Liquid biopsy – a promising and effective method for surveying non-small cell lung cancer minimal residual diseases and anti-cancer drug response after treatment

Using peripheral blood CTC to detect MRD

Captured CTCs can be used to analyze protein expression and subcellular localization along with single-cell analysis of tumor heterogeneity [48], 49], and this is because the molecular features of CTCs are correlated with those in primary tumors [50]. For example, EGFR mutations between CTCs and primary tumors are mutable but highly consistent in patients with advanced NSCLC [51]. Additionally, a meta-analysis of 20 trials involving 1,576 patients with NSCLC showed a significant association between the presence of CTCs and reduced disease-free survival (DFS) and overall survival (OS) [52]. Therefore, monitoring CTCs as an indicator of MRD contributes to predicting recurrence and assessing the risk of metastasis.

Surgery is the primary treatment for patients with early-stage (stages I to III) NSCLC, as recommended by the National Comprehensive Cancer Network (NCCN) [53]. During the perioperative period, the CTC count fluctuates in patients with NSCLC who were surgically treated. Additionally, several studies have confirmed that a high pre-/intra-/post-operative CTC count is associated with a poorer prognosis [54], [55], [56]. Consequently, a clinically significant prognostic threshold of ≥5 CTCs per 7.5 mL of blood has been established for NSCLC prognosis using the CellSearch System [57]. Based on this, the CTCs in peripheral blood samples from 56 NSCLC patients who underwent radical surgery were assessed before and one month after the surgery. The presence of CTC at one month following surgery was found to be substantially linked with a shorter DFS (hazard ratio [HR]=5.75; 95 % CI: 1.50–21.946; p=0.010), according to multivariate analysis [58].

The detection of CTCs is a demanding technology due to the small CTC count, as few as one per 10⁶-10⁷ peripheral blood mononuclear cells, in patients with solid tumors [59], 60]. Similarly, previous studies have reported that the detectable CTC levels in the pulmonary vein blood were higher than those in the peripheral blood before surgery, and that the number of CTCs was associated with tumor size [61]. Furthermore, compared to the resected main tumor, early dispersed CTCs exhibited a larger mutational overlap when metastases occurred 10 months later [49]. These results provide novel insights into CTC detection. Consequently, further extensive prospective investigations are required to validate these findings.

In patients with NSCLC receiving chemoradiotherapy, CTCs monitoring serves as an adjunct guide for treatment. A small retrospective study showed that CTCs could be detected before the advent of radiological evidence of recurrence in patients with NSCLC (stage II–III) treated with chemoradiotherapy [62]. Another clinical trial of 101 patients with NSCLC receiving chemotherapy showed that the CTCs count decreased after administration of chemotherapy, which was associated with longer progression-free survival (PFS) (5.4 vs. 1.9 months; p<0.001) and OS (8.3 vs. 3.3 months; p<0.001) [57]. Moreover, while a decrease in CTCs after chemotherapy indicates remission, an increase indicates reactivation, which would lead to treatment adjustment [63].

When at least two CTCs aggregate by physical contact through intercellular junctions, they form clusters, also known as circulating tumor microemboli [64], 65]. CTC clusters have greater metastatic potential than individual CTCs, and may be protected from immune invasion and survive longer because of the shape of the clusters [66], 67]. Accordingly, one study recommended that the percentage of cluster CTCs was significantly correlated with the TNM stage. In addition, 9 out of 36 patients with NSCLC had post-operative recurrence, and at least one CTCs cluster was detected in six of them. Therefore, future studies should investigate the potential of using CTC clusters as biomarkers for monitoring NSCLC recurrence [68].

Using peripheral blood ctDNA to detect MRD

CtDNA is a small fragment of DNA resulting from tumor cells necrosis, apoptosis, or secretion and is released into the peripheral circulation at approximately 90–150 base pairs in length [69]. The DNA fragment can be stably detected in peripheral blood with an abundance of ≥0.02 %, including lung cancer driver genes and other class I/II gene variants. Studies have shown that ctDNA is highly consistent with tumor DNA [70], 71] and demonstrated a linear relationship between ctDNA and tumor volume [72].

Consequently, multiple retrospective analyses have confirmed that the median predictive time from post-operative dynamic monitoring of ctDNA to clinical or radiographic reports of lung cancer recurrence is 70 days to 5.2 months [43], 73], 74]. Tracking NSCLC evolution through therapy (Rx) (TRACERx) [75], by analyzing the blood of 24 patients with NSCLC after surgery, researchers accurately identified more than 90 percent of them as destined to relapse a year before clinical imaging confirmed recurrence. Additionally, >75 % of the tumors exhibited variants of PIK3CA, NF1, and genes involved in chromatin modification and DNA damage repair. Moreover, amplification of CDK4, FOXA1, and BCL11A was associated with an increased risk of death (HR=4.9; p=4.4 × 10⁻⁴). Notably, this time gap between relapse prediction and occurrence offers opportunities for clinical intervention and management.

A prospective multicenter cohort study analyzed perioperative ctDNA in 950 plasma samples from 330 patients with stage I–III NSCLC obtained at three different perioperative time points with NGS (before surgery, three days, and one month after surgery). The samples were utilized for ctDNA-based MRD analysis and demonstrated that the presence of MRD (ctDNA positivity at postoperative three days and/or one month) was a substantial predictor for disease relapse (HR=11.1; 95 % CI: 6.5–19.0; p<0.001), and that preoperative ctDNA positivity was linked with poorer recurrence-free survival (RFS; HR=4.2; 95 % CI: 2.6–6.7; p<0.001). The three tumor driver genes with the highest mutation frequencies were EGFR (64 %), TP53 (49 %), and RBM10 (17 %). Similarly, another study showed that the likelihood of recurrence in MRD-positive patients detected using these ctRNAs was 11 times higher than that in MRD-negative patients [42]. Another study analyzed cancer personalized profiling (CAPP)-seq personalized profiled ctDNA from 255 samples (patients with stage I–III lung cancer and healthy individuals) using deep sequencing. Receiver operating characteristic (ROC) analysis revealed an area under the curve of 0.97, with maximal sensitivity and specificity of 93 and 96 % respectively [73]. In another study that assessed MRD using ctDNA from 77 patients with NSCLC with a preoperative ctDNA-positive status demonstrated a significant 3.8–4.0-fold risk of recurrence and death, which was also associated with a lower RFS (HR=3.812; 95 % CI: 1.681–8.855; p=0.0005) and OS (HR=5.004; 95 % CI: 1.731–14.756; p=0.0009). The most common mutations were in TP53 (60 %), EGFR (21 %), and KEAP1 (9 %). Furthermore, the risk of recurrence, metastasis, or death in preoperative ctDNA-positive patients increased by 3.4-fold and 4.0-fold, respectively, compared with that in ctDNA-negative patients [43]. Therefore, detecting MRD using ctDNA is of great value for predicting recurrence and survival rates.

One of the biggest advantages of liquid biopsy over tissue biopsy is that it can be performed repeatedly for dynamic monitoring to further improve MRD surveillance [12]. A prospective study of surgical patients with lung cancer that investigated perioperative dynamic changes in ctDNA demonstrated that ctDNA half-life in patients with MRD was significantly longer than in those without MRD (103.2 min vs. 29.7 min, p=0.01) and determined the appropriate detection time of ctDNA-based surveillance to be the third day after R0 [76]. In another clinical study, patients with NSCLC with ctDNA detected during the post-treatment monitoring period had lower DFS than those with consistently negative ctDNA after surgery (HR=8.5; 95 % CI: 3.7–20; p<0.001). In addition, 79 % of relapsed patients tested positive for at least one ctDNA during disease surveillance. The top three ctDNA-positive findings are EGFR (55.4 %), KRAS mutation (16.1 %), and ALK fusion (8.9 %) [40]. Consequently, the longitudinal ctDNA monitoring is efficient in identifying relapsed patients. Moreover, patients with lung cancer receiving immune checkpoint inhibitors undergo dynamic monitoring of ctDNA and benefit from the detection of MRD for clinical intervention. For instance, ctDNA tracking was performed in 31 patients with NSCLC after a median of 26.7 months of programmed cell death receptor-1/the anti-programmed death 1 ligand (PD-1/PD-L1) blocking therapy. CtDNA was undetectable at the time of monitoring in 27 patients, among whom 93 % (25/27) did not progress. In addition, all four patients with detectable ctDNA experienced disease progression [77]. Moreover, it is difficult to monitor a patient’s condition after immunotherapy, since no clinical symptoms are visible. Therefore, the use of ctDNA provides an opportunity to administer therapy earlier in patients with the potential to relapse.

Using peripheral blood exosomes to detect MRD

The contents of cancer cells carried in exosomes represent critical cancer-specific biomarkers that help detect and analyze the induction of angiogenesis and the suppression of immune responses, migration, and metastasis [78]. A unique advantage of exosomes biopsy is that biological molecules are protected by a lipid bilayer membrane of exosomes that confers a high degree of stability and can be isolated and analyzed [79]. Different RNAs and proteins carried in exosomes have been demonstrated to indicate the poor prognosis of patients with NSCLC, such as miR-17-3p, miR-21, lncRNA RP5-977B1, lncRNA SNHG15, and the eukaryotic translation initiation factor 4G2 [80], [81], [82]. Thus, the detection of such biological information would help integrate MRD detection and risk of recurrence to adjust different clinical treatments. Although ctDNA and CTCs are the mainstream methods for monitoring MRD, the addition of exosomes can achieve complementarity [46]. In addition, combining exosomal RNA (exoRNA) with cell-free DNA (cfDNA) can enhance the identification of EGFR mutations in patients with NSCLC. Therefore, MRD and EGFR mutations can be detected simultaneously [47].

Liquid biopsies of non-blood liquid detect MRD in NSCLC

As some biomarkers may be present in higher amounts in non-blood liquids than in the peripheral blood, liquid biopsy of tumor biomarkers in different bodily fluids is progressively bringing optimism [83]. PE usually occurs in advanced patients with advanced NSCLC, and is currently detected using ctDNA and exosomes. Tong et al. analyzed 30 patients with NSCLC and found that, compared with the control group, the median concentration of ctDNA in the PE supernatant was 278.1 ng/mL, which was significantly greater than the median concentration in the plasma (20.4 ng/mL, p<0.001). Moreover, the detection rate of EGFR-driven mutations was the highest. Other mutations included ALK, BRAF, EGFR, ERBB2, KRAS, KRAS, NF1, PIK3CA, and RET. The results showed that monitoring MRD using ctDNA in PE has the advantage of the abundance of ctDNA [44]. Similarly, Zhong et al. found the number of CTCs detected in BALF was significantly higher than that in peripheral blood CTCs (6.76 ± 0.89 vs. 5.78 ± 0.57, p=0.016) with the same sample size, as well as other biomarkers used to diagnose recurrence, such as NSE, CEA, and CA125 (p<0.05) [41]. CTCs moving from tumors to peripheral blood undergo complex processes such as epithelial-mesenchymal transition, while in BALF, they are directly derived from the primary lesions, retaining original tumor cell characteristics and releasing metabolites. Based on these results, monitoring MRD by detecting CTCs in the BALF is a promising alternative to monitoring peripheral blood. Additionally, liquid urine biopsy is an effective tool for monitoring MRD in patients with NSCLC. One study on urinary ctDNA profiling in EGFR-positive patients with NSCLC used ddPCR sed to detect mutant EGFR in all peel samples collected at various time points. Researchers noticed a notable increase in positive EGFR mutations in urine samples after a 6-month treatment period. Ninety-one percent of individuals with higher urine DNA concentrations experience recurrence [45]. As the kidneys filter out many proteins, it is simpler to extract DNA from urine samples, which may lessen PCR amplification interference [83].

Evaluation of chemotherapy outcome by liquid biopsy

In patients with NSCLC receiving platinum-based therapy, researchers have investigated different subtypes of CTCs using multiple immunofluorescence stains. The results have demonstrated a significant relationship between an increasing ratio of stem cell-like (CD133-positive) to epithelial cells (pan-CK-positive) and mesenchymal N-cadherin-cells (2 vs. 8 months, p=0.003, HR=4.43; 5 vs. 8 months, p=0.03, HR=2.63) [84]. In addition, the expression profiles associated with CTCs provide promising results for the evaluation of platinum-based efficacy and prognosis in patients with NSCLC. For example, the enzyme excision repair cross-complementation group 1 (ERCC1) can remove cisplatin-induced DNA adducts. In a small-scale clinical trial of 17 metastatic patients with NSCLC receiving platinum-based therapy, Das et al. detected the expression of ERCC1 in CTCs and found that increased ERCC1 expression in CTCs correlated with decreased PFS (p<0.04, F-test, linear regression) [108]. Similarly, metastasis and recurrence were linked to higher expression levels of epithelial cell adhesion molecule (EpCAM) on CTCs, in 76 patients with NSCLC receiving postoperative adjuvant chemotherapy, which occurred before CT imaging and diagnosis. Particularly, a statistically significant difference (p=0.008) was observed in the EpCAM expression in CTCs one day prior to the fourth adjuvant treatment [85]. Pt-(GpG) intrastrand crosslinks are one of the major DNA adducts formed by platinum. Previous studies have shown that Pt-(GPG) adducts can be detected in peripheral blood CTCs of patients with NSCLC, and that a substantial difference exists between the effective and ineffective groups [109]. Moreover, platinum resistance mutations can be identified by assessing genetic information from CTCs. Accordingly, CTCs from nine patients with platinum-resistant advanced NSCLC were isolated and subjected to whole-exome sequencing (WES) at the single-cell level. Functional annotation demonstrated that cancer-driver mutations, including EGFR and TP53, and cell cycle-regulated or stem cell-related gene mutations, including SHKBP1, NUMA1, ZNF143, MUC16, ORC1, PON1, and PELP1, were found in CTCs and were to chemotherapy resistance and metastasis [86].

CfDNA has been used to detect mutation sites because it can be conveniently extracted dynamically and noninvasively. For instance, in a study involving 57 patients with NSCLC receiving platinum-based chemotherapy and 13 patients receiving neoadjuvant chemotherapy, WES was used to identify somatic mutations in 225 genes. The platinum response was found to be associated with copy number variations in chromosomes 8q24.3 and 22q11.21, as well as non-synonymous variants in EGFR, TTN, TP53, and KRAS. Additionally, the failure of DNA double-strand breaks and calcium signaling pathways is correlated with the mutational signatures of these variants [87]. In a similar study of cfDNA mutations detected by WES, researchers observed 1,559 point mutations across 98 plasma samples and found that the whole mutation burden decreased after platinum-based chemotherapy (p=0.034). Moreover, fewer non-silent mutations were detected after chemotherapy in patients with a better platinum therapeutic response than in insensitive patients, which reflected the sensitivity of platinum chemotherapy. In addition, some driver mutations, such as TP53 (p.K159X) and EGFR (p.E709 K and p.G719A), were undetectable, while some hotspot mutations, such as EGFR (L858R), KRAS (p.G12C), persisted after platinum chemotherapy [88]. With an in-depth exploration of the longitudinal monitoring of ctDNA, researchers found that ctDNA levels peaked at a median of 7 h after the beginning of chemotherapy (Interquartile Range [IQR]: 2–26 h). CtDNA was detected at baseline in 75 % of patients receiving systemic chemotherapy; CTCs were detected at baseline and had post-treatment genomic alterations in 28 % of patients. The abundance of ctDNA in the plasma increased due to tumor lysis when tested immediately after the initiation of treatment, allowing the detection of genetic changes hidden in baseline testing. Furthermore, experimental data have demonstrated that cfDNA can help predict the prognosis of patients receiving platinum-based chemotherapy. A previous multivariate analysis reported that a higher baseline concentration of cfDNA was an independent prognostic factor, with patients with high cfDNA concentrations at baseline showing poorer DFS and OS than those with lower concentrations (p=0.001) [89].

Exosomes have been confirmed to confer chemoresistance to cancers through the transfer of miRNAs. Exosomal miRNAs are easy to extract and measure from peripheral blood, making them potential biomarkers for predicting chemotherapy response and resistance. Ma et al. observed that individuals tolerant to chemotherapy drugs had higher baseline levels of miR-425-3p in their exosomes. Exosomal miR-425-3p targets AKT1 to promote autophagic activation of recipient cells, ultimately leading to chemoresistance [90]. Additionally, miR-1273a was the most significantly downregulated miRNA in the cisplatin-treated exosomes by microarray analysis. Downregulation of exosomal miR-1273a upregulates the expression of syndecan-binding proteins, eventually increasing cisplatin resistance in NSCLC [91].

Furthermore, an increasing number of studies have suggested that circulating miRNAs are correlated with the prognosis of patients with NSCLC receiving platinum-based chemotherapy. This suggests that they are potential biomarkers for predicting drug response and survival. Accordingly, in an investigation of the expression and clinical significance of different plasma miRNAs in patients who received platinum-based therapy, researchers found that high expression of miR-128 and miR-155 resulted in shorter OS than those with low expression (p<0.05). MiR-128 promotes epithelial-mesenchymal transition and cell migration targeted by Drosha ribonuclease III and Dicer ribonuclease III in NSCLC [110]. The carcinogenic effect of miR-155 is correlated with the inhibition of PTEN and cytokine signaling 1 and 6 [111]. Circulating miRNAs involved in macrophage polarization function were predictive variables in 125 patients with NSCLC treated with first-line platinum-based therapy. Particularly, high miR-202 expression was an independent prognostic factor for shorter PFS (p=0.021) and OS (p=0.024) [92].

Pemetrexed inhibits thymidylate synthase and folate-dependent enzymes, interfering with the biosynthesis of thymidine and purine nucleosides, eventually blocking tumor growth [112]. However, no reliable biomarker is currently available for determining the effectiveness of pemetrexed-based chemotherapy. Therefore, Franchina et al. determined the expression levels of several circulating miRNAs that are possibly involved in the folate pathway in 22 patients with NSCLC treated with pemetrexed-based chemotherapy and 27 healthy controls. Patients with progressing diseases had considerably higher miR-22 expression than those without disease progression (p=0.03). Thus, the correlation between high expression and a lack of response in patients indicates that miR-22 is a predictive biomarker for pemetrexed-based chemotherapy [113]. Accordingly, researchers investigated the relationship between circulating endothelial cells (CECs) counts and the efficacy of platinum plus pemetrexed first-line chemotherapy in 69 patients with NSCLC. Peripheral blood CECs were detected at baseline, and after the second and third cycles. The results showed that the CEC count increased by >50 % between the first and second cycles and was significantly associated with disease progression (p=0.008). Additionally, patients with baseline CEC counts above average (>153 cells/4 mL) had longer PFS and OS (with no statistically significant difference) [93].

The role of liquid biopsy in tyrosine kinase inhibitor treatment

Tyrosine kinase inhibitors (TKIs) are generally prescribed for patients with metastatic NSCLC (stage III or IV) and certain genetic alterations. EGFR mutations are the most common genetic alterations in patients with NSCLC. Liquid biopsy is used in TKI therapy to select the targeted therapy according to the type of molecular mutation, monitor efficacy, and detect drug resistance mechanisms [114].

Deep sequencing of plasma DNA has been used to detect EGFR mutations in ctDNA. When tested in 288 patients with NSCLC, the diagnostic specificity of the exon 19 deletion was 98.0 % and the specificity of L858R was 94.1 %. Thus, owing to their high specificity, EGFR-TKIs can be directly recommended based on positive plasma DNA results [115]. Conversely, tissue biopsies and ctDNA detection yielded false-negative results for EGFR mutations, which may be partly due to the release, distribution, and clearance of ctDNA. For instance, 28/1,017 (2.75 %) patients with NSCLC with EGFR mutation were not detected via tissue biopsy but were detected using ctDNA, which may be correlated with the spatial heterogeneity of the tumor. The PFS of patients who received EGFR TKIs was higher in that patient group than that of those who received chemotherapy, thus identifying that this group had special therapeutic significance. Currently, if ctDNA fails to detect EGFR mutations, tissue biopsy is used as a supplementary test, which may save approximately 30 % of tissue samples [116].

Furthermore, resistance to EGFR-TKIs, driven by various genes and different genomic alterations, has been confirmed using NGS of cfDNA [94]. In addition to EGFR T790 and C797S mutations, rare EGFR mutations or other resistance mechanisms, such as G724S, and L858R/A859S/Y891D triple mutations, FGFR3-TACC3 fusions, and NTRK1 fusion also exist [95], 117], 118]. EGFR mutations coexisting with EML4-ALK gene translocations in ctDNA were first reported in a patient with NSCLC with multiple metastases [119]. Therefore, dynamic monitoring of EGFR-TKI resistance can identify drug resistance sites and predict treatment failure in a timely manner, thus providing a window of opportunity for intervention.

A previous study showed that a decline in ctDNA concentration 48 h after the initiation of EGFR targeted therapy identified good responders to TKI and had longer PFS (14.7 months vs. 8.5 months, p=0.013) than patients with stable or increased ctDNA levels [96]. Thus, the detection of EGFR mutations by ctDNA can predict patient outcomes and assist in the timely adjustment of the appropriate treatment regimen for patients receiving EGFR-TKIs. Moreover, persistent EGFR mutations in ctDNA at weeks 3 and 6 after osimertinib administration predicted shorter PFS in patients who were considered to benefit from additional chemotherapy afterwards [91]. Therefore, it is possible to select appropriate EGFR-TKI therapeutic regimens for patients with NSCLC based on ctDNA levels and identify patients with significant improvements in survival during treatment. Accordingly, in an analysis of 830 plasma samples from 228 patients with stage IV EGFR-positive NSCLC who received first-line osimertinib (3rd generation TKI) or sequential TKI therapy, the OS of the two treatment plans barely differed. Furthermore, the low-risk patients identified receiving sequential TKI therapy at a mutant allele frequency (MAF) <7 % at diagnosis and the high responders who were ctDNA negative after 3 or 6 months of treatment and with an MAF <7 % before treatment had two-thirds lower risk of death than in the opposite case. Consequently, sequential treatment may be advantageous for strong responders and low-risk patients [120].

The First-Line AUtopia of osimertinib in EGFR-mutated NSCLC (FLAURA) study confirmed that EGFR-mutated NSCLC patients treated with osimertinib as first-line treatment had 6.8 months longer OS than patients treated with gefitinib or erlotinib. Unfortunately, resistance is inevitable, and the spectrum of its mechanisms is broadly heterogeneous [121], 122]. Currently, elevated levels of exosome-derived miRNAs, such as miR-210-3p, miR-184, miR-3913-5p, and miR-494-3p, which are involved in epithelial-mesenchymal transition, activation of bypass pathways (RAS-MAPK path anomaly and PI3K pathway activation), and other potentially unknown mechanisms, are capable of participating in osimertinib resistance since their elevated plasma levels corresponded to poor patient outcomes [97], [98], [99].

Liquid biopsy and antiangiogenic drugs

Given the importance of angiogenesis in tumors, CECs and circulating endothelial progenitor cells (CEPCs) have been used to evaluate tumor angiogenic activity and the efficiency of antiangiogenic drugs.

Different lung adenocarcinoma histological subtypes show differences in angiogenic status. The CEPC count and the level of vascular endothelial growth factor (VEGF) were higher in solid adenocarcinomas than in non-solid adenocarcinomas (p<0.001, p=0.005). Therefore, identifying patients with solid adenocarcinoma is crucial because they may benefit from anti-angiogenic therapy [100].

CEC count was confirmed to be correlated with PFS in antiangiogenic therapy. For instance, docetaxel and bevacizumab have been demonstrated to synergistically reduce endothelial cell proliferation and inhibit CECs mobilization. Researchers estimated the level of CECs on days 1 and 8 and found that patients with ≥10 count increase in CECs had longer PFS than patients with <10 count increase in CECs (median PFS of 11.0 vs. 6.90 months, respectively) [101]. Furthermore, 49 patients with NSCLC with anlotinib were stratified according to the ratio of minimal CECs counts to baseline (CECs min/baseline) as <1 or ≥1. Subsequently, CECs were identified using flow cytometry as CD31+ cells and CD105+ cells. The results revealed that patients with CECs (CD31+) min/baseline <1 had longer PFS (p<0.05) than those with CECs (CD31+) min/baseline ≥1. In addition, CD31 showed more responsive alterations than CD105 through the PI3K-AKT pathway, supporting CD31+ CECs as a more sensitive marker for predicting the efficacy of anlotinib treatment [102]. Another study showed that pre-therapeutic patients with vimentin (Vim) + CTECs at baseline showed significantly shorter median progression-free survival (mPFS) than those with Vim-CTECs. Post-therapy patients with epithelial cell adhesion molecule (EpCAM) + CTCs and CTECs, regardless of Vim expression, showed significantly reduced mPFS [103]. Thus, comprehensive co-detection of various subtypes and karyotypic molecular characterization of CTCs and circulating tumor endothelial cells (CTECs) may be useful tools for prognosticating and tracking the effectiveness of treatment in patients with NSCLC receiving bevacizumab with chemotherapy.

Evaluation of immunotherapy by liquid biopsy

Immune checkpoint inhibitors (ICIs), including antibodies targeting PD-1, PD-L1, and cytotoxic T-lymphocyte antigen 4 (anti-CTLA-4), have been approved as first-line treatments for NSCLC or after progression to chemotherapy [123], 124]. Liquid biopsy may help with screening patients who can benefit from ICIs or who are prone to immune-related adverse events (irAEs), in addition to dynamically monitoring the efficacy of ICIs.

Non-synonymous mutations in tumor cells lead to neoantigen production, which is associated with increased tumor immunogenicity [125]. Tumor mutational burden (TMB) is defined as the total number of somatic non-synonymous mutations per megabase in the coding region of a gene, including small insert-deletion variants [126]. Studies have indicated that tumors with a higher TMB and more neoantigens may respond better to immunotherapy [127], 128]. Additionally, clinical trials have suggested that PD-L1 expression is positively correlated with TMB and response to immunotherapy [129]. Moreover, the TMB detection can overcome the dynamic acquisition difficulties and spatial heterogeneity of PD-L1. Accordingly, in June 2020, the U.S. Food and Drug Administration (FDA) recognized the elevated TMB as a requirement for pembrolizumab therapy for several tumor types [130].

Liquid biopsy has been used to evaluate the prognosis of patients with NSCLC receiving immunotherapy. Moreover, TMB, as measured by ctDNA in the blood (bTMB), may be utilized as a biomarker for patients on anti-PD-1/PD-L1 medication. Accordingly, a previous study found that bTMB ≥6 was associated with favorable progression-free survival [104]. Menno et al. detected CTCs in patients with advanced NSCLC with a worse response rate to checkpoint inhibitors and high tumor-derived EVs. CTCs counted ≥18/7.5 mL in EVs were found associated with shorter survival [131]. Similarly, PD-L1 expression was measured in 96 patients with NSCLC treated with nivolumab and was confirmed to be higher in CTCs than in tissues (83 % vs. 41 %), indicating that dynamic monitoring of PD-L1 in CTCs is more convenient and sensitive [107]. Regulatory T (Treg) cells are a subset of T cells that express FoxP3 [132]. Research indicates that FoxP3 may be a favorable prognostic indicator for patients receiving anti-PD-1 treatment. One week after anti-PD-1 therapy, patients with NSCLC had higher response rates, longer progression-free survival, and improved OS in the presence of a greater number of circulating FoxP3+ Tregs (p<0.05). Similarly, low human polymorphonuclear myeloid‐derived suppressor cells (PMN-MDSCs), monocytic myeloid‐derived suppressor cells, and CD39+ CD8+ T cells have the potential to serve as favorable predictors for patients with advanced NSCLC receiving anti-PD-1 therapy [133].

However, most patients with NSCLC treated with ICIs develop early disease progression, making it impossible to identify patients who are likely to have durable clinical benefits (DCB) by imaging examination. In this regard, pre-treatment ctDNA and peripheral CD8+ T-cell levels have recently been demonstrated as independent predictors of DCB. Investigators found that patients with NSCLC with lower baseline circulating CD8+ T cell levels can achieve DCB from ICIs. Additionally, they developed and validated a non-invasive response classifier incorporating pre-treatment ctDNA and immune profiling, which could accurately identify patients with NSCLC with DCB [105]. Similarly, Duchemann et al. calculated the ratio of CD8+ PD-1+ to CD4+ PD-1+ (PD-1-Expressing Ratio on Lymphocytes in a Systemic blood sample, or ‘PERLS’) using cytometry in patients with NSCLC before receiving PD-1/PD-L1 blockers. PERLS+ was suggested to be significantly associated with PFS (9.63 months vs. 2.20 months; p=0.004) and OS (not reached vs. 7.98 months; p=0.02). The PERLS scores of patients with DCB were significantly higher than those of patients without DCB (mean 1.43 vs. 1.06; p=0.002) [106]. The neutrophil-to-lymphocyte ratio represents a proinflammatory diathesis that indicates poor prognosis in patients with NSCLC. Accordingly, in a phase 3 OAK trial (NCT02008227), pre-treatment NLR≥4 was strongly correlated to mortality after atezolizumab (HR=1.64; 95 % CI: 1.35–2.01). Consequently, a low baseline NLR could recognize patients with NSCLC who would benefit more from atezolizumab [134].