In the present study, we have found lower expression of CD44, a surface marker of CSC, in lung adenocarcinoma. Moreover, we analyzed the association between PD-L1 and CD44 in lung adenocarcinoma. Indeed, there have been studies investigating the association between PD-L1 and CSC in a multitude of cancers. Upregulated PD- L1 expression were reported to be found in breast and colon CSCs [15]. One study showed significant correlation between PD-L1 expression and CSC markers Oct4A, Nanog and BMI1 in a large breast cancer data set. It has also shown that two of the CSC markers, Oct4A and Nanog, could be induced by ectopic PD-L1 expression whereas downregulation of PD-L1 expression led to attenuated self-renewal ability of breast CSCs [16]. Another study has shown that PD-L1 expression was correlated with CSC markers and in chemoresistant colorectal cancer specimens [17]. One study led by Dong also revealed that stemness of tumor cells could be boosted by PD-L1 [18]. CD44 is involved in multiple signaling functions, cell proliferation, apoptosis, survival, migration and differentiation. Studies have also determined the functional role of CD44 in cytokine production and secretion and deemed it a typical surface marker in lung adenocarcinoma [19]. In the present study, we have found that CD44 was positively correlated with PD-L1 expression (r=0.46, P=2.86e-28) in lung adenocarcinoma using TIMER database. Their positive association was further validated in tumor specimens among patients with lung adenocarcinoma (r=0.267, P=0.02). Our finding on the association between PD-L1 and CSC was consistent with these previous reports.
In the current study, we further analyzed the microenvironment surrounded by CSC. To this end, we classified lung adenocarcinoma into six discrepant subtypes, namely wound healing, IFN-gamma dominant, inflammatory, lymphocyte depleted, immunologically quiet and TGF-beta dominant. Each of the six subtypes represents a specific microenvironment. The niche where CSCs reside is quite complex. It contains a collection of heterogeneous cells, which may result in different subtypes in one cancer [20-22]. Tumor with the same type may be inconsistent in immune-competence among different individuals. Immuno-competence, to some extent, partially reflects the microenvironment in which the tumor is involved. As our results clearly demonstrated, CD44 expressions were relatively higher in C3 (inflammatory), C4 (lymphocyte depleted) and C6 (TGF-beta dominant) subtypes.
Since tumor microenvironment is composed of large amounts of immune cells, we next detect CSC’s association with a bunch of immune cells. In the current study, B cell, CD8+ T cell, CD4+ T cell, macrophage, neutrophil and dendritic cell (DC) were all analyzed for their correlation with CSC, as reflected by CD44, in lung adenocarcinoma. B cells positively modulate immune responses and inflammation through antibody production and to promote T-cell activation and proliferation through antigen presentation. CD4+ T cells have been reported to contribute to the microenvironment remodeling required for sustained tumor regression. It is known that neutrophils in tumors often predict worsened outcomes. Dendritic cells (DCs) are reported to be responsible for the balance between CD8+ T cell immunity and tolerance to tumor antigen. Tumor-associated macrophages (TAM) can enhance tumor cell invasion and metastasis [23-27]. There have been studies reporting the interaction between CSCs and immunes cells, as mediated by growth factors and cytokines [28,29]. For example, one study recently demonstrated that TAM could create a niche suitable for CSCs. One study has proven that tumor associated macrophages (TAMs) increase the number, tumorigenicity and drug resistance in CSCs through STAT3 activation. In turn, CSCs induce M2 phenotype in TAMs and block anti-tumor CD8+ responses during chemotherapeutic treatment [30]. One study showed that CSC-pulsed DCs induced the antigen-specific TH1 immune response [31]. In this study, we have found that CD44 expression was correlated with T cells CD4 memory resting and mast cells resting, using GEO database GSE103584. Moreover, in the analysis of TIMER, we have detected positive association between immune cells (B cell, CD8+ T cell, CD4+ T cell, macrophage, neutrophil and dendritic cell) with CSC, as detected and manifested by CD44.
Finally, we analyzed the role of CD44 in affecting OS in lung adenocarcinoma patients. Kaplan Meier analysis demonstrated that CD44 as a negative role in affecting OS and Cox regression analysis proved it served as a negative prognostic factor. Interestingly, it has to be noted that although CD44 was linked with increased infiltration of immune cells whereas found to be a negative biomarker associated with worsened OS in lung adenocarcinoma patients. We assume the following reasons may be possible: 1) Despite the infiltration of numerous immune cells surrounding CSC in lung adenocarcinoma, these immune cells do not necessarily exert anti-tumor activities. For instance, tumor associated neutrophils and TAM exert pro-tumor effects that would boost the malignant properties. Their pro-tumor effect outweighed the anti-tumor effect, which would lead to increased malignancy. 2) Since CSC was associated with PD-L1 accumulation in lung adenocarcinoma in the present study, the inhibition of anti-tumor effect exerted by PD-L1 could possibly blunt the suppressive function of immune cells on tumor cells. Studies have view PD-1/PD-L1 axis as a major pathway exerting immune-inhibitory effect, triggering a suppressive microenvironment that protects cancer cells from immune destruction [32]. Moreover, Schatton et al reported that CSCs could downregulate T cell activation [33], which may also serve as one of the explanations for CD44’s negative role in predicting OS in lung adenocarcinoma.
We have first conducted the study to ascertain the association between CD44 and PD-L1 in lung adenocarcinoma. In addition, we tested the infiltration of immune cells surrounding CSC. Furthermore, the role of CD44 in predicting OS in patients with lung adenocarcinoma was also analyzed. We have demonstrated that CD44 is positively correlated with PD-L1 expression, immune cells infiltration and serves as a negative prognostic biomarker in lung adenocarcinoma. We therefore speculate that lung adenocarcinoma patients with higher CD44 expression may be surrounded by robust immune cell infiltration whereas these surrounding cells do not exert proper anti-tumor effects against CSC. We propose that lung adenocarcinoma patients with higher PDL1 expressions may evade immune cells attack. Undeniably, some points could be improved in our study. For instance, T cells and macrophages could be classified more intensively to be indicative of subtypes and activation status.
Notably, our finding is of some clinical relevance. Currently, there have been many clinical trials of immunotherapeutic approaches targeting CSCs. For instance, CSC-DC vaccine, aiming to inhibit immune suppression induced by CSC, is under investigation [34].