Predictive biomarkers of anti-PD-1/PD-L1 therapy in NSCLC

Immunotherapy, especially anti-programmed cell death protein 1/programmed cell death ligand 1 (PD-1/PD-L1) treatment has significantly improved the survival of non-small cell lung cancer (NSCLC) patients. However, the overall response rate remains unsatisfactory. Many factors affect the outcome of anti-PD-1/PD-L1 treatment, such as PD-L1 expression level, tumor-infiltrating lymphocytes (TILs), tumor mutation burden (TMB), neoantigens, and driver gene mutations. Further exploration of biomarkers would be favorable for the best selection of patients and precisely predict the efficacy of anti-PD-1/PD-L1 treatment. In this review, we summarized the latest advances in this field, and discussed the potential applications of these laboratory findings in the clinic.


Background
Lung cancer has a high incidence rate worldwide and is the main cause of cancer deaths [1]. The 5-year survival rate varies in different regions [2]. Non-small cell lung cancer (NSCLC) accounts for approximately 80-85% of all lung cancers [3,4]. Recently, the anti-programmed cell death protein 1/programmed cell death ligand 1 (PD-1/ PD-L1) treatment has substantially changed the treatment patterns of NSCLC. The anti-PD-1/PD-L1 treatment with or without platinum-based chemotherapy has become the first-line strategy for NSCLC without driver gene mutations [5].
The anti-PD-1/PD-L1 treatment blocks the interaction of PD-1 and its ligands, interferes with inhibitory signal transduction, restores the vitality of T cells, and thereby restarts the anti-tumor immune effect [17,18]. NSCLC has high level of heterogeneity. The heterogeneity of molecular immune subtypes and immune microenvironment results in the differences in the efficacy of PD-1/PD-L1 inhibitors [19]. The low response rate to PD-1/PD-L1 inhibitors hinders the clinical application [20]. Therefore, it is urgent to find reliable biomarkers to effectively predict the efficacy of PD-1/PD-L1 inhibitors. In this review, we summarized the latest advances in Page 2 of 13 Niu et al. Exp Hematol Oncol (2021) 10:18 the predictive biomarkers of anti-PD-1/PD-L1 therapy in NSCLC.

PD-L1 expression level
A known mechanism for PD-1/PD-L1 to promote tumor immune escape is adaptive immune resistance [21]. Multiple clinical trials have been performed to evaluate the relationship between the expression of PD-L1 on tumor cells and the response rate to PD-1/PD-L1 inhibitors (Fig. 1). The high level of PD-L1 expression heralds the potential benefit of anti-PD-1/PD-L1 treatment [22,23].

Neoantigens
Neoantigens are derived from somatic mutation [35], which bind to major histocompatibility class I (MHCI) and are expressed on the surface of cancer cells. Neoantigens endow the tumor with high immunogenicity and induce anti-tumor immune response ( Fig. 1) [36]. Neoantigens are released by tumor cells and captured by professional APC, and then the effector T cells targeting cancer specific antigens are activated [37]. Activated T cells migrate and infiltrate into tumor bed, specifically recognize the antigens on tumor cells and kill cancer cells [37]. The tumor clones with potent immunogenicity are eliminated, and the cancer cells with weak immunogenicity escape immune surveillance [38]. Many studies proved that anti-PD-1/PD-L1 therapy combined with radiotherapy or oncolytic virus increased the release of neoantigens and amplified the specific immune response [39][40][41]. Compared with no durable clinical benefit (NDB) patients, DCB patients had higher burden of candidate neoantigens. High candidate neoantigen burden was associated with improvement in PFS (HR = 0.23, 95%CI: 0.09-0.58, p = 0.002) [42] ( Table 1). The efficacy of immunotherapy was not only related to the quantity of neoantigens, but also related to the quality of neoantigens [43]. High-quality neoantigens especially clonal neoantigens, could bind to multiple HLA alleles [43]. The clonal neoantigens promoted the activation and infiltration of neoantigen reactive T cells expressing high level of PD-1, and tumors enriched clonal neoantigens were more sensitive to PD-1 blockers [34]. The incidence rate of DCB in patients with high mutation burden and low neoantigen subclonal fraction was higher than patients with high subclonal neoantigen fraction or low clonal neoantigen burden (92 vs. 11%) [34] (Table 1). Immune elimination of neoantigen-containing tumor cell subpopulations and genetic events such as chromosomal deletions or loss of heterozygosity in tumor cells lead to the loss of neoantigens, which contribute to the emergence of acquired resistance to anti-PD-1/PD-L1 treatment [44].

Inflammation related genes
Some expression signatures reflect the inflammatory state of tumors, such as genes related to T cell activation, chemokine expression, and adaptive immune resistance ( Fig. 1) [64,65]. Patients with significantly elevated inflammatory profile scores tended to be sensitive to PD-1/PD-L1 inhibitors. Compared with non-responders, responders had significantly higher inflammation signature scores [65]. In addition, inflammation scores was correlated with epithelial-mesenchymal transition (EMT) scores. Thompson's study showed that the combination of EMT phenotypic feature scores and inflammation gene scores increased the accuracy of prediction [65]. Therefore, it is predicted that reversal of EMT may improve the resistance to anti-PD-1/PD-L1 therapy [65].
Further study found that in the same NSCLC cohort, the eight genes associated with antigen processing machinery (APM) scores could more effectively predict the efficacy than inflammation scores [66]. Also, our previous study indicated that some immune response-related signatures related to the efficacy of immune checkpoint inhibitor in lung adenocarcinoma [4].

microRNA(miRNA)
MiRNA modifies the expression of target genes by regulating protein translation [67]. miRNA dysregulation is closely associated with carcinogenesis and can promote or suppress cancer by targeting a group of genes ( Fig. 1) [68]. In addition, miRNA regulates anti-tumor immunity. Some miRNAs interfere with antigen processing and presentation, upregulate human leukocyte antigen (HLA)-G expression and downregulate natural killer group 2, member D (NKG2D) ligand to form immune escape [69].

Tumor-infiltrating lymphocyte (TIL)
Previous reports shown that PD-L1 expression was significantly associated with intratumoral T cells infiltration in NSCLC [72]. The transcription factor thymocyte selection-associated high mobility group box gene (TOX) in tumor-infiltrating CD8 + T cells promotes T cell exhaustion by upregulating the expression of immune checkpoint proteins PD-1, T cell immunoglobulin and mucin-domain containing-3 (TIM-3) [73], T cell immunoglobulin and ITIM domain (TIGIT) [74], and cytotoxic T lymphocyte antigen 4 (CTLA-4), thereby attenuates the outcome of anti-PD-1 therapy (Fig. 1) [75]. Based on PD-L1/TIL status, NSCLC tumor immune microenvironments were divided into type I (PD-L1 + , TIL + ), type II (PD-L1 − , TIL − ), type III (PD-L1 + , TIL − ) and type IV (PD-L1 − , TIL + ) [76]. The difference in clinical factors related to different tumor immune microenvironment types determines the patient selection for combination immunotherapies [76]. Type I tumors benefit greatly from anti-PD-1/PD-L1 therapy. However, Type III tumors are resistant to anti-PD-1/PD-L1 monotherapy, which could be reversed by the combining adjuvant therapy to recruit T cells into tumor bed [77]. The proportion of CD8 + cells among the overall population of CD3 + TILs has a close relationship with anti-PD-1/ PD-L1 treatment outcomes. It has been shown that High CD8-to-CD3 ratio was positively correlated with diseasefree survival (DFS) and OS (DFS: HR = 0.954, 95%CI: 0.965-0.983, p = 0.002; OS: HR = 0.965, 95%CI: 0.931-1.001, p = 0.057) ( Table 1) [78]. The early proliferation of CD8 + T cells after anti-PD-1 therapy heralded a good clinical response to anti-PD-1 therapy [79]. T cell receptor (TCR) is expressed on the surface of T cells and composed of α chains and β chains, which form diversity and specificity through somatic DNA rearrangement [80]. TCR binds to MHC/antigen short peptide complex and triggers immune response (Fig. 1) [82]. Consolidation therapy with durvalumab after concurrent chemo-radiotherapy (cCRT) could significantly improve the overall survival and median progressionfree survival of patients as compared with placebo group [83]. Radiotherapy stimulated anti-tumor immunity by promoting the release of tumor neoantigens and driving the immune attack of CD8 + TILs [84]. Post-cCRT PD-L1 upregulation might be in response to radiotherapyrelated immune attack, which provided theoretical basis for the application of PD-L1 blockers following cCRT [85]. In addition, increased CD8 + TIL density after cCRT was associated with favorable survival [85].

Extracellular vesicles (EVs)
EVs are a collection of membrane-bound carriers, which carry lipids, proteins, and nucleic acids [94]. Budding inward through endosomal pathways to form exosomes and sprouting out of the plasma membrane to form microvesicles [95]. EVs bind to target cells and initiate signal transduction through receptor-ligand interactions or internalize through endocytosis [96]. EVs mediate cancer cell sensitivity to chemotherapy and radiotherapy, and are promising strategy in liquid biopsy for cancer diagnosis and predictive markers [97,98]. The exchange of EVs between immune cells affects innate immunity and adaptive immunity [99]. Local dendritic cells (DCs) secreted-EVs could induce T cell activation [95]. EVs are key components in the microenvironment that bridge the communication between tumor cells and stromal cells [100]. By extracting EVs miRNAs from advanced NSCLC patients receiving anti-PD-1/PD-L1 therapy for sequencing analysis, a remarkable difference in the concentration of specific miRNAs between responders and non-responders was found [101]. As a non-invasive liquid biopsy, early detection of tumor-derived EVs may help to predict the efficacy of anti-PD-1/PD-L1 therapy [102][103][104].

Circulating cancer-associated macrophage-like cells (CAMLs)
Tumor associated macrophage (TAM) promotes the invasion characteristics of malignant cells by secreting growth factors and cytokines such as VEGF, MMP, TNF-α [105]. TAM and circulating tumor cells (CTC) migrate to the blood circulation through lymphatic or capillary barrier, which enhance tumor invasion and distant metastasis [106]. As a diffuse TAM (Fig. 1), the isolation of CAMLs from peripheral blood of various cancer patients may be evidence of tumor metastasis and neovascularization [107]. CAMLs were quantified by the CellSieve system using multiplex immunostaining [108]. CAMLs ≥ 50 μm was defined as giant CAMLs. The size of CAMLs after completion of CRT was related to disease progression and patient's survival [109]. The presence of giant CAMLs before anti-PD-L1 maintenance therapy indicated a poor prognosis (median PFS: 8 months, HR = 2.5, 95% CI: 1.1-5.8, p = 0.025; median OS: 25 months, HR = 3.5, 95% CI: 1.3-9.6, p = 0.034) ( Table 1). The tumor-stimulating effect of CAMLs may limit the efficacy of anti-PD-L1 therapy [109].

Other peripheral blood cells
Among many indicators that reflect inflammation, the high neutrophil to lymphocyte ratio (NLR) heralded a poor prognosis in many malignant tumors [117,118]. Multiple studies found that NSCLC patients with high NLR had low response rate to immune checkpoint inhibitors (ICIs) [119,120] [119]. Similarly, another retrospective study also verified the predictive value of NLR for anti-PD-1 treatment [120]. Lactate dehydrogenase (LDH) is an indicator of cancer-related inflammation [121]. According to the values of LDH and NLR, lung cancer patients were divided into 3 groups (good, 0 factors; intermediate, 1 factor; poor, 2 factors). Compared with the good group, the intermediate group and poor group were more easily resist to anti-PD-1/PD-L1 treatment [121]. In addition, NLR and LDH might be useful indicators for predicting irAEs [122]. Neutrophils were highly correlated with myeloid phenotype, which promoted lymphocyte depletion [123]. Tumor-infiltrating CD8 + T cells to neutrophils (CD8/PMN) ratio could distinguish responders treated with anti-PD-1 therapy [123]. Combining neutrophil antagonists improved immunotherapy outcomes [123]. Besides, the amount and activity of NK cells in responders were highly elevated [124].

Gut microbiota
Gut microbiota has a symbiotic relationship with the host [125]. In addition to playing a barrier role in the gastrointestinal tract, microorganisms are related to the immune function of the plora [126]. Immune cells are activated through cross-reactivity between microbial proteins and tumor antigens [127]. DCs induce activated T cells outside the intestine, recognize tumor antigens and exert anti-tumor effect [127]. In addition, the microbial proteins translocate from the intestine to the blood circulation, trigger initial immunity in secondary lymphoid organs and induce the activation of T cells. T cells migrate to the tumor site and participate in immune surveillance ( Fig. 1) [127]. The composition of microorganisms may affect the efficacy of PD-1 inhibitors [128].
A study showed that the fecal Akkermansia muciniphila could be detected in 69% (11/16) and 58% (23/40) of patients exhibiting partial response or stable disease, whereas it was detectable in 34% (15/44) of patients who progressed or died [129]. Gut microbiota profiles of fecal specimens could be assessed by 16S ribosome RNA gene sequencing. Alipis putredinis, Prevotella copri and Bifidobacterium longum were enriched in the responders, and Ruminococcus_unclassified was enriched in nonresponders. Patients with higher microbiota diversity had significantly longer PFS (HR = 4.2, 95%CI: 1.42-12.3, p = 0.009) ( Table 1) [130]. The microbiota associated with clinical benefit varies in different studies, which implied that the difference between diet, host genetics, lifestyle factors, and human species may contribute to the diversity of gut microbiome and further affect the efficacy of ICIs [131,132]. The application of cumulative antibiotics (ATB) could reduce the diversity of gut microbiota and disrupt the microbial balance [133,134], which significantly weakened the efficacy of PD-L1 inhibitors and affected survival outcomes (median PFS: 1.9 months, HR = 1.5, 95%CI: 1.0-2.2, p = 0.03; median OS: 7.9 months, HR = 4.4, 95%CI: 2.6-7.7, p < 0.01) [135]. A study indicated that proton pump inhibitor (PPI) affected the diversity of gut microbiota through gastric acid [136]. The data of the phase II POPLAR and phase III OAK trial showed that in the population of anti-PD-L1 therapy, patients treated with ATB or PPI had shorter OS (HR = 1.20, 95%CI: 1.04-1.39) (Table 1), and the application of PPI was significantly related to shorter PFS (HR = 1.26, 95%CI: 1.10-1.44) [137]. As a promising treatment method, fecal microbiome transplantation (FMT) could improve the diversity of gut microbiota and the efficacy of immunotherapy [138,139].

Conclusion
Anti-PD-1/PD-L1 treatment is a promising treatment strategy for NSCLC. However, there are still numerous patients who are difficult to benefit from anti-PD-1/ PD-L1 treatment. Various biomarkers for predicting efficacy are being explored. In the present stage, PD-L1 expression is the most widely adopted biomarker in clinical practice. TMB, TIL and neoantigen are significantly correlated with the efficacy of anti-PD-1/PD-L1 therapy. Gut microbiota, inflammatory genes, and dysregulated miRNA play an important role in anti-tumor immune regulation. Combining of multiple biomarkers may increase the predictive robustness and guide the implementation of cancer precision medicine.