1. Regulation and Function of PD-L1 Expression
PD-L1 (B7-H1, CD274) is a type 1 transmembrane glycoprotein encoded by the CD274 gene. PD-L1 is highly expressed on tumor cells, as well as on hematopoietic cells, including macrophages, dendritic cells, mast cells, neutrophils, myeloid-derived suppressive cells (MDSCs), platelets, and T and B lymphocytes, and on non-hematopoietic cells, such as epithelial cells, endothelial cells, mesenchymal stem cells, placental trophoblasts, and many others
[1]. Expression of PD-L1 is regulated at the transcriptional, post-transcriptional, and post-translational levels
[2][3][4][5]. At the transcriptional level, PD-L1 expression is modulated by JAK/STAT1, NF-kb, and other transcriptional factors. Additionally, glycosylation and acetylation of PD-L1 are two important post-translational regulators of PD-L1 expression. For example, acetylation of PD-L1 impacts the translocation of PD-L1 from the cytoplasm to the nucleus, whereas glycosylation of PD-L1 can enhance its ability to interact with PD-1 to increase anti-cancer T cell immune responses
[6][7]. Expression of PD-L1 is also impacted by protein degradation pathways such as autophagic degradation and ubiquitination. PD-L1 degradation pathways not only affect the efficacy of anti-PD-(L)1 immunotherapy, but also affect its pharmacokinetics and pharmacodynamics—parameters that are critical for the timing of sampling of PD-L1 for biomarker assessment
[8][9][10]. The regulation of PD-L1 expression differs significantly between different cancer types, as well as between tumor cells and non-tumor cells due to variation of aberrations of activation of many signaling pathways such as RAS/RAF/ERK, PI3K/AKT, and JAK/STAT3, among others
[3]. Of note, gene amplification, gene translocation, copy-number gains, or copy-number loss also occur in tumor cells
[9][11]. PD-L1 copy number changes are associated with PD-L1 expression in several types of cancers such as NSCLCs, urothelial carcinoma, and breast carcinoma, among others. However, PD-L1 gene amplification is not associated with PD-L1 expression when tumor tissues are assessed by IHC in a subset of Hodgkin lymphoma patients
[12][13][14][15].
The major function of PD-L1 is suppression of T cell activation via extrinsic binding to its cognate receptor PD-1 expressed on T cells and subsequent downregulation of T cell-mediated anti-tumor responses in general, thereby resulting in a concomitant inhibition of T cell-mediated anti-tumor responses. Recently, intrinsic PD-L1 signaling has been shown to also play a critical role in promoting tumor progression, metastasis, immune evasion, and inducing responsiveness to anti-PD-(L)1 immunotherapy via a PD-1 dependent or independent means
[16][17][18][19]. The cell-intrinsic PD-L1 signaling can be modulated by: (
1) its post-translational modifications such as acetylation or glycosylation, or by forming cis-heterodimers (i.e., the ligand binding to receptors expressed on the same cell), such as PD-L1/PD-1 or PD-L1/CD80 heterodimers (CD80 being a shared receptor for CTLA4 and CD28); (
2) PD-L1 signaling in cancer cells triggered by engagement of PD-1; and (
3) PD-L1 binding to co-receptor integrins. From the perspective of biomarker identification, the cell-intrinsic PD-L1 signaling dictates what assays should be used for testing PD-L1 expression. For example, anti-glycosylated PD-L1 specific antibody needs to be used for testing expression of glycosylated PD-L1 on cells, whereas an intracellular staining protocol is needed to assess acetylated PD-L1 located in the cell’s nucleus.
PD-L1 can be cleaved by cell surface proteases such as A Disintegrin. Additionally, Metalloproteinase 10 and 17 (ADAM10 and ADAM17) and this cleaved form of PD-L1 (cPD-L1) can be detected in the serum or plasma of patients with NSCLC, melanoma, and many other cancers
[20][21][22]. Additionally, several PD-L1 splice variants (sPD-L1) can also be detected in plasma of patients with NSCLC and other cancer types
[23][24][25][26]. In addition, the expression levels of PD-L1 in serum or plasma or other components detected in blood include circulating tumor cells (CTCs), exosomes, and various immune cells that have been assessed and proposed to be a predictive biomarker for melanoma, NSCLC, gastrointestinal cancer, breast cancer, and many others
[27][28][29][30][31][32][33].
2. PD-L1 on Circulating Tumor Cells
Circulating tumor cells (CTCs) are rare cancer cells found in the blood that are derived from solid tumors. The rationale for using PD-L1 positive CTCs (PD-L1
+CTCs) as a potential biomarker include: (
1) PD-L1 expressed on CTCs plays an important role in tumor progression and metastasis, as well as resistance of anti-PD-(L)1 immunotherapy
[34][35]. (
2) CTCs derived from primary tumors or metastatic lesions can be detected in blood, providing the opportunity that assessment of the levels of PD-L1 expression in CTCs might recapitulate the PD-L1 expression pattens in tumor tissues
[1]. This is important since that it has been shown that the levels of PD-L1 expression are not equal between primary tumors and metastatic tumors in many patients
[36]. PD-L1 expression in metastatic tumors, but not in primary tumors, is associated with poor prognosis of melanoma patients
[37]. (
3) The number of CTCs has been shown to be a prognostic biomarker for NSCLC and many other cancers and is also associated with the tumor responses to anti-PD-(L)1 immunotherapy
[38][39]. Finally, (
4) many assays that are currently available to assess the PD-L1
+CTCs are imperfect and are facing many technical challenges (discussed below). Therefore, assessment of the number of PD-L1
+CTCs as a biomarker is currently being explored in NSCLC, melanoma, and many other cancers
[33][40][41][42].
The association between the number of baseline PD-L1
+CTCs and the efficacy of anti-PD-(L)1 immunotherapy is inconclusive and conflicting. Several studies demonstrate that NSCLC patients with high baseline PD-L1
+CTCs respond worse to treatment with nivolumab or pembrolizumab
[40][42][43][44][45][46]. Similar findings are also reported in melanoma and other types of cancers. However, other studies indicate that baseline numbers of PD-L1
+CTCs are not associated with patient responsiveness to nivolumab, including NSCLC patients. In addition, some studies demonstrate that the number of PD-L1
+CTCs is correlated with PD-L1 expression in primary tissues, whereases no correlation was found in other studies
[40][43][45][47][48]. Moreover, the detection rates of PD-L1
+CTCs and its cut-off value in many types of cancers including lung cancer vary dramatically, ranging from less than 30% to 90% across different studies
[39][43][47][48][49][50]. The number of PD-L1
+CTCs from different patients within the same study also vary dramatically, ranging from one cell to hundreds of CTCs in blood samples from prostate cancer patients, or from one cell to 20 in patients with colorectal cancer
[47][48][51]. The median percentage of PD-L1 positive cells from different tumors also varies significantly
[52].
The inconsistency and heterogeneity of PD-L1
+CTCs highlight the challenge of reconciling the results of these clinical studies. Among several factors that may contribute to the inconsistency, some of them are listed here. (
1) Different CTC isolation platforms were used in these studies. It has been shown that the CTC isolation yield differs significantly when different platforms are used
[53][54]. Although the CellSearch system (an affinity-based assay for enrichment of EpCAM positive tumor cells) is considered the gold standard CTC detection platform
[55], many challenges remain in isolating the rare and highly heterogeneous CTC population
[56][57][58]. For example, the CellSearch system cannot be used for isolating EpCAM negative CTCs due to epithelial-to-mesenchymal transition
[59]. The procedures and challenges of isolating CTC have been comprehensively reviewed elsewhere
[53][58][60][61]. (
2) The described studies are retrospective analyses with low statistical power due to the limited sample size. (
3) The results from tumor biopsies demonstrate that PD-L1 is a not a universal biomarker for different lines of therapy. For example, PD-L1 is a predictive biomarker for first-line atezolizumab treatment for NSCLC patients, but the anti-cancer efficacy is not associated with PD-L1 expression for second-line atezolizumab treatment
[1][62][63]. However, many studies combined mixed lines of treatments, including first-line to multiple line treatment, making it difficult to interrogate data from these studies. (
4) The predictive effect of PD-L1 is significantly different among different drugs and cancer types. For example, PD-L1 is a predictive biomarker for first-line and second-line pembrolizumab treatment for NSCLC patients, but not for nivolumab treatment in the same setting
[1]. Several studies mentioned above included patients treated with different anti-PD-1 therapeutic antibodies, making it difficult to interrogate data from these studies as well.
Advances in understanding the biology and regulation of PD-L1 might provide additional opportunities to explore the potential use of PD-L1
+CTCs as a biomarker. These options include: (
1) testing PD-L1
+CTCs from different sources such as obtaining tissues from primary tumors, draining lymph nodes, or tumor metastases. It has already been shown that PD-L1 expression can vary significantly between these different cell sources
[60][64][65]. The functions and heterogeneity of PD-L1
+CTCs suggest that different sources of PD-L1
+CTCs may be used as a biomarker. (
2) Testing post-translational modifications of PD-L1, such as acetylation or glycosylation of PD-L1, or its nuclear localization. It has been shown that nuclear expression of PD-L1 may be associated with a poorer prognosis in late stage colorectal or prostate cancer, whereas the nuclear expression is not associated with CTCs. In addition, it has been shown that levels of deglycosylated PD-L1 may have a better predictive value for anti-PD-(L)1 immunotherapy than that of glycosylated PD-L1 when tumor biopsy samples were used
[66][67][68][69]. (
3) CTCs can form CTC clusters. These clusters have shown more metastatic potential than single CTCs in breast cancer and other types of cancers and play a role in cell survival and immune evasion
[70][71]. Higher levels of CTC clusters are associated with a worse prognosis in breast cancer patients
[72]. In addition, CTCs have also been shown to form clusters with a subset of neutrophils, resulting in the promotion of cancer cell metastasis. Overall, increased levels of CTC–neutrophil clusters are associated with a poorer prognosis in breast cancer patients
[73], suggesting that the PD-L1 positive CTCs or CTCs/immune cell clusters might also have a potential to be a biomarker for anti-PD-(L)1 immunotherapy
[74][75]. (
4) Combining the use of PD-L1
+CTCs with other biomarkers of immune checkpoint receptors. The formation of cis-PD-L1/CD80 heterodimers significantly impacts the efficacy of anti-PD-(L)1 immunotherapy, suggesting that co-expression of CD80 and PD-L1 in CTCs might be a useful biomarker
[76]. However, it is unlikely that combining PD-L1
+CTCs with the approved biomarkers TMB-H or MSI-H would have a predictive value, as multiple studies, based on tumor biopsy results, have demonstrated that PD-L1 is an independent biomarker from TMB-H or MSI-H
[77][78][79][80]. Finally, (
5) PD-L1 gene amplification is associated with patient responsiveness to anti-PD-(L)1 immunotherapy
[15][81], and vice versa
[82]. Therefore, the level of PD-L1 gene expression in CTCs could be a useful biomarker and is worth pursuing.
3. PD-L1 on Circulating Immune Cells
Numerous efforts have been made to explore whether PD-L1 expression on peripheral immune cells can be used as a biomarker for anti-PD-(L)1 immunotherapy. Several lines of evidence support PD-L1 expressed on circulating immune cells as a biomarker for anti-PD-(L)1 immunotherapy. First, nearly all types of immune cells participate in tumor development, progression, metastasis, and resistance to anti-cancer therapies including anti-PD-(L)1 immunotherapy
[83][84][85][86][87][88][89][90]. Second, the levels of PD-L1 positive immune cells alone in tumor tissues have prognostic value for many cancer types such as NSCLS, melanoma, renal carcinoma, and many others
[91][92]. Moreover, the levels of PD-L1 expression on tumor-infiltrating immune cells or in combination with PD-L1 expression on tumor cells as assessed by IHC are associated with efficacy of atezolizumab or pembrolizumab treatment
[1]. Third, the increased microvascular permeability in the tumor tissue promotes the immune cells trafficking to and from the blood stream, suggesting that PD-L1-positive immune cells from tumor tissues can be detected in blood
[93][94][95]. Supporting this, it has been shown that T cells expanded in the tumor microenvironment are present in peripheral blood mononuclear cells (PBMCs)
[96].
The current paradigm regarding the mechanisms underlying anti-PD-(L)1 immunotherapy is that the effective anti-tumor T cell responses are dependent on activation of the dysfunctional tumor-infiltrating CD8
+ T cells in the tumor microenvironment. Accumulating evidence demonstrates that systemic immunity (defined here as activation of immune responses independent of activation of tumor-infiltrating immune cells) has emerged to be critical for immune-mediated tumor eradication
[83][97]. Spitzer et al. demonstrate in a mouse tumor model that activation of T cells in the draining lymph node, bone marrow, and blood contributes to tumor eradication that is independent of the anti-tumor immune response in the tumor microenvironment
[98]. Supporting this notion, analyses by single-cell RNA sequencing for T cell receptors in tumor tissues demonstrate that T cells in the tumors are replenished by T cells activated outside of the tumor tissues
[96]. Lastly, activated T cells or other immune cells can be found in many organs, including skin, liver, and heart, among others, in patients who develop irAE. Although the origins of these immune cells are not clear, it is possible that activated Immune cells from these organs might also be detected in the peripheral blood
[99][100][101]. Nonetheless, testing for PD-L1 expression on immune cells in peripheral blood is an attractive approach, as PBMCs might contain both tumor-infiltrating immune cells as well as immune cells activated in the periphery by anti-PD-(L)1 immunotherapy
[102][103][104].