The study population was patients who were diagnosed with PCNSL between January 2009 and February 2017 and registered for our prospective cohort studies after providing written informed consent (NCT00822731 and NCT01877109). In our prospective cohort studies, we collected serum samples and the pre-treatment characteristics of patients at diagnosis. Treatment and outcome-related data, including treatment regimens, tumor response, date of progression, and date of death, were regularly updated. These cohort studies were approved by the Institutional Review Board of Samsung Medical Center, and all investigations were conducted according to the principles expressed in the Declaration of Helsinki and its contemporary amendments. Because patients with all subtypes of lymphoma were enrolled, the evaluations for work-up and treatments were performed according to our clinical practice for each subtype. For patients with PCNSL, the initial evaluation was done according to the International Primary CNS Lymphoma Collaborative Group recommendations . Cerebrospinal fluid (CSF) analyses and ophthalmic examinations were also performed in most patients to test for leptomeningeal and ocular invasion. As the primary treatment for newly diagnosed PCNSL, HD-MTX-containing chemotherapy with or without WBRT was used. Response was assessed according to the response criteria for PCNSL recommended by the International Primary CNS Lymphoma Collaborative Group : complete response (CR) was defined as no contrast enhancement in brain magnetic resolution imaging (MRI) and negative findings in ocular and CSF examinations; partial response (PR) was defined as at least a 50% decrease in the enhancing tumor lesion; progressive disease (PD) was defined as at least a 25% increase in the lesion or any new lesion in the CNS or systemic sites; and stable disease (SD) was defined as less than a PR but not PD. Response evaluation was performed after the completion of primary treatment chemotherapy, and surveillance brain MRI was done to monitor the occurrence of disease relapse.
We retrospectively analyzed 68 patients who had archived serum samples available for measurement of sPD-L1 among patients enrolled in the aforementioned cohort studies, after excluding patients with secondary CNS involvement in systemic DLBCL. Using serum samples and ELISA, we first measured the sPD-L1 levels and correlated them with the clinical and pathological characteristics of the patients at diagnosis. Then, response to first-line therapy and the survival outcomes of patients were compared according to the level of sPD-L1. Second, we analyzed the expression of PD-L1 in tumor cells and non-tumor cells in 52 patients whose paraffin-embedded tissue blocks were available for immunohistochemistry analyses. Third, we measured serum cytokines using multiplex ELISA to explore additional biomarkers that might predict the outcomes of PCNSL patients and influence the level of sPD-L1 or the tissue expression of PD-L1. To confirm the DLBCL histology of our patients with PCNSL, two pathologists (I.C and Y.K) reviewed patients’ histopathology slides using the 2017 World Health Organization classification . Relapsed disease was defined as disease recurrence in patients who had no evidence of disease after cessation of therapy, and PD was defined as SD or PD during the primary treatment. Multiple diseases were defined as more than one lesion found in a radiologic evaluation, and deep regions of the brain were defined as the basal ganglia, brainstem, periventricular regions, and cerebellum. We updated the survival status in March 2019 for the survival analysis, and this study was approved by the Institutional Review Board of Samsung Medical Center (IRB No. 2019-05-054).
Measurement of serum sPD-L1
Serum samples were collected at diagnosis and stored at -80°C until analysis. Serum aliquots had not been previously thawed before use in our multiplex chemokine assay. The level of sPD-L1 was measured using ELISA kits (PDCD1LG1 ELISA kit, USCN Life Science, Wuhan, China) according to the manufacturers’ instructions. Briefly, the microplate provided in the kit was pre-coated with an antibody specific to PDCD1LG1. Standards or samples were then added to the microplate wells with a biotin-conjugated antibody specific to PDCD1LG1. Next, avidin conjugated to horseradish peroxidase was added to each microplate well and incubated. After the enzyme-substrate reaction, the color change was measured spectrophotometrically at a wavelength of 450nm. To estimate the reference ranges of sPD-L1, we measured the levels of serum sPD-L1 in 12 normal individuals (6 males and 6 females, median age 51 (range 24 – 78). They voluntarily donated residual serum samples that were left after blood tests during their regular health check-up. The sPD-L1 values in the blood serum specimens of healthy controls were determined by the same method. The measurement of each sample was done in duplicate.
Immunohistochemistry for tissue PD-L1 expression
Immunohistochemistry was performed on paraffin tissue sections (4-μm thick), and the PD-L1 antibody (Spring Bioscience, CA, USA; clone SP142, M4421, rabbit anti-human PD-L1/CD274, monoclonal antibody, 1:25 dilution) was used to assess the expression of PD-L1. The antibody was incubated for 120 min at 37°C using the Ventana BenchMark XT platform after antigen retrieval for 92 min with CC1 buffer. Signal visualization was done using the OptiView DAB immunohistochemistry detection kit (Ventana, Tucson, Azusa) and OptiView Amplification kit (Ventana, Tucson, Azusa). Tonsil squamous epithelium was used as a PD-L1 immunohistochemistry positive control . The slides were semi-quantitatively analyzed by two pathologists (I. C and Y. K). The extent of PD-L1 expression in tumor cells was defined as the proportion of tumor cells showing PD-L1 expression with any intensity in the tumor area . Macrophages and lymphocytes infiltrating the tumor area were considered non-tumor immune cells, and the proportion of PD-L1 expression in them was assessed in the same manner as in the tumor cells. Tumor cells were discriminated from tumor infiltrating lymphocytes using morphology because the tumor cells had unequivocal morphologic characteristics that allowed discernment. The assessment of PD-L1 expression in tumor infiltrating macrophages was done by measuring PD-L1 expression in CD68-positively stained macrophages. PD-L1-positive tumor cells were defined as those positively stained for PD-L1 with a distinct membranous, cytoplasmic, or punctate/granular pattern of any intensity based on previously published descriptions [23, 24]. The following additional antibodies were used to assess CD68 expression and identify the cell of origin: CD68 (Leica Biosystem, Newcastle, NCL-L-CD68, mouse monoclonal, 1:50 dilution), CD10 (Novocastra, Newcastle, NCL-L-CD10-270, mouse monoclonal, 1:100 dilution), BCL6 (Novocastra, Newcastle, NCL-L-Bcl-6-564, mouse monoclonal, 1:80 dilution), and MUM1 (Dako, CA, M7259, mouse monoclonal, 1:100 dilution). To assess the positivity of Epstein-Barr virus (EBV) in tumor tissue, EBV-encoded RNA (EBER) in situ hybridization (ISH) was also performed because EBV-positivity could be associated with PD-L1 expression. EBER was detected using ISH and an EBV ISH kit (Leica Microsystems, Bannockburn, IL, USA). We used EBV-negative lymphoid tissues and the hybridization mixture without EBV oligonucleotides as negative controls.
Multiplex cytokine assay
We measured eotaxin-1, GROα, interferon (IFN)-α, IFN-γ, IL-1α, IL-1β, IL-1RA, IL-2, IL-4, IL-5, IL-6, IL-7, IL-8, IL-9, IL-10, IL-12p70, IL-13, IL-15, IL-17α, IL-18, IL-21, IL-22, IL-23, IL-27, IL-31, interferon γ-induced protein (IP-10), monocyte chemoattractant protein 1 (MCP-1), macrophage inflammatory protein-1α (MIP-1α), MIP-1β, regulated on activation T cell expressed and secreted (RANTES), stromal cell-derived factor 1α (SDF1α), tumor necrosis factor (TNF)-α, and TNF-β levels in duplicate with a ProcartaPlex™ multiplex immunoassay kit (Invitrogen, Camarillo, CA, USA) and the Bio-Plex Cytokine Assay System (Bio-Rad Laboratories, Hercules, CA, USA) according to the manufacturer’s instructions.
In the survival analysis, overall survival (OS) was designated as the time from the date of diagnosis to the date of death or last follow-up and progression-free survival (PFS) was designated as the time from the date of diagnosis to the date of progression or relapse, death, or last follow-up. The optimal cutoff value for the survival analysis was determined using the receiver-operating characteristic (ROC) curve method. Kaplan-Meier survival graphs and log-rank tests were used for the univariate survival analyses, and Cox proportional regression models were used for the multivariate analyses. A bi-variate analysis of sPD-L1 and tissue expression of PD-L1 was done to analyze their correlation. The chi-square test or Fisher's exact test were used to analyze their associations with clinical and pathological characteristics. In all comparisons, P-values less than 0.05 were considered statistically significant, and all P-values correspond to two-tailed significance tests. Statistical analyses were carried out using SPSS software version 21.0 (IBM, Armonk, New York, USA).