Multiplex Immunohistochemistry Applies to Examination of PD-1/PD-L1 on Immune Cells for Prognosis Prediction in Non-Small Cell Lung Cancer


 Background: Tumor cells expressing programmed cell death 1 (PD-1) and programmed cell death ligand 1 (PD-L1) correlate with a better prognosis of immunotherapy in non-small cell lung cancer (NSCLC) patients. Expression of PD-1 and PD-L1 on immune cells is also concerned by more and more researchers.Methods: This study included 174 patients with NSCLC, and collected from the month of December in 2012 to April 2019. Formalin-fixed paraffin-embedded (FFPE) samples from NSCLC patients were performed by multiplex immunohistochemistry (IHC) staining using CD8, CD57, CD68, CD163, PD-1 and PD-L1. Marker localization included each type of immune cell subset with PD-1 or PD-L1 was quantified and analyzed.Results: The present study revealed distribution characteristics of PD-1 and PD-L1 on CD8+ T cells, CD57+ NK cells, CD68+ macrophages and CD163+ M2 macrophages in NSCLC patients using multiplex IHC, which indicated that expression of PD-1 was higher on CD8+ T cells and expression of PD-L1 was higher on CD8+ T cells and CD68+ macrophages. Immune clustering analysis showed that the low immune feature group displayed more survival rate than the high group. The reason was due to higher ratios of CD8/PD-L1 in the low group compared with the high group in the NSCLC cohort. Further, the Kaplan-Meier analysis of survival rate according infiltration of different immune cells also indicated that low CD57+ NK cells and low CD68+ macrophages were associated with a higher survival rate. The similar results were observed in the Kaplan-Meier analysis of expression of PD-1 and PD-L1 on immune cells.Conclusions: Taken together, we displayed the expression characteristics of PD-1 and PD-L1 on tumor-infiltrating immune cells and revealed that high expression of PD-1 and PD-L1 on immune cells was associated with poor survival rate. The present study provided further evidence to better guide clinical treatment in NSCLC.


Background
Lung cancer is one of the most common malignant tumors in the world and one of the leading causes of cancer death [1,2]. In the past decade, immune checkpoint inhibitors (ICIs) for non-small cell lung cancer (NSCLC) have been made signi cantly progress, especially the programmed cell death 1 (PD-1) and programmed cell death ligand 1 (PD-L1) [3]. PD-1 and PD-L1 antibodies or inhibitors have been approved for patients with advanced NSCLC who do not respond to platinum-based chemotherapy [4,5]. Numerous studies indicate that biomarker screening can effectively select patients suitable for immunotherapy [6][7][8][9][10].
In the tumor microenvironment, interaction between PD-1 and PD-L1 can cause apoptosis of anti-tumor in ltrating immune cells [11,12]. Tumor cells expressing PD-L1 can escape the surveillance of the immune system, especially when PD-1 is expressed on immune cells, thereby promoting tumorigenesis and progression [13]. Some studies have observed that PD-1 could be detected on T cells, macrophages and natural killer (NK) cells [14][15][16]. PD-1, as an immunosuppressive receptor, is expressed only on the surface of activated cells in vivo, but not on resting T cells [13,17]. Thus, expression of PD-1 may indicate immune cell activity. In addition, several studies have found that expression of PD-1 on CD8 + cells could be used to predict the effectiveness of ICI treatment in patients with NSCLC [18][19][20]. The choice of immunotherapy is mainly based on the expression level of PD-L1 in tumor cells in NSCLC, which is closely related to the prognosis [21][22][23]. Recent studies indicated that expression of PD-L1 has been found not only on tumor cells, but also on immune cells, including cytotoxic T cells, NK cells, macrophages and B cells [10,[24][25][26]. Thus, expression of PD-L1 on immune cells, as a potential predictive biomarker, is concerned by more and more researchers [27,28].
In the present study, we aimed to investigate distribution characteristics of PD-1 and PD-L1 on immune cells including CD8 + T cells, CD57 + NK cells, CD68 + macrophages and CD163 + M2 macrophages in NSCLC patients using multiplex immunohistochemistry (IHC) and explore the predictive role of PD-1 or PD-L1 expression on different immune cells for prognosis in NSCLC.

Patients and clinical data
This study included 174 patients with NSCLC, and collected from the month of December in 2012 to April 2019. This study was performed in accordance with the Declaration of Helsinki. And it was approved by the Internal Review and the Ethics Boards of the Yuebei People's Hospital of Shaoguan. Most of patients underwent chemotherapy or targeted therapy or chemotherapy combined with targeted therapy. The clinical and pathological data, including age, gender, smoking history, pathologic stage and tumor histology, was collected for analysis after patient consent (Table 1).

Statistical analysis
We compared differences between two groups using the Mann-Whitney U test (unpaired, nonparametric, two-tailed). For every feature using X-tile software (version 3.6.1; Yale University School of Medicine, New Haven, CT) de ned the optimum cutoff score based on the association with the patients' survival rate. In univariate analysis for different variable values, survival curves were performed using the Kaplan-Meier method and were compared using log rank test. Statistical analysis was generated using GraphPad Prism (version 7.01), all tests P < 0.05 was considered signi cant. In multi-factor analysis, we applied random forest trees for analysis (python software).

Distribution characteristics of PD-1 and PD-L1 on immune cells
The NSCLC cohort contained 174 cases, including 138 adenocarcinoma (ADC) cases and 36 squamous cell carcinoma (SQCC) cases ( Table 1). The age, gender, smoking history pathologic stage and tumor histology were indicated in Table 1. The pathologic stage of these patients mainly concentrated in stage III and IV and most of patients underwent chemotherapy or/and targeted therapy. We rst explored distribution characteristics of PD-1 and PD-L1 in different immune cells in 174 NSCLC cases. As shown in Fig. 1A and 1B, PD-1 or PD-L1 expression was detected on CD8 + T cells, CD57 + NK cells, CD68 + macrophages and CD163 + M2 macrophages and displayed a similar tendency in stroma area (SA) and tumor area (TA). Expression of PD-1 was higher on CD8 + T cells and expression of PD-L1 was higher on CD8 + T cells and CD68 + macrophages in SA and TA ( Fig. 1A and 1B). Representative multiple IHC images were showed about expression of PD-1 on CD8 + T cells (Fig. 1C) and expression of PD-L1 on CD8 + T cells (Fig. 1D) and CD68 + macrophages (Fig. 1E). These images also indicated high expression of PD-1 and PD-L1 on CD8 + T cells and high expression of PD-L1 on CD68 + macrophages.
Expression of PD-1/PD-L1 and in ltration of immune cells predicted the prognosis We divided 174 NSCLC cases into high immune feature group and low immune feature group according to immune clustering molecules ( Fig. 2A). Speci cally, the clustering molecules included the positive rate of T cells (CD8 + cells), NK cells (CD57 + cells), macrophages (CD68 + cells), M2 macrophages (CD163 + cells) and of each immune subpopulation expressing PD-1 or PDL-1 in the TA and SA. As shown in Fig. 2A, the high and low groups contained 100 and 74 NSCLC cases, respectively. According to the high and low groups, we found that the low group displayed more survival rate than the high group (P = 0.0081) (Fig. 2B).
Because of high expression of PD-1 and PD-L1 on CD8 + T cells in Fig. 1, we analyzed ratios of CD8/PD-1 and CD8/PD-L1 in the TA and SA between the low and high immune feature groups. The results indicated that the ratio of CD8/PD-1 was not signi cant difference in the TA and SA between the low and high groups ( Fig. 2C and 2D). However, the ratio of CD8/PD-L1 in the low group was signi cantly increased in TA and SA compared with the high group ( Fig. 2E and 2F). These data showed that expression of PD-L1 on CD8 + T cells in the low group was low, which might be one of important reasons to bene t from survival. Random forest tree analysis further showed that ve indicators contributed to the most survival analysis were CD68 + PD-L1+, CD57 + PD-L1+, CD8 + PD-L1+, CD68 + in SA and CD68 + PD-L1 + in TA (Fig. 2G). Among them, CD68 + PD-L1 + and CD57 + PD-L1 + have been reported in some literatures and could be used as a screening biomarker for lung cancer immunotherapy [10,25]. These results indicated that the immune cells expressing PD-L1 could affect the clinical outcome of patients with NSCLC.
In ltration of different immune cells predicted the prognosis in the NSCLC cases We further explored effects of different immune cells on prognosis in the NSCLC cohort using X-tile software to de ne the optimum cutoff value for CD8+, CD57+, CD68 + and CD163+. The results indicated that low CD57 + NK cells (Fig. 3A) and low CD68 + macrophages (Fig. 3B) in SA and TA showed a higher survival rate compared with the high group (Table S1). But high/low groups of CD8 + T cells (Fig. 3C) and CD163 + M2 macrophages (Fig. 3D) according to the optimum cutoff value did not show signi cant difference (Table S1). The data suggested that NSCLC patients could bene t from low CD57 + NK cells and CD68 + macrophages.

Expression of PD-1/PD-L1 on immune cells predicted the prognosis in the NSCLC cases
We next analyzed expression of PD-1 or PD-L1 on immune cells to predict the survival in the NSCLC cohort. According to the optimum cutoff value (Table S2), low expression of PD-1 on CD8 + T cells (Fig. 4A) and CD57 + NK cells (Fig. 4C) in SA was associated with a higher survival rate. And low expression of PD-L1 on CD8 + T cells (Fig. 4B) and CD57 + NK cells (Fig. 4D) in SA and TA showed a better survival rate (Table S2). Interestingly, on CD68 + macrophages, low expression of PD-1 or PD-L1 in either SA or TA displayed a bene cial prognosis ( Fig. 4E and 4F, Table S2). But on CD163 + M2 macrophages, low expression of PD-1 or PD-L1 in SA was bene cial for survival in NSCLC patients ( Fig. 4G and 4H, Table S2). Taken together, we provided further demonstration which PD-1 and PD-L1 expression on immune cells as predictive biomarkers was applied to prognosis analysis in NSCLC patients using multiple IHC detection.

Discussion
In recent years, immunotherapy has made great progress in the treatment of NSCLC, especially the PD-1 and PD-L1 inhibitors. In order to more accurately screen patients who could bene t from immunotherapy, many studies have reported some biomarkers related to the e cacy of immune checkpoint inhibitors [6-9, 18, 29]. Among them, PD-L1 expression on tumor cells was the most widely used. However, several studies have found that some patients whose tumor cells without PD-L1 expressing could also achieve durable response from immunotherapy [30,31]. Moreover, PD-L1 has been observed not only to be expressed on tumor cells, but also broadly expressed on immune cells, which have an important role in immunotherapy. In addition to the expression of PD-L1, the tumor microenvironment was also inextricably linked to the e cacy of ICI [32], and the tumor-in ltrating lymphocytes (TILs) could effectively stratify patients with NSCLC [33,34]. Therefore, a better understanding of the expression of PD-1 and PD-L1 distribute over different cells and their impact on clinical outcomes, which could provide insights for the treatment of NSCLC patient. To achieve this objective, we rst conducted a detailed retrospective study of the tumor tissue of 174 patients with NSCLC. The results indicate that immune cells expressing PD-1 or PD-L1 were able to predict the clinical prognosis of patients with NSCLC.
A large number of mechanistic research data showed that PD-1 was expressed primarily on T cells and PD-L1 was mainly expressed on tumor cells. Combination of PD-1 and PD-L1 results in tumor immune evasion. In the present study, we found that PD-1 and PD-L1 were expressed on T cells, NK cells and macrophages using multiplex IHC detection. The average positive rates of PD-1 expression on CD8 + T cells, CD57 + NK cells, CD68 + macrophages and CD163 + M2 macrophages were 14.00%, 4.17%, 5.96% and 9.61% in SA, and 12.27%, 2.64%, 3.07% and 4.89% in TA, respectively. The average positive rates of PD-L1 expression on CD8 + T cells, CD57 + NK cells, CD68 + macrophages and CD163 + M2 macrophages were 13.30%, 5.61%, 12.23% and 8.5% in SA, and 6.86%, 2.88%, 6.22% and 2.5% in TA, respectively. PD-L1 distribution in immune cells was mainly expressed on macrophages and T cells. This type of macrophages (CD68 + PD-L1+) were probably related to the promotion of immune escape and belonged to tumor associated macrophages (TAM) [35]. However, there were few studies about PD-L1 expression on T cells, which is needed further explore.
According to the positive rates of immune pro ling, the NSCLC cohort were divided into a strong immune feature group (high) and a weak immune feature group (low). Interestingly, we observed a better survival rate in patients from the low group, suggesting that tumor microenvironment in the low group may be biased towards the activated state, while the high group was the opposite. The ratio analysis of CD8/PD-1 and CD8/PD-L1 in TA and SA revealed that the low group had higher ratio of CD8/PD-1 and CD8/PD-L1, especially the CD8/PD-L1 ratio was signi cantly different between two groups. Meanwhile, multivariate analysis of random forest trees in two groups indicated that CD68 + PD-L1+, CD57 + PD-L1+, CD8 + PD-L1+, CD68 + in SA and CD68 + PD-L1 + in TA had a major effect on survival rate. These results demonstrated that more immune cells expressing PD-1 or PD-L1 in the high group cause the suppression of immune cells, leading to decrease of survival rate.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. SupplementaryMaterial.docx