Over the past decade, the advent of immunotherapy has changed the paradigm of cancer treatment. Many therapeutics targeting programmed cell death protein-1 (PD-1) or its ligand (PD-L1) have been approved by regulatory agencies as monotherapy or in combination with chemotherapy and other targeted drugs. PD-L1 immunohistochemistry (IHC) and tumor mutational burden (TMB) are the main predictive biomarkers discovered to date in clinical studies, however, PD-L1 IHC remains the primary adjunct diagnostic test used in clinical practice to inform immune checkpoint inhibitor (ICI) therapy.
PD-L1 is expressed on the surface of tumor cells and tumor-infiltrating immune cells, and studies of multiple tumor types have identified an association between PD-L1 levels and the clinical benefits of single-agent anti-PD-1/PD-L1 therapy. While patients with high PD-L1 expression generally reap the greatest clinical benefits, patients with low or no PD-L1 expression still benefit, prompting further exploration of the causal relationship between evolving cancer genomic changes and infiltrative immune cell subtypes.
Considering the limitations of PD-L1 as a single biomarker, we need to understand the complex tumor microenvironment (TME) in order to develop better biomarkers and diagnostics, predict clinical benefit more accurately, and, crucially, generate an effective immune response to tumors in a more targeted manner.
Predictive biomarkers
PD-L1 IHC
Unlike mutational binadiness, PD-L1 is a linear biomarker and can therefore be assessed at multiple expression thresholds. Due to the complex biology of PD-L1 expression on tumor cells and immune cells, approved PD-L1 IHC assays employ not only different expression thresholds, but also different scoring methods. The PD-L1 scoring algorithm has analytical specificity, indications, dependence, and varies according to the treatment route. Some assays score only tumor cell expression (ignoring immune cells), while others only assess the PD-L1 of immune cells (ignoring tumor cells), while others measure both tumor and immune cell PD-L1. The table below shows cut-offs for the primary PD-L1 IHC analysis used in non-small cell lung cancer (NSCLC). It is important to note that the approvals and uses of these validated cut-offs are country-specific.
PD-L1 expression is dynamic, exhibiting significant differences in tumor type, tumor site, and even within the same tumor specimen. The expression of PD-L1 in tumor cells is regulated by genomic, epigenetic, and transcriptional mechanisms. The strong correlation between PD-L1 on tumor cells and the degree of CD8+ T cell infiltration in NSCLC is most likely due to transcriptional regulation in responses to inflammatory signals, such as IFN-γ, which are generated by active anti-tumor immune responses.
Non-uniform expression of PD-L1, which is often confined to areas of immune infiltration, suggests that PD-L1 is adaptively induced due to high local concentrations of relevant cytokines within TME and further highlights heterogeneity within tumors. The intratumoral heterogeneity and plasticity of PD-L1 may influence patient management decisions, and in addition, false-negative PD-L1 IHC results may be obtained due to sampling. The location of the biopsy and the treatment that has been performed are important parameters for PD-L1 expression. Obtaining a new biopsy at the start of treatment may not be feasible for clinical reasons, but due to the plasticity of the immune response, particularly PD-L1, archival tumor specimens do not necessarily reflect TME in recurrent or metastatic lesions.
PD-L1 is expressed by a variety of immune cell types, primarily macrophages, dendritic cells (DCs), as well as T cells, NK cells, and B cells. It is unclear which specific immune populations or spatial features of immune cells are most clinically relevant. Further characterization of the spatial interactions of tumor cells with specific immune cell subsets using multiplex IHC techniques is needed to determine whether immune cells need to continue to be classified into a broader class, which requires more nuanced and complex analysis.
Other biomarkers
Clinical trials have shown that tumors with a large number of effector T cells are more sensitive to ICI therapy. IFN-γ-centric gene expression signatures have been used to show a correlation between overall response rate (ORR) and progression-free survival (PFS) in melanoma patients treated with the anti-PD-1 antibody pembrolizumab. Consistent with these findings, the IMpower150 trial in NSCLC reported a T effector cell (Teff) gene expression profile consisting of PD-L1, CXCL9, and IFN-γ that was beneficial for PFS when comparing atezolizumab in combination with chemotherapy versus chemotherapy alone.
In addition, tumor patients with high mutational burden have greater clinical benefit from CI treatment. Concordance studies of TMB and PD-L1 IHC appear to suggest that these biomarkers are orthogonal and independently predict the benefit of CI treatment. As a result, pembrolizumab recently received regulatory approval for the treatment of unresectable or metastatic high TMB solid tumors that have progressed on prior therapy and for which there are no alternative treatment options.
As research into more predictive biomarkers continues to advance, there is a need to move away from a single analytical approach and integrate the various biological processes responsible for carrying out the immune response. Preliminary exploratory analyses suggest that combined biomarkers can be translated into clinical benefits. In the Checkmate-026 and Checkmate-227 studies, patients whose tumors express high levels of PD-L1 and have a high tissue TMB status benefit most from treatment with the anti-PD-1 antibody nivolumab. Similar results were reported in the OAK study.
Phenotypic diversity of immune effector cells
Many studies have described the relationship between tumor-infiltrating lymphocytes and their spatial distribution and therapeutic benefit. Identification of the spatial distribution of various immune effector cells needs to be complemented by functional analyses such as activation markers and antigen reactivity, and by analyzing the intrinsic antitumor responsiveness of CD8+ T cell TCR sequences in tumor samples, suggesting that the ability to recognize autologous tumors is limited to a small number of tumor-infiltrating CD8+ T cells. To complicate matters further, there may be significant TCR heterogeneity in CD4+ and CD8+ T cell populations in different regions of the tumor (central versus marginal invasion), and the degree of heterogeneity has prognostic significance.
In general, three distinct populations of CD8+ T cells can be identified based on a common transcriptional profile:
primitive or memory cells expressing CC-chemokine receptor 7 (CCR7) and transcription factor 7 (TCF7);
穿孔素1(PRF1)、颗粒酶A(GZMA)和GZMB阳性的细胞毒性细胞;
A "dysfunctional" diverse cell population characterized by markers associated with a state of exhaustion, such as PD-1 (PDCD1), LAG3, and TIM3 (HAVCR2).
The relative proportions of these three populations appear to be highly variable, not only in different tumor types, but also in tumors of the same histology. Transcriptional profiling and TCR sequencing information have been used to determine whether the phenotype of CD8+ cells within the tumor is altered. Preliminary models suggest that TME-induced phenotypically heterogeneous dysfunctional cell populations evolved from naïve CD8+ cells, which produce cytotoxic cells in a TME-independent manner. The extent to which cells in the "dysfunctional" cell bank can transform into a "cytotoxic" state is currently unknown, however, understanding this is important for designing therapies that facilitate and trigger this transition.
In addition, intratumoral CD8+ T cells with specificity for tumor-expressed neoantigens have been identified in several studies, but they are only a small fraction of the total phenotypically heterogeneous population. Interestingly, CD39 appears to be able to distinguish between tumor-specific (high CD39) and bystander (low CD39) CD8+ T cells. Other phenotypic and functional analyses showed that CD103 was co-expressed by CD39+ subsets as well as PD-1, TIM3, and CTLA-4, and these markers are often associated with depletion status. Co-expression of CD39 and CD103 relies on prolonged stimulation by TCR and exposure to TGF-β. Interestingly, high levels of CD39+ CD103+ double-positive cells appear to translate into survival benefits in patients with squamous cell head and neck cancer.
Spatial distribution of immune effector cells
Recent studies have analyzed the spatial relationship between CD8+ T cells and MHCII-expressing cells, the most likely antigen-presenting cells. In patients with renal cell carcinoma, TCF1+/CD8+ T cells preferentially colocalize to the region of MHC class II expressing cells, while TCF1−/CD8+ cells are diffuse throughout the tumor. The incidence of TCF1+/CD8+ cells also correlated with the incidence of DCs, whereas TCF1−/CD8+ cells did not exhibit this correlation. Interestingly, tumors containing TCF1+/CD8+ T cells and high-density regions expressing MHC class II APCs have better clinical efficacy, and APC-dense regions may provide an intratumoral environment for stem cell-like CD8+ T cell differentiation into effector cells capable of maintaining anti-tumor immune responses.
It is evident that tumor infiltration of mature, active dendritic cells into the tumor bed increases the recruitment of immune activation and active immune effector cells. Accurate identification and quantification of DCs is of great value, however, markers such as CD11c, CD11b, CD163, and MHC-II are not DC-specific and are expressed on other cell types such as macrophages. The observation that DCs can promote or inhibit nascent immune responses further complicates the fact that there is currently no effective marker to distinguish between the two DCs.
During tumor progression, TME transforms into an environment that actively protects tumor cells from immune attack, and TME cells that are thought to have tumor-protective functions include T regulatory cells (Tregs), tumor-associated macrophages (TAMs), tumor-associated fibroblasts (CAFs), and myeloid suppressor cells (MDSCs).
Tumor protection is often mediated by secreted cytokines and chemokines that prevent immune effector T cells from invading the tumor bed or rendering these cells functional, e.g., TGF-β. Combination therapy with TGB-β inhibitors and anti-PD-L1 antibodies can lead to intratumoral penetration of Teff, changing the phenotype from "exclusion" to "inflammatory", reducing metastases and improving long-term survival.
MDSCs are a heterogeneous population of neutrophils and monocyte-like myeloid cells and are increasingly recognized as key mediators of immunosuppression in various cancers. In cancer patients, an increase in the number of circulating MDSCs is associated with advanced clinical stage, increased incidence of metastatic disease, and immunosuppression. In addition to immunosuppressive functions, MDSCs can actively shape the tumor microenvironment through complex crosstalk with tumor cells and surrounding stroma, thereby increasing angiogenesis, tumor invasion, and metastasis.
Intrinsic resistance of tumor cells
At present, there are still many gaps in our understanding of the different responses and resistance mechanisms of tumor immunotherapy. While we hope to find a panacea, the possible truth is that there are many mechanisms that contribute to the generation of resistance. While mutations in tumor suppressors (e.g., p53) increase the mutational burden, help amplify the number of neoantigens, and make tumors more sensitive to the immune system, these mutations can also drive mutations in antigen presentation pathways and key cellular mechanisms that trigger transcriptional and metabolic changes that in turn have indirect effects on tumor-infiltrating immune cells.
A key component of the tumor antigen presentation mechanism is the MHC class I molecule, which consists of the HLA class I (HLA-I) subunit and β2 microglobulin (β2M). Downregulation of MHC-I is present in a variety of tumor types and may occur through a variety of mechanisms, including genetic mutations, epigenetic silencing, transcriptional changes, and post-translational modifications.
How does the down-regulation of MHC-I affect TME? The study found that the positive expression of HLA-I was related to the immune infiltration density, but not to the PD-L1 status. All HLA-I+/PD-L1+ tumors have a high degree of CD8+ T cell infiltration, and HLA-I deletion is associated with a reduced number of T lymphocytes within the tumor, whose space is confined to the stroma surrounding the tumor. On the other hand, HLA-I-/PD-L1- tumors are larger and have a lower density of CD8+ T cells. This study demonstrates that the tumor immune escape phenotype combines two independent immune escape mechanisms: loss of HLA-I and upregulation of PD-L1.
Experience gained in neoadjuvant therapy
Recent explorations of immunotherapy for early-stage disease have provided new opportunities to understand TME and how it changes in treatment response. Neoadjuvant studies are particularly rich because diagnostic biopsies are often collected prior to treatment and can be compared to surgical resection performed immediately after neoadjuvant therapy.
The ABACUS trial is a single-arm phase 2 study investigating neoadjuvant therapy with atezolizumab prior to cystectomy. Elevated PD-L1 levels (SP142; IC≥5%), baseline Teff gene expression, and intraepithelial CD8+ T cells were associated with clinical benefit. Interestingly, an increase in the number of intraepithelial CD8+ T cells and PD-L1+ immune cells was observed in post-treatment tumor samples compared to preoperatively, and this increase was more pronounced in responders than in patients with relapse. Elevated TGF-β, fibroblast activating protein, and cell cycle genes were found to be associated with drug resistance in post-treatment samples.
PURE-01 studies reported similar results in patients with bladder epithelial cancer who received three cycles of pembrolizumab prior to radical cystectomy. Patients with elevated PD-L1 IHC (22C3; CPS≥10%) achieved a higher pathologic complete response rate compared with those with lower PD-L1 status. Resected PD-L1 CPS levels were increased and TMB levels decreased after treatment compared to pre-treatment biopsies. In addition, there was an increase in immune-related genes, including genes related to IFN-γ signaling, antigen presentation, and T cell function, in post-treatment samples compared to pre-treatment samples.
The NABUCCO study evaluated the effects of ipilimumab and nivolumab prior to resection for stage III urothelial carcinoma. Compared to the ABACUS and PURE-01 studies, treatment outcomes were not related to baseline status of intratumoral CD8+ T cell infiltration or Teff signaling. Conversely, the study observed a trend between elevated TMB and increased frequency of DDR gene alterations in patients who achieved a complete response. Patients who were unable to obtain a complete response were enriched with TGF-β gene expression signatures in baseline samples, suggesting a possible mechanism of resistance. In addition, the correlation between TIL density and response was studied, and interestingly, there was no correlation between baseline TIL and efficacy, but TIL enrichment was observed in patients who acquired CR after treatment.
These observations are not unique to bladder cancer. The NEOSTAR trial evaluated nivolumab or nivolumab plus ipilimumab as neoadjuvant therapy for patients with operable NSCLC. Patients who obtained an imaging or major pathologic response generally had higher PD-L1 than the baseline tumor sample. The frequency of TIL, tissue-resident memory T cells, TEFF, and effector memory T cells was higher after nivolumab plus ipilimumab treatment compared with nivolumab alone.
Overall, baseline PD-L1 appears to be a predictor of response to ICI. The likes of TMB, TIL, and DDR also appear promising as biomarkers, but further evaluation and validation are needed. Differences in treatment regimens, as well as heterogeneity in patient populations, may lead to inconsistencies in the results of different trials. With neoadjuvant immunotherapy and chemo-immunotherapy combinations gaining regulatory approval, identifying validated biomarkers to predict immune response is critical to aid in treatment decisions.
As the transition from chemotherapy to targeted therapy and immunotherapy continues to evolve, researchers will employ more complex clinical trial designs, such as combinations of targeted therapy and immunotherapy, combinations of different CI drugs, and a greater focus on the design of biomarkers that are more sensitive in predicting response or resistance to single- or multi-agent treatment regimens.
In the future, it is likely that a multiparametric approach will be needed to identify patients most likely to respond to ICI therapy, and it is important to develop robust tissue-based composite techniques. This helps to characterize TME more readily and to analyze interactions between tumors and effector immune cells, as well as between effector cells and APCs. The expression profile of the marker of interest may indicate the activation, inhibition, or exhaustion status of effector cells and suggest mechanisms of resistance.
Copyright Notice
The copyright of this article belongs to "Xiaoyao Says Medicine".