The impact of the molecular prole of the tumor microenvironment on the prognosis of NSCLC

Purpose The present study was performed to clarify the correlation between macrophages, tumor neo-vessels and programmed cell death-ligand 1 (PD-L1) in the tumor microenvironment (TME) and the clinicopathological features of non-small cell lung cancer (NSCLC) and to explore the prognostic factors of stromal features in NSCLC. Methods Tissue microarrays containing 92 NSCLC patients were studied with immunohistochemistry (IHC). The distribution and quantitative data of CD68-and CD206-positive tumor-associated macrophages (TAMs) in tumor islets and tumor stroma, and the expression of tumor neo ‐ vessels and PD-L1, were analyzed by inverted microscopy and Image-Pro Plus 6.0 software. Prognostic analyses with the clinicopathological characteristics and tumor microenvironment features were performed. Results The number of CD68-positive macrophages in each location of the tumor islets and tumor stroma was signicantly higher than that of CD206-positive macrophages, and they were signicantly correlated (P < 0.0001). Survival analysis revealed that CD68- and CD206-positive TAMs in the tumor stroma and tumor islets were signicant prognostic factors (P < 0.05, respectively). Comprehensive analysis of CD206-positive stromal TAMs showed that CD105 and PD-L1 were signicant prognostic factors (P=0.045). Moreover, CD68-positive TAMs in tumor islets and the expression of PD-L1 were independent predictors of poor prognosis for NSCLC. Conclusion Thus, the key elements in the tumor microenvironment, including tumor neo ‐ vessels, macrophages and PD-L1, were heterogenic in NSCLC tissues and had signicant roles in cancer invasion and metastasis. The combined analysis of key components in the tumor microenvironment was an important prognostic factor. CD206-positive of CD68-positive TAMs and CD206-positive TAMs in the tumor stroma with shorter OS, with the tumor stroma the most suitable intratumoral area for evaluating TAMs. study found of CD206-positive stromal TAMs, CD105 and PD-L1 had a certain relationship with prognosis. The number of positive cells in each (tumor islets and stroma) summed, and CD68-positive TAMs or CD206-positive TAMs were combined with the other two key components to analyze the results, results obtained. discrepant role and regulatory mechanism of TAMs in different interspaces may interact with cancer cells and together accelerate cancer invasion and metastasis. The present study implied that the different molecular proles of the tumor microenvironment were intimately linked with the prognosis of NSCLC patients. The combined analysis of key components in the tumor microenvironment was an independent prognostic factor, which showed the importance of comprehensively analyzing the tumor microenvironment.


Introduction
Lung cancer is the most common cause of cancer-related deaths, of which non-small cell lung cancer (NSCLC) comprises approximately 85-90%(Siegel et al. 2020). Despite the great progress of comprehensive treatment strategies based on surgery for resectable NSCLC in recent years, approximately 50% of earlystage NSCLC patients will relapse or develop distant metastases within 5 years after radical surgery (Asamura et al. 2008).
Paget's "seed-soil" theory makes people realize that the tumor microenvironment plays a vital role in tumor metastasis (Paget 1989  Immunohistochemistry (IHC) and Multiplexed Immuno uorescence Tissue microarrays containing 92 NSCLC patients were studied. Brie y, tissue microarrays were treated by depara nization in xylene, hydration with graded alcohol and subjected to antigen retrieval. Then, the tissue microarrays were placed in 3% hydrogen peroxide (H 2 O 2 ) for 10 min at room temperature to inactivate endogenous peroxidases. After washing three times in PBS, the slides were blocked with 2% bovine serum albumin (BSA, B2064, Sigma, USA) for 30 min at room temperature, followed by incubation with primary antibodies against PD-L1 (1:100, ab205921, Abcam, UK), CD105 (1:500, ab28364, Abcam, UK) and CD68 (1:200, ab34710, Abcam, UK) at 4 °C overnight. After washing with PBS, the slides were incubated with secondary antibody (1:200, ab150077, Abcam, UK) for 60 min at 37 °C. The slides were then washed in PBS three times, followed by Dako REALTM EnVisionTM (DAB, PW017, Sangon Biotech, China) detection and counterstaining with hematoxylin. Similar to IHC, after antibody incubation, all sections were covered using Fluoroshield containing 4′,6diamidino-2-phenylindole (DAPI, Abcam) for 10 min at room temperature to identify nuclei. Normal lung tissue was used as a positive control.
Quanti cation of immunohistochemical staining All slides were scanned with an Olympus BX51 microscope equipped with an Olympus DP72 camera (Olympus Optical Co., Ltd., Tokyo, Japan) and a CRi Nuance multispectral imaging system (Cambridge Research & Instrumentation, Inc., Woburn, MA, USA). Positive staining was indicated by brownish granules. Then, after obtaining the images of signal unmixing, TAMs were analyzed based on the expression of CD68 and CD206. For each slide, 3 high-power elds of the tumor islets and tumor stroma per tissue section were separately selected. Tumor islets were de ned as areas where tumor cells accounted for more than 70% of the total cells and tumor stroma as areas where tumor stromal cells accounted for more than 70% of the total cells (Li et al. 2018). The average number of these three elds represented the CD68-or CD206-positive cell number for each component of the tumor (Figure. 1A). Tumor neo-vessels marked by CD105 and tumor cells expressing PD-L1 were counted in six high-power elds selected at the tumor site, and the mean cell counts were documented. Image-Pro Plus 6.0 software was used to count positive cells. According to the HIS color selection scheme (H=0-30; I=0-255; S=0-255), de ne the area range and lter the below 50 pixels, which is nonspeci c positive color noise. To determine the density of in ltrating macrophages, tumor neo-vessel density and PD-L1 expression, the cut-off value to classify subgroups was according to the median of each value.
For the combined group, according to the expression levels of the three components, taking the median value of each component as the cutoff value, the expression of tumor neo-vessels, macrophages and PD-L1 could be divided into low-and high-density groups. For the combined group, according to the expression levels of the three key stromal components, patients were divided into three subgroups according to the density of TAMs, tumor neo-vessels and PD-L1: group 1, all components were expressed at a low level; group 2, one of components was expressed at a high level; and group 3, all components were expressed at a high level. The represented digital images were independently collected by two pathology investigators who were blinded to the clinicopathological characteristics of all tissue specimens.

Statistical analysis
Statistical analyses were performed with SPSS 25.0 (SPSS Inc., Chicago, IL, USA). For categorical data, χ2 test was performed. Spearman rank correlation analysis was used to analyze the correlation between macrophages, tumor neo-vessels and PD-L1 expression. Differences in the CD68-positive TAMs and CD206-positive TAMs among the groups were analyzed by the Mann-Whitney test. The Kaplan-Meier method was used to estimate the survival curve for OS, and the log-rank test was used to assess the difference in survival between groups. The Cox regression model was used to perform univariate and multivariate analyses. A two-tailed P<0.05 was considered statistically signi cant.

Results
Distribution and expression of CD68-positive TAMs, CD206-positive TAMs, CD105 and PD-L1 in NSCLC In the low and high CD68-positive TAM subgroups, the high tumor neo-vessel density cases were 20 (43.5%) and 26 (56.5%), respectively. In the low and high PD-L1 expression subgroups, high densities of CD68-positive TAMs were observed in 18 cases (40.9%) and 26 cases (59.1%), respectively. In the low and high CD206-positive TAM subgroups, 21 (45.7%) and 25 (54.3%) patients had high tumor neo-vessel density, respectively. In the low and high PD-L1 expression subgroups, there were 15 (33.3%) and 30 (66.7%) cases with high CD68-positive TAM density, respectively. Of note, tumor neo-vessels, CD68-positive TAMs and PD-L1 expression were not signi cantly correlated with any of the clinicopathological characteristics, which indicated that these key components of the tumor microenvironment were independent of clinical features, including tumor size, tumor histological type, degree of differentiation, lymph node metastasis and tumor staging (Table. 1a). Overall, CD206-positive TAMs in tumor islets and stroma were signi cantly correlated with lymph node metastasis. However, there was no signi cant correlation between TAMs, which were positive for both indicators in any part, and the other clinicopathological factors of NSCLC (Table. 1b).  macrophages and PD-L1 indicated that the mortality risk in combined group 4 was signi cantly increased (P = 0.045, Figure 4D). As presented in Table 2, the factors with P ≤ 0.20 in the univariate analyses and the components of interest were entered into the multivariate analyses. Smoking status, differentiation degree, CD68-positive TAMs in tumor islets and PD-L1-positive cells were independent prognostic factors for OS (P<0.05 for all) ( Table 2).

Discussion
It is generally believed that cancer invasion, including NSCLC, is not merely a local problem but a multifactor and multistep continuum, with a variety of molecular dysfunctions and cell signaling dysregulations. Genetic or epigenetic changes in cancer cells are the initial factors driving carcinogenesis, and the responses of stromal cells in the tumor microenvironment may promote or regulate cancer invasion and metastasis, which ultimately results in an altered tumor microenvironment favoring cancer invasion and progression. In this study, CD68 was selected as the TAM marker. CD68 is commonly considered to be a pan-macrophage marker but cannot distinguish between M1 and M2 subtypes (Falini et al. 1993). M2 macrophages have a variety of surface markers, including CD163, CD204 and CD206. Among them, CD206 is known to be expressed on the surface of most classes of macrophages and dendritic cell subpopulations and is routinely used to identify the M2 phenotype (Gordon 2003). Recent studies on M2 macrophage markers (such as CD163 and CD204) have also shown that M2 macrophage density is more closely related to poor prognosis than CD68-positive TAMs (Komohara et al. 2014) . TAMs have different effects on the prognosis of different types of cancer. In NSCLC, the prognostic relevance of TAMs is still under debate. The reasons for the inconsistent reports may be related to the choice of markers, different statistical powers, and differences in evaluation modes. CD105 is an endoglin used to evaluate blood vessels such as CD31 and CD34 (total endothelial markers), while CD105 is considered to recognize only abnormal blood vessels induced by tumors (Weidner et al. 1992). To our knowledge, the present study is the rst to compare the TAM distribution using CD68 and CD206 in two intratumoral areas and to compare the distribution of TAMs, tumor neo-vessels and PD-L1 in NSCLC.
The analyzed results showed that the number of CD68-positive TAMs and CD206-positive TAMs was higher in the tumor stroma but lower in tumor islets, which was consistent with several previous reports (Dai et al. 2010;Li et al. 2018). We also found that there was a strong correlation between the distribution of CD68-positive TAMs and CD206-positive TAMs in tumor islets and tumor stroma. Furthermore, the mean numbers of CD68-positive TAMs in each location of the tumor islets and tumor stroma were signi cantly higher than those of CD206-positive TAMs. Univariate analysis showed that a large number of CD68positive TAMs and CD206-positive TAMs in the tumor stroma were associated with shorter OS, which was consistent with the results of Li Z et al (Li et al. 2018) and showed that the tumor stroma is the most suitable intratumoral area for evaluating TAMs. This study also found that the comprehensive analysis . We con rmed that PD-L1 expression was highly correlated with the prognosis of NSCLC and could be used as an independent prognostic factor for patients with NSCLC. PD-L1 mediates immunosuppressive signals. The results also showed that patients with low PD-L1 expression in tumor cells may have a longer OS. However, several studies suggest that PD-L1 overexpression achieves longer survival in early NSCLC (Cooper et al. 2015), breast carcinoma (Schalper et al. 2014), gastric cancer (Zheng et al. 2014) and colorectal cancer (Droeser et al. 2013) . Other studies have shown that there is no correlation between PD-L1 expression and OS (Sorensen et al. 2014) . Many recent studies have reported that high PD-L1 is associated with poor prognosis in NSCLC (Igawa et al. 2017;Keller et al. 2018;Li et al. 2019). In different studies, the de nition of PD-L1 positivity or high density was different, leading to di culties in concluding the relationship between PD-L1 expression and NSCLC prognosis. Evidence suggests that PD-L1 expression is actually an adaptive mechanism and may be a response of tumor cells to host immune pressure (Taube et al. 2012). It can also be understood that the expression of PD-L1 is related to the endogenous immune response, such as tumor in ltrating lymphocytes (TILs) in NSCLC and indoleamine 2,3-dioxygenase-1 (IDO-1) expressed by dendritic cells (DCs) (Mandarano et al. 2019). Any possible prognostic signi cance is not directly related to a single immune signal but to the overall balance between the host's antitumor immune response and tumor-mediated immunosuppression.
It has been reported that tumor cells can induce increased expression of M2 macrophages and PD-L1 (Wen et al. 2018). Our study demonstrated that the expression of PD-L1 in cancer cells was correlated with the density of macrophages. A Japanese study suggested that M2 macrophages were associated with PD-L1 expression on NSCLC cytotoxic T cells (Sumitomo et al. 2019) . There are also correlations in other solid tumors, such as gastric adenocarcinoma (Harada et al. 2018).
In conclusion, key components in the tumor microenvironment may interact with cancer cells and together accelerate cancer invasion and metastasis. The present study implied that the different molecular pro les of the tumor microenvironment were intimately linked with the prognosis of NSCLC patients. The

Informed consent
This retrospective study was approved by the Ethics Committee of Zhejiang Cancer Hospital. Informed consent was waived owing to the retrospective nature of this study.
(A) The evaluation of IHC staining. Representative CD68 IHC-stained slides scanned by a pathology digital imaging system at 4× ( ). For each slide, 3 representative 0.1 mm2 elds were separately selected for tumor islets marked with a red frame ( ) and tumor stroma marked with a black frame ( ). 20× (B) Immunostaining of TAMs in NSCLC with CD68 and CD206 antibodies. a-b: Cases with a high number of CD68+ TAMs and CD206+ TAMs in tumor islets. c-d: Cases with a low number of CD68+ TAMs and CD206+ TAMs in tumor islets. e-f: Cases with a high number of CD68+ TAMs and CD206+ TAMs in the tumor stroma. g-h: Case with a low number of CD68+ TAMs and CD206+ TAMs in the tumor stroma. 20× (C) The distributions of CD68+ TAMs and CD206+ TAMs in tumor islets and tumor stroma (above). The correlations between CD68+ TAMs and CD206+ TAMs in tumor islets and tumor stroma (below). (D) CD68+ TAMs and CD206+ TAMs in tumor islets and stroma with patient OS in NSCLC. P<0.05. (A) The evaluation of IHC staining. Representative CD68 IHC-stained slides scanned by a pathology digital imaging system at 4× ( ). For each slide, 3 representative 0.1 mm2 elds were separately selected for tumor islets marked with a red frame ( ) and tumor stroma marked with a black frame ( ). 20× (B) Immunostaining of TAMs in NSCLC with CD68 and CD206 antibodies. a-b: Cases with a high number of CD68+ TAMs and CD206+ TAMs in tumor islets. c-d: Cases with a low number of CD68+ TAMs and CD206+ TAMs in tumor islets. e-f: Cases with a high number of CD68+ TAMs and CD206+ TAMs in the tumor stroma. g-h: Case with a low number of CD68+ TAMs and CD206+ TAMs in the tumor stroma. 20× (C) The distributions of CD68+ TAMs and CD206+ TAMs in tumor islets and tumor stroma (above). The correlations between CD68+ TAMs and CD206+ TAMs in tumor islets and tumor stroma (below). (D) CD68+ TAMs and CD206+ TAMs in tumor islets and stroma with patient OS in NSCLC. P<0.05