germinoma tissues. Representative immunohistochemical images for PD-1 and PD-L1 in intracranial germinoma are shown in Fig 1. In these tissues, all markers showed membrane-accentuated expression, also often accompanied by cytoplasmic expression. To visualize tumor cells, germinoma marker OCT4 was co-stained with PD-L1 in all cases [16].

TIL subtype densities in recurrent germinomas
During follow-up of 6 months to 31.5 years, 3 patients (Cases 14-16) were refractory to treatment and experienced recurrence, and Case 15 eventually died from the tumor. Moderate and high expressions of PD-1 and PD-L1 were observed in these 3 recurrent patients. In the dead case (Case 15), relatively high CD3+, CD4+, CD8+, and Foxp3+ TIL densities were observed (Fig 5A). MRI showed dissemination of germinoma at the final recurrence (Fig 5B). In all recurrent cases, high expressions of PD-1 and PD-L1 were detected. No difference of CD3+, CD4+, CD8+, and Foxp3+ TIL densities were found between initial and recurrent germinoma (Fig 5C).

Associations between clinicopathological parameters and PD-1, PD-L1, expression and TIL subtype densities
We performed univariate analysis with expressions of PD-1 and PD-L1 and clinicopathological parameters in the 22 primary cases. Detailed results are shown in Table 3. We identified correlations between high expression of PD-1 in TILs and high serum β-hCG levels (cutoff, 100 mIU/mL; n = 18 for β-hCG < 100 mIU/mL and n = 4 for β-hCG ! 100 mIU/mL;   [20]. Consistent with the current study, Fankhauser et al. reported frequent expression of PD-L1 in testicular seminoma (73%), a kind of non-brain germ cell tumor with identical histology to intracranial germinoma [17]. Intracranial germinoma is generally considered as the counterpart of "seminoma" in the testis and "dysgerminoma" in the ovary, constituting the neoplastic counterpart of the primordial germ cell, just as embryonal carcinoma is the neoplastic counterpart of totipotent embryonal stem cells and teratoma the neoplastic counterpart of the derivatives of the three embryonic germ layers [21]. However, little is known of the exact genetic correlation between germinoma and seminoma. Cierna et al. reported similar PD-L1 expression in testicular seminoma with a frequency of 76% [22]. Frequency of PD-L1 expression in our cohort was comparable with that in other brain tumors such as glioblastoma (88.0-100%) [14,23], anaplastic astrocytoma (71.4%) [14] and medulloblastoma (56.3%) [8]. In many types of non-brain tumor, frequent tumoral expression of PD-L1 was also reported with tumors such as ovarian carcinoma (86.7%) [24], non-small cell lung cancer (53%) [25], and colorectal cancer (36%) [26], renal cell carcinoma (32%) [27] and melanoma [28]. The Expression of PD-1/PD-L1 axis in intracranial germinoma discrepancies between PD-L1 expression in germinomas among studies might be attributable to the use of primary antibodies from different manufacturers or different immunoreactivities between the manual operation in our study and the automated staining system used in the study by Aoki et al. Taken together, previous investigations into the same category of pathology and our study of 24 patients offer positive confirmation of the frequent expression of PD-L1 in intracranial germinoma. The histology of intracranial germinoma is characteristic, exhibiting numerous TILs among large tumor cells, as the so-called "two-cell pattern", suggesting a peculiar hostimmune response to the tumor. In our cohort, numerous TILs were also found in all 25 samples of intracranial germinoma tissue (Fig 2). The majority of TILs have been reported to be T-lymphocytes. In 1979, Neuwelt and Smith reported T-lymphocyte membrane marker as present in 51% of small cells in germinoma tissues [29]. Saito et al. also reported that 70-80% of TILs were T lymphocytes, infiltrating either diffusely or in clusters, whereas 20-30% of TILs were B lymphocytes that tended to cluster in tumor tissues [30]. TILs have recently received close attention in terms of anti-tumor immune function. Since functions of immune (C) Comparison of TIL subtype densities between initial and recurrent germinoma tissues. No difference of CD3+, CD4+, CD8+, and Foxp3+ TIL densities was found between initial and recurrent cases. All P values were calculated according to the un-paired Welch's t test.
https://doi.org/10.1371/journal.pone.0194594.g005 Expression of PD-1/PD-L1 axis in intracranial germinoma checkpoint signals, such as the PD-1/PD-L1 axis, were discovered in cancer, most studies have focused on correlations between PD-1, PD-L1 and TILs using semiquantitatively analyses, but limited studies have reported the details of infiltrating T-lymphocyte subtypes, particularly in central nervous system tumors. Berghoff et al. reported that, in glioblastoma, PD-1+ TIL density correlated positively with CD3+ TIL density and CD8+ TIL density [23]. Our previous study revealed that high expression of PD-L1 was associated with reduced infiltration of CD8 + T cells and poor prognosis in medulloblastoma [8]. The current study analyzed the density of TIL subtypes with quantified data. As a result, expression of PD-1 in TILs correlated positively with CD8+ TIL and CD3+ TIL densities. In addition, expression levels of PD-1 in TILs did not influence CD4+ TIL density. However, the underlying mechanisms require further research for interpretation. Interactions between immune and tumor cells play cardinal roles during malignant progression [31]. Furthermore, dysfunction of the anti-tumor immune system may result in the progression of germinoma to the development of symptoms. Intracranial germinomas arise mostly in the neurohypophysis (30.1%) and pineal regions (51%), but infrequently in the basal ganglia (3.3%), cerebellopontine angle (2.6%), lateral ventricle (2%), cerebellum (2%) and multiple sites (8.5%) [32]. Germinoma cells actually show the malignant characteristic of infiltration into brain parenchyma, especially around the subventricular regions, and display recurrence at different areas and dissemination into the CSF space. Intracranial germinomas are highly radiosensitive and potentially curable using radiotherapy alone [33,34]. As well documented in the literature, CNS radiation for pediatric patients leads to dysfunctions in growth [35], the endocrine milieu [36], and neurocognition [37]. A reduced dose of radiotherapy to the whole ventricular area with chemotherapy has been introduced into treatment and has yielded promising therapeutic responses [6, 38,39]. However, the side effects of radiotherapy remain inevitable, even when dose and target volumes have been reduced. Meanwhile, clinical trials to substitute radiation with chemotherapy have resulted in increased relapse rates and inferior outcomes, and chemotherapy alone is thus currently not an acceptable treatment option [40,41]. Despite these efforts to establish reduced radiotherapy protocols in combination with chemotherapy, 10-20% of intracranial germinomas are bound to recur after first-line treatment and no standard second-line treatments have been defined [6, [42][43][44]. To conquer radiation-related side effects as well as recurrent tumors, a major task is to develop a novel therapeutic option for germinoma. Our results suggest immune checkpoint therapy as one possible option. Antibodies targeting checkpoint molecules such as PD-1 have already shown therapeutic effects on non-small cell lung cancer and melanoma, and PD-L1 expression exhibited a positive correlation with response to PD-1 inhibition in those tumors [45,46]. Preliminary efficacy of PD-1 antibodies was reported in seven patients with relapse of extracranial GCT after high-dose chemotherapy and stem cell transplantation [47]. Three of those patients received PD-1 antibodies for at least 6 months and long-term tumor response was achieved in two of the three patients, with tumor tissues showing highly positive results for PD-L1 staining. This report encourages examination of the efficacy of immune checkpoint therapy for intracranial germinoma. Given the existence of blood-brain barrier (BBB), the penetrance of PD-1 antibodies into CNS via BBB is not fully known. However, the BBB disruption in brain neoplasms could be observed with a marked heterogeneity [48]. In addition, radiotherapy, a standard therapeutic approach in intracranial germinomas, could further facilitate the penetrance of activated anti-tumor immune cells [49] and possibly the access of PD-1 antibodies. Actually, many clinical trials are ongoing to explore PD-1 in patients with primary (NCT02017717) or metastatic (NCT02320058) brain tumors. However, planning a suitable clinical trial is difficult, because intracranial germinoma is rare, and relapsed cases are even rarer. An international study for this context is thus desirable to develop novel approaches to the treatment of intracranial germinoma.

Conclusion
In the present study, expressions of PD-1 and PD-L1 were detectable in 96% (24/25) and 92% (23/25) of germinoma tissues, respectively. PD-1 expression in lymphocytes was associated with Foxp3+ TIL density. Increases in PD-1+ TILs was associated with accumulation of CD8 + TILs, which might be caused by immune dysfunction. Induction of Foxp3+ TILs from CD4 +TILs causes immune tolerance. These results intimate the breakdown of the tumor immune system that might be induced by PD-1/PD-L1 axis in intracranial germinoma, suggesting the potential efficacy of immune checkpoint therapy.