Predictive and Prognostic Biomarkers for Lung Cancer Bone Metastasis and Their Therapeutic Value

Lung cancer is the leading cause of cancer-related death worldwide. Bone metastasis, which usually accompanies severe skeletal-related events, is the most common site for tumor distant dissemination and detected in more than one-third of patients with advanced lung cancer. Biopsy and imaging play critical roles in the diagnosis of bone metastasis; however, these approaches are characterized by evident limitations. Recently, studies regarding potential biomarkers in the serum, urine, and tumor tissue, were performed to predict the bone metastases and prognosis in patients with lung cancer. In this review, we summarize the findings of recent clinical research studies on biomarkers detected in samples obtained from patients with lung cancer bone metastasis. These markers include the following: (1) bone resorption-associated markers, such as N-terminal telopeptide (NTx)/C-terminal telopeptide (CTx), C-terminal telopeptide of type I collagen (CTx-I), tartrate-resistant acid phosphatase isoform 5b (TRACP-5b), pyridinoline (PYD), and parathyroid hormone related peptide (PTHrP); (2) bone formation-associated markers, including total serum alkaline phosphatase (ALP)/bone specific alkaline phosphatase(BAP), osteopontin (OP), osteocalcin (OS), amino-terminal extension propeptide of type I procollagen/carboxy-terminal extension propeptide of type I procollagen (PICP/PINP); (3) signaling markers, including epidermal growth factor receptor/Kirsten rat sarcoma/anaplastic lymphoma kinase (EGFR/KRAS/ALK), receptor activator of nuclear factor κB ligand/receptor activator of nuclear factor κB/osteoprotegerin (RANKL/RANK/OPG), C-X-C motif chemokine ligand 12/C-X-C motif chemokine receptor 4 (CXCL12/CXCR4), complement component 5a receptor (C5AR); and (4) other potential markers, such as calcium sensing receptor (CASR), bone sialoprotein (BSP), bone morphogenetic protein 2 (BMP2), cytokeratin 19 fragment/carcinoembryonic antigen (CYFRA/CEA), tissue factor, cell-free DNA, long non-coding RNA, and microRNA. The prognostic value of these markers is also investigated. Furthermore, we listed some clinical trials targeting hotspot biomarkers in advanced lung cancer referring for their therapeutic effects.


INTRODUCTION
Lung cancer, comprising non-small-cell lung cancer (NSCLC) and small-cell lung cancer (SCLC), is the leading cause of cancerrelated death worldwide, with approximately 1.6 million deaths reported annually (1). Approximately 36% of all lung cancers and 54.5% of the stage II-IIIA NSCLC cases showed postoperative recurrence or metastasis. Bone metastasis, which is the most common site for tumor distant dissemination, is detected in more than one-third of patients with advanced lung cancer (2,3). Patients with bone metastases experience complications, including severe bone pain, pathological fractures, spinal cord and nerve compression, and other skeletal-related events, which significantly reduce patient quality of life (2,4). According to the guidelines of the European Society for Medical Oncology and several clinical studies, use of bisphosphonates (e.g., zoledronic acid) or antireceptor activator of nuclear factor kB (NF-kB) ligand (anti-RANKL) monoclonal antibody (e.g., denosumab) is recommended for the treatment of lung cancer patients with bone metastasis (5)(6)(7)(8)(9).
The detection of lesions in patients at early disease stages before the occurrence of distinct clinical symptoms is challenging. This delayed diagnosis shortens the optimal treatment period. Therefore, early diagnosis of bone metastasis at the initial staging work-up could lead to markedly improved clinical outcomes. Thus far, the most accurate method for the detection of bone metastasis is bone marrow biopsy. Moreover, imaging diagnostic methods, including X-ray, computed tomography, magnetic resonance imaging, technetium 99mmethyl diphosphonate ( 99m Tc-MDP) bone scan, and 18 Ffluorodeoxyglucose positron emission tomography/computed tomography play critical roles (10,11). However, the clinical application of these procedures is limited due to their low specificity, invasive operation, complex process, and high costs. Thus, other indirect markers are unavoidably required for predicting lung cancer bone metastasis in addition to its prognosis.
It is widely acknowledged that adenocarcinoma, pathological stage III disease, and advanced age are significantly related to a high risk of bone metastasis (12)(13)(14). Moreover, lymph node metastasis has been associated with higher odds of multiorgan metastases and a worse overall survival (OS) (15). Programmed cell death 1 ligand 1 (PD-L1) immunochemistry is commonly used as a prognostic biomarker of immune checkpoint inhibitor for clinical decision making. Nevertheless, some researchers suggest that the usefulness of PD-L1 expression is controversial (16)(17)(18). Moreover, the tumor mutational burden is measured by whole exome and next-generation sequencing to predict the clinical outcome after treatment with immune checkpoint inhibitor, as well as bone metastasis (19). Nevertheless, these factors provide limited information. Thus, it is important to investigate novel methodologies and therapeutics that would increase the proportion of patients responding to such treatments, while simultaneously improving their side effect profile.
Studies on potential markers in the serum, urine, and tumor tissue were undertaken to predict the bone metastases in patients with NSCLC. Some of these markers are stable and easy to detect, and potentially helpful in disease classification, diagnosis, and treatment.
In this review, we summarize the findings of recent clinical research studies on markers detected in samples obtained from patients with lung cancer, and analyze their capacity for the prediction of treatment response and prognosis of bone metastasis. For a logical classification, these markers are divided into bone resorption-associated markers, bone formation-associated markers, bone metastasis-associated signaling markers, and others (20). The overall concept pictureof lung cancer bone metastasis is showed in Figure 1.

OSTEOCALST-AND BONE RESORPTION-ASSOCIATED BIOMARKERS
The components of the bone microenvironment, including osteocytes, immune cells, and bone matrix, play critical roles in bone metabolism and tumor cell seeding, dormancy, and growth (21). The interaction between osteoclasts and osteoblasts promotes bone degradation through the release of several factors. Disseminated lung cancer cells activate osteoclasts around metastatic foci, which facilitate expansion in the mineralized matrix (22). During these processes, the rate of resorption can also be quantified through several bone resorption markers, which are described below.

N-terminal Telopeptide (NTx) and C-Terminal Telopeptide (CTx)
Type I collagen is the principal collagen in the skeletal system, accounting for approximately 90% of the organic chemical constituents of bone. NTx and CTx are the degradation products of collagen which are released into blood when bone is absorbed, and subsequently collected in urine after kidney drainage (23,24). Studies performed by Izumi et al. and Chung et al. indicated that urinary NTx plays a significant role in the diagnosis of patients with bone-metastatic lung cancer (25,26). In addition, the levels of serum NTx were increased in patients with NSCLC as well as other solid tumors with bone metastasis (23,27). Bayrak et al. (28) demonstrated that, at the threshold value of 25.69 nmol bone collagen equivalents (BCE), the sensitivity of serum NTx was 90.42% and the specificity was 43.4%. Kaira et al. (29) revealed that normalization of elevated urinary NTx at 1 month after chemotherapy was correlated with the therapeutic response of patients with bone-metastatic lung cancer. Moreover, Zhang et al. (30) analyzed 1,279 patients and revealed that serum concentration of NTx was significantly high in the Chinese population with bone metastasis from solid tumors including lung cancer. Similarly, the measurement of CTx is considered very helpful for the complementary diagnosis of bone metastasis from lung tumor (31)(32)(33). However, some researches indicated that the levels of CTx were not significantly different between patients and healthy individuals (28). These inconsistent results may be attributed to the effects of diets or injuries on the levels of CTx. Therefore, both NTx and CTx are beneficial in detecting bone metastasis in lung cancer and worth further exploring.

C-Terminal Telopeptide of Type I Collagen (I-CTP)
Pyridinoline (PYD) cross-linked I-CTP is a metabolite released during the degradation of type I collagen. It is recognized as one of the most accurate markers for bone metastasis. In 2005, Yokoyama et al. (34) conducted a clinical research study of 87 patients with primary lung cancer, showing that serum I-CTP was significantly higher in patients with bone metastasis than those without. The most efficient cut-off value for I-CTP computed in this study was 6.4 ng/ml. A study performed by Tanaka et al. (35) in 2013, which included a larger number of patients, yielded similar results. Charpko et al. (36) utilized the 99m Tc-MDP scan technique to directly demonstrate the correlation between serum I-CTP and the extent of metastatic disease in bone. Kong et al. (37) compared the clinical effect of serum b-type collagen carboxy-terminal peptide and I-CTP in 126 patients, demonstrating that the latter has higher sensitivity and accuracy with a similiar receiver operating characteristic curve [area under the curve (38)] value. Tang et al. (33) combined tartrate-resistant acid phosphatase isoform-5b (TRACP-5b) and I-CTP as prognostic factors, revealing an elevated AUC of 0.895 compared with that of each factor alone (p<0.0001); the cut-off values were 7.6 U/l and 8.4 U/l, respectively. These studies showed that I-CTP may be a reliable marker and assessment using several markers can be more beneficial in the prediction of bone metastasis.

TRACP-5b
TRACP-5b is an enzyme secreted mainly by activated osteoclasts, thus reflecting the enrichment and vitality of osteoclasts. Its serum levels reflect the degree of osteolytic bone metastasis and the tumor burden within the bone milieu (39). Moreover, this enzyme responds rapidly and significantly to anti-resorptive treatment and is minimally affected by diet and renal or hepatic disease compared with NTx and CTx. These features render TRACP-5b to be a bone-specific assessment factor. Studies have demonstrated that the levels of TRACP-5b are significantly higher in breast cancer and prostate cancer with bone metastasis (40,41). Notably, in patients with lung cancer bone metastasis, some clinical research has produced controversial results regarding its accuracy (32,33,(42)(43)(44). In clinical cases, the elevation of TRACP-5b can be caused by physiological processes, such as osteoporosis and hyperparathyroidism. Thus, other markers should be considered in combination with TRACP-5b for prediction. sign of metastasis in patients without bone abnormality assessed by imaging techniques. This is because D-PYD is not internally metabolized, but excreted in the urine. Hence, the measurement of D-PYD in urine is of great importance in monitoring bone resorption.
Parathyroid Hormone-Related Peptide (PTHrP) PTHrP, secreted by malignant tumor cells, acts as a potent stimulator of osteoclastic bone resorption and may induce hypercalcemia (48). Miki et al. (49) designed a series of studies with anti-PTHrP neutralizing antibody. They found that PTHrP played an essential role in bone metastasis, but not visceral metastases, in SCLC, indicating to some extent the specificity of PTHrP. Katseli et al. (50) collected the circulating tumor cells in peripheral blood and concluded that the mRNA levels of PTHrP correlated positively with lung cancer stage, presence of bone metastasis, and squamous cell carcinoma. Deng et al. (51) detected the upregulated expression of PTHrP in bone metastasis sites of NSCLC by reverse transcription-polymerase chain reaction(RT-PCR) and immunohistochemistry. Teng et al. (52) constructed a novel model, including PTHrP, for the early identification of lung cancer bone metastasis. This model included 205 patients, and the average predictive time for the occurrence of bone metastasis was 9.46 months (shorter than that of bone scan diagnosis). Moreover, the response to some osteoclast-targeted therapies, including erlotinib and reveromycin A, could also be estimated by the serum levels of PTHrP (53,54).

RANKL/RANK/Osteoprotegerin (RANKL/ RANK/OPG) System
The RANKL/RANK/OPG pathway is a vital regulator of bone metabolism in normal and pathological conditions, such as the "Vicious Cycle" (55). RANK is a surface receptor specifically expressed on osteoclasts and their progenitors. It is activated by binding to RANKL expressed on the surface of osteoblasts and stromal cells, with the assistance of macrophage-colony stimulating factor (M-CSF) (56). The RANK/RANKL interaction subsequently activates transcription factors, such as NF-kB, and eventually increases the number of activated osteoclasts (57). OPG, a decoy receptor secreted by osteoblasts and stromal cells, inhibits the positive effects of RANKL on osteoclasts by preventing the RANKL/RANK interaction. This leads to hindrance of osteoclastogenesis and bone resorption (58).
The RANKL/RANK/OPG axis has been used in the prediction of bone metastasis. Karapanagiotou et al. (43) demonstrated that increased serum levels of RANKL were detected in NSCLC patients with bone metastases, while increased levels of OPG were observed in all patients with lung cancer. Terpos et al. (44) revealed an unusually low serum TRACP-5b/RANKL ratio in patients with existing or subsequent metastasis. Santini et al. (59) performed an immunohistochemical analysis of tumor tissues collected from 74 patients. The results displayed that 89% of the patients with bone metastasis were RANK-positive, and 59.5% of patients showed >50% positive tumor cells. Moreover, there was no significant difference noted in the expression of RANK between primary tumors and metastases (59). According to these research studies, the RANK/RANKL pathway shows great stability and reliability in the diagnosis of bone metastases of lung cancer, and RANKL-targeting agents could serve as a rational approach for the prevention of bone metastases (60,61).
The therapeutic effect of RANK-targeting denosumab in the treatment of advanced breast cancer, prostate cancer, and multiple myeloma was demonstrated in a series of phase III studies (62)(63)(64)(65). Recently, a randomized, open-label, phase III SPLENDOUR trial was designed to evaluate the treatment of denosumab in patients with advanced lung cancer. The addition of denosumab to chemotherapy in the intent-to-treat population and subgroups with or without bone metastases did not lead to improvement in OS (66). Of note, denosumab was welltolerated, without unexpected safety concerns. Clinical trials of denosumab that are currently recruiting patients are listed in Table 5. Moreover, some studies revealed the effect of serum RNA levels of RANK, RANKL, and OPG in the evaluation of zoledronate acid therapy in patients (61). Additionally, the use of thrombospontin was considered for the prevention of osteoporosis and bone metastases (67).

OSTEOBLAST-AND BONE FORMATION-ASSOCIATED BIOMARKERS
Lung cancer cells implanted in bone loci also affect the activity of osteoblasts in the microenvironment. Osteoblasts are activated by a large amount of growth factors and mediated by certain osteoclast-secreted proteins, which could to some extent predict the skeletal invasion of cancer cells (22).
Total Serum Alkaline Phosphatase/Bone Specific Alkaline Phosphatase (TAP/BAP) TAP is widely distributed in the body with a series of isoenzymes. TAP lacks diagnostic specificity because of its sensitivity to several non-neoplastic diseases, such as liver cancer or cholecystitis. 68] Thus, few research studies focus on TAP. Instead, BAP is thought to be a reliable marker reflecting osteoblast activity and mineralization as one of the isoenzymes of TAP. Aruga et al. (68) and Ebert et al. (42) proved that the concentrations of TAP and BAP were significantly higher in patients with bone metastasis than those without (p<0.01). Lumachi et al. (69) proposed that there is no correlation between age and BAP (R=−0.10, p=0.61) among patients with bone metastases. Bayrak et al. (28) obtained a BAP threshold value of 22.38 mg/l with sensitivity of 60.87% and specificity of 69.05%. Similarly, Tang et al. (33) obtained cut-off values of 21.8 mg/l with an AUC of 0.760. Notably, there are two age-dependent physiological peaks of serum BAP at infancy and puberty. These peaks in levels are associated with altered longitudinal growth and may affect the reliability of results (70). These studies revealed that BAP plays an important role in predicting bone metastasis of lung cancer, while its clinical use require more limitation, such as age.

Osteocalcin (OC)
OC is the most abundant non-collagenic protein in the bone extracellular matrix. It is only produced by osteoblasts and odontoblasts under the control of vitamin D3 (71,72). OC and fragments of the peptide are released during bone resorption, indicating its potential role as a marker of bone metastasis. Karapanagiotou et al. (43) and Terpos et al. (44) demonstrated that decreased OC serum levels were detected in NSCLC patients with bone metastases. In contrast, Bayrak et al. (28) argued that the levels of OC were not significantly different between the bone metastasis and no metastasis groups. Therefore, considering its limited accuracy and stability, OC may not be an ideal choice for predicting bone metastasis.

Osteopontin (OPN)
OPN is a sibling glycoprotein first identified in osteoblasts. It is defined as an extracellular matrix protein, which plays a vital role in the mineralization and absorption of bone matrix (72). In the other side, OPN regulates the migration and adhesion of tumor cells by binding to its receptor CD44. A study confirmed that OPN is overexpressed in NSCLC tumor tissues compared with their adjacent normal counterparts, and significantly correlated with TNM stages and lymph metastasis (73). CD44 is a cell-surface receptor for OPN and mediates epithelial-mesenchymal transition; hence, it is related to tumor migration (76). Liu et al. (77) genotyped the CD44 gene in DNA extracted from blood lymphocytes of patients with NSCLC and found that the rs187115 (AG+GG) genotype was a significant predictor of bone metastasis and poor survival. It is widely accepted that the detection of single nucleotide polymorphisms in CD44 and other genes would be helpful in assessing the invasion ability of tumors (78).
Amino-Terminal Extension Propeptide of Type I Procollagen/Carboxy-Terminal Extension Propeptide of Type I Procollagen (PINP/PICP) PICP and PINP are produced by the extracellular cleavage of a molecule of type I procollagen at the moment of its incorporation into the bone matrix (71). The concentrations of PICP and PINP, reflecting the rate of synthesis of type I collagen, are altered by the liver where they are degraded. Kobayashi et al. (79) demonstrated that the serum levels of PINP were significantly correlated with the clinical stage, extent of bone metastasis, survival time, levels of D-dimer, and tumor size in lung cancer. In 2004, Elbert et al. (42) demonstrated that the concentration of PINP was significantly higher in patients with bone metastasis than those without (p<0.01), with sensitivity and specificity of 18.4% and 97.5%, respectively. Other researchers, including Lüftner et al. (80), assessed the predictive effect of the serum levels of PINP on the osseous spread of breast and prostate cancer, while studies on lung cancer are limited and require more attention.

Bone Sialoprotein (BSP)
BSP, a sulfated phosphoprotein enriched in bone and other mineralized tissue, is critical in elucidating the formation of bone metastases (81). In 1997, Bellahcene et al. (82) found that BSP was not detected in normal lung tissue, except the cartilage in bronchi. On the contrary, most of the adenocarcinomas and all squamous carcinomas among the 86 tested lung tissues had detectable levels of BSP. Moreover, Zhang et al. (83) further declaimed that BSP expression in patients with tumor-resected NSCLC is significantly correlated with bone metastasis and can be utilized to identify high-risk patients after primary tumor resection. Similarly, He et al. (84) revealed that the serum levels of BSP could be used to evaluate bone metastasis, with a cut-off value of 33.56 ng/ml, sensitivity of 77.8, and specificity of 81.1%. Collectively, these findings indicated that BSP is a stable component of a bone metastasis predictive model, like in the study conducted by Zhou et al. (74).

BONE METASTASIS-ASSOCIATED SIGNALING MARKERS
Osteoclast and osteoblast functions are thought to be essential in bone metastasis for different types of lung cancer, and varying degrees of osteolytic and osteoblastic activities have been observed. As expected, more potential markers are involved in the progression of osteolytic and osteoblastic metastasis, by mediating downstream proteins or the signaling axis. These markers should be taken into consideration for the early diagnosis of lung cancer bone metastasis, although the mechanisms involved in this process remain unknown (81).

Epidermal Growth Factor Receptor/ Kirsten Rat Sarcoma/Anaplastic Lymphoma Kinase (EGFR/KRAS/ALK) Axis
Lung cancer is often driven by molecular alterations, such as EGFR and KRAS mutations, and ALK rearrangements expressed in tumor tissues of patients with NSCLC. Fujimoto et al. (85) demonstrated that patients with EGFR mutations (N=98) had significantly more metastatic lesions in the brain and bone than those in the wild-type group. In addition, EGFR mutations were significantly more frequent in patients with multiple lung metastases than a single lung metastasis (24/40 vs. 12/42, respectively; p=0.004). Kuijper et al. (86) found that EGFR+ tumors were linked to more frequent metastases to the bone (31.5% vs. 53.8%, respectively) and pleura (24.1% vs. 37.5%, respectively) compared with triple-negative tumors. Of note, Hendriks et al. (87) and Dormieux et al. (88) showed that the incidences of brain and bone metastases were not different between EGFR+, KRAS+, and wild-type patients. Nevertheless, Zang et al. (13) and Amelot et al. (89) revealed that OS was associated with the EGFR status in patients with NSCLC spinal metastasis.
EGFR mutations may also be a useful prognostic factor for the response to EGFR tyrosine kinase inhibitors (TKIs) (90). The majority of these mutations are located in the catalytic kinase domain of EGFR. Typical mutations include a deletion within exon 19 (E19del) or a leucine-to-arginine point mutation at codon 858 (L858R) within exon 21 (91). The detection of E19del and L858R by immunohistochemistry predicts the effectiveness of first-generation EGFR TKIs (e.g., gefitinib and erlotinib) and second-generation TKIs (e.g., afatinib) (92,93). Other studies identified T790M mutations in EGFR exon 20 as predictors for the development of resistance to first-generation EGFR TKIs (94)(95)(96). According to the phase III AURA-3 trial, treatment of patients with T790M mutations using osimertinib resulted in a better clinical outcome compared with chemotherapy, with an objective response rate of 71% versus 31%, respectively, and median progression-free survival (PFS) of 10.1 versus 4.4 months, respectively (97). In the phase III FLAURA study involving 556 patients with untreated EGFR-mutated NSCLC, osimertinib significantly enhanced the objective response rate, median PFS, and median OS compared with gefitinib or erlotinib (98,99). Importantly, 49% of patients lost the T790M mutation during treatment with osimertinib, which predicted their resistance to osimertinib and worse outcome. This mutation is usually associated with the emergence of competing resistance mechanisms, such as KRAS mutations, c-MET amplification, small-cell transformation, and gene fusion (100,101). Furthermore, mutations (e.g., L792, L718, G719, and exon 20 insertions) may predict exceedingly poor responses to osimertinib and standard TKIs; however, their exact prognostic and predictive roles are not fully clarified (102)(103)(104). Some potential TKIs targeting these mutations, including roziotinib and TAK-788, remain under clinical investigation ( Table 1). It is assumed that the EGFR mutations of primary lung cancer could forecast the prognosis of patients with bone metastasis and tumor response to TKI therapy.
The ALK rearrangement is present in approximately 8% of NSCLC patients with oncogenic drivers (109). Dormieux et al. (88) demonstrated that the group with ALK fusions developed There was a significant relationship between serum NTx concentration and BM from solid tumors in the Chinese population.

I-CTP
Yokoyama et al.
Serum 87 I-CTP was significantly higher in patients with BM than those without BM. The most efficient cut-off value for I-CTP computed in this study was 6.4 ng/ml. Charpko et al.
The levels of bone pathological degradation (I-CTP) and bone formation reflected the extent of metastatic disease in bone measured by 99m Tc-MDP scans. Kong et al. (37) Serum 126 The levels of both b-CTX and I-CTP were significantly higher in lung cancer patients with BM (both p<0.001) and ICTP had a better sensitivity and accuracy than b-CTX. Tanaka et al. more brain and lung metastases compared with the negative group, with rates of 42% versus 29% and 37% versus 24%, respectively. Despite the absence of a direct correlation between ALK fusions and bone metastasis, the detection of these rearrangements contributes to the classification of target therapies for lung cancer patients with bone metastasis. The phase III PROFILE 1014 trial compared crizotinib with chemotherapy and demonstrated better objective response rate and PFS in ALK-positive patients (110). Moreover, the secondgeneration ALK-TKIs ceritinib, alectinib, and brigatinib, and the third-generation ALK TKI lorlatinib, showed an enhanced effect on the treatment of ALK-rearranged NSCLC compared with crizotinib. A variety of secondary ALK point mutations (e.g., L1196M, G1202R, S1206Y, and 1151Tins) were detected in patients with progressed disease, some of which predicted resistance to crizotinib (111)(112)(113). Recent clinical trials are presented in Table 2.
KRAS mutations are among the most commonly found mutations in NSCLC (20-30% of cases) (114)(115)(116). Its mutations mainly occur in codons 12, 13, and 61, which are thought to be diagnostic factors of bone metastasis of lung cancer. Renaud et al. (117) claimed that patients with KRAS G12C rearrangement developed significantly more bone metastases compared with those in the control group of the cohort (59% vs. 16%, respectively; p<0.0001) after thoracic surgery for NSCLC. Currently, there are two investigational KRAS G12C inhibitors, namely AMG 510 and MRTX849 ( Table 3), which have shown 40-50% response rate in some pre-treated patients (106,123,124). Further investigation of other specific therapies may reveal that KRAS mutations could also be useful for the classification of KRAS-targeted therapy.
Most of the EGFR mutations are measured by tumor tissue immunohistochemistry through wounded puncture or surgery. Yao et al. (125) found a correlation between EGFR mutations and 18 F-fluorodeoxyglucose positron emission tomographycomputed tomography imaging, and attempted to use this non-invasive inspection as an independent predictor. It was also verified that circulating tumor DNA (ctDNA) in the plasma can be used to detect EGFR mutations in patients with NSCLC, providing information similar to that obtained from biopsies (126). Therefore, the predictive effect of EGFR mutations, ALK rearrangements, and KRAS mutations are with great clinical potential.

C-Mesenchymal-Epithelial Transition
Factor/Hepatocyte Growth Factor (c-MET/ HGF) Signaling Axis c-MET is a receptor tyrosine kinase, and HGF is its sole ligand. The dysregulation of c-MET and HGF in tumor progression and invasion interplay with other signaling pathways, such as the EGFR pathway (92,118,119). The prevalence of MET amplification and MET exon 14 skipping mutation (METex14) is approximately 5% and 2%, respectively.
In 2009, Navab et al. (120) performed a study involving the overexpression of c-MET and HGF, resulting in enhanced metastases to distant organs (e.g., bone, brain, and kidneys). Following a series of studies, c-MET was thought to be a biomarker for NSCLC, which could be measured by immunohistochemistry and qualified with the H-score ranging from 0 to 300. Tsakonas et al. (119) demonstrated that a c-MET H score ≥20 is a positive prognostic biomarker for OS in patients with early-stage NSCLC, and could also be used to predict the effect of platinum-based adjuvant chemotherapy. Notably, Grano et al. (127) conducted the first important study describing the expression and roles of c-MET and HGF in osteoblasts and osteoclasts. Moreover, Whang et al. (128) demonstrated that the c-MET and HGF signaling axis played an essential role in the metastatic bone microenvironments. Moreover, the expression of c-MET and HGF was correlated with the progression of bone metastasis. The expression of this axis was also viewed as a prognostic biomarker, representing the clinical response of patients with advanced lung cancer and prostate cancer to c-MET-and vascular endothelial growth factor receptor 2 (VEGFR2)-targeted therapy (92,118,129). Currently, there is no approved therapy targeting c-MET, while several c-MET inhibitors (e.g., crizotinib, cabozantinib, tepotinib, telisotuzumab, etc.) are being investigated in clinical trials. Table 4 presents the details of these clinical trials. According to a retrospective study, MET inhibitors are associated with prolonged OS of patients with metastatic METex14 NSCLC (130,131). Secondary c-MET amplification is the most frequent cause of EGFR-TKIs tolerance. Following the administration of osimertinib as first-line therapy, MET amplification (encountered in 15% of patients) was the most common mechanism of intrinsic resistance (99). Moreover, >50% of the EGFR T790M-positive patients retained the c-MET amplification, and c-MET mutations are likely to be associated with EGFR C797S, CDK6, and BRAF amplifications (131,132). Despite the absence of specific research regarding the predictive effects of c-MET and HGF, their predictive roles in the prognosis of patients with bone metastases are worthy of further investigation.
Higher levels of CXCR4 are correlated with poor prognosis and a high risk of skeletal-related events in patients with NSCLC. The EGFR-mutated group had significantly more metastatic lesions in the brain and bone than the wild-type group. EGFR mutations were significantly more frequent in patients with multiple lung metastases than those with a single lung metastasis. Hendriks et al. (74) Tumor tissue 189 The incidence of brain and bone metastases was not different between EGFR+, KRAS+, and wild-type patients.
Tumor tissue 65 Univariate analysis suggested that high expression of CXCR4 was significantly correlated with bone metastasis (p=0.004). In addition, it was marginally correlated with brain metastasis (p=0.068) and lymph node metastasis (p=0.085), as well as worse OS (p=0.004) and PFS (p=0.005). Liao et al.
Tumor tissue 32 CXCR4 was highly expressed in the bone destruction area of metastatic NSCLC samples. Also, it was related to poor survival in NSCLC patients with bone metastasis, with an increase in VCAM1 and a decrease in ADAM17.
Tumor tissue 95 Patients with high levels of C5AR1 had significantly shorter RFS and OS (p<0.001). They found significantly higher C5AR1 levels in primary NSCLC tumors from patients who developed bone metastases during disease progression compared with those from patients who developed metastases to other non-skeletal sites.
Regarding patients with lung cancer, a univariate analysis conducted by Li et al. (133) suggested that high expression of CXCR4 was significantly associated with bone metastasis, as well as worse OS and PFS, and was marginally correlated with brain and lymph node metastasis. Zhou  Several studies have shown that the CXCR4/CXCL12 interaction may foster local tumor growth and metastatic potential by maintaining an immune-quiescent microenvironment, which could be blocked by CXCR4 antagonism (134)(135)(136)(137)(138). It was also reported that the majority of NSCLC tumors expressed CXCL12 in the cytomembranous compartment, and strong staining for CXCL12 was associated with a higher incidence of disease recurrence (139). Other studies suggested that high CXCL12 expression in lymph nodes promoted the metastasis of CXCR4expressing cells, benefiting from autocrine CXCL12/CXCR4 (121). Some meta-analyses have evaluated the whole effects of CXCL12/ CXCR4 expression on NSCLC. The results supported that CXCR4 expression is correlated with lymph node metastasis, distant metastasis, tumor stage, and OS (140)(141)(142). Collectively, these studies proposed that CXCR4 is a predictive biomarker for bone metastasis and its prognosis.
The CXCL12/CXCR4 axis greatly contributes to certain tumormediated metastases and is thought to be a potential target of CXCR4 antagonists (143,144). CXCR4 antagonists can be categorized into four major groups: nonpeptide CXCR4 antagonists (e.g., AMD3100); small peptide CXCR4 antagonists (e.g., BL-8040); antibodies to CXCR4 (e.g., LY 2624587 and ulocuplumab); and modified agonist CXCL12 antagonists (e.g., NOX-12A) (134,145,146). A series of clinical trials utilizing CXCR4 inhibitors were conducted in the field of hematological malignancies, while only a few clinical trials evaluated their safety and effectiveness in solid malignancies (147). Notably, small peptide LY2510924 exhibited limited clinical efficacy in the treatment of advanced SCLC (148). Therefore, it is essential to produce adequate evidence on the therapeutic role of CXCL12/CXCR4 in lung cancer and bone metastasis.
Moreover, the CXCL12/CXCR4 axis participates in other mechanisms of cancer progression. Recent findings indicated that CXCR4 expression on tumor cells plays a positive role in key downstream pathways, including PI3K/AKT/mechanistic target of rapamycin kinase (PI3K/AKT/MTOR) and extracellular signal-regulated kinase 1/2 (ERK1/2), which promote tumor proliferation and migration (149). Vascular cell adhesion molecule 1 (VCAM1) was also shown to be an activator of Elevated serum CEA and CYFRA levels appear to provide useful clinical information on the presence of BM and liver metastasis, as well as multiple-organ metastases, although they were not a powerful indicator of prognosis.

Tissue factor
Xia et al. CXCR4, with ADAM metallopeptidase domain 17 (ADAM17) as a downstream mediator (122). Zuo et al. (150) demonstrated that overexpression of CXCR4 enhanced cell motility and invasion via EGFR and matrix metallopeptidase 9 (MMP9), owing to the positive correlation among them. Furthermore, CXCR4 regulates the migration of lung cells through activation of Rac family small GTPase 1 (RAC1) and matrix metalloproteinases (MMP2 and MMP14), and through the action of inhibitor of kappa kinase (IKK), ERK, NF-kB, and integrins (151,152). Research has demonstrated the potential of CXCR4 antagonists to enhance the efficacy of immune checkpoint inhibitors in some types of solid tumor by inducing an immune suppressive tumor microenvironment (153)(154)(155). In summary, the predictive role of CXCL12/CXCR4 axis is reliable; however, the exact mechanism and therapeutic utilization of this axis warrant further verification.
Complement Component 5a/Complement Component 5a Receptor (C5A/C5AR) Axis The complement system represents a central component of innate immunity mediated by the proteolytic cleavage of C3 and C5. Cleaved products C3b and C5b participate in a variety of adaptive immune processes via membrane attack complex-dependent killing. Ajona et al. (107) demonstrated that blockage of the C5A/ C5AR axis led to a substantial improvement in the efficacy of antiprogrammed cell death 1 (anti-PD-1) immune responses in patients with lung cancer. Since then, C5AR-targeted therapy has been considered an advanced complementary therapy for clinical use. Furthermore, the C5A/C5AR axis is thought to be a reliable predictor of lung cancer bone metastasis. This is because patients who developed bone metastases had significantly higher C5AR1 levels in the primary tumor versus those with metastases to other non-skeletal sites during disease progression (108). To further investigate the mechanism of the C5A/C5AR axis, Vadrevu et al. (156) performed in vitro experiments and hypothesized that C5AR facilitates metastasis by suppressing effector CD8+ and CD4+ Tcells responses in the lungs. Recruitment of immature myeloid cells, generation of T regulatory cells, and secretion of factors, such as CXCL16, transforming growth factor beta (TGF-b), and interleukin 10 (IL10), may be involved in the mechanisms of this suppressor (108,157).

Notch Signaling Pathway
NOTCH3, a transmembrane receptor and a member of the Notch signaling pathway, plays an essential role in the development of lung cancer (158). It is overexpressed in approximately 40% of NSCLCs and has been associated with poorer disease-free survival and OS, higher TNM stage, poorer response to chemotherapy, and an increased rate of lymph node metastasis. Therefore, it is an ideal biomarker for the prediction of prognosis of NSCLC (105,(159)(160)(161). In NSCLC and SCLC, NOTCH3 signaling acts as a tumor-promoting and -suppressing pathway, respectively (162). The functional differences, which depend on the type of tumor, are worthy of exploration. Li et al. (163) found that upregulation of Wnt family member 3A (WNT3A) enhanced the expression of NOTCH3 and its downstream genes, hes family bHLH transcription factor 1 (HES1) and hes related family bHLH transcription factor with YRPW motif like (HEYL), and promoted metastasis of NSCLC. Liu et al. (164) demonstrated that NOTCH3 increased the invasion ability of NSCLC by upregulating zinc finger E-box binding homeobox 1 (ZEB1), which contributed to TGF-binduced transformation and bone metastasis in NSCLC. High levels of NOTCH3 are also associated with chemoresistance and resistance to radiotherapy, and inhibition of Notch enhances the sensitivity to EGFR kinase inhibitor (159,165). Notch signaling inhibitors, including MRK-003, represent new approaches for the targeted therapy for lung cancer. These inhibitors, to some extent, contribute to the diverse treatment options for bone metastasis (161). In summary, mRNA expression in tissue or immunohistochemistry staining of NOTCH3 from primary tumor tissue are likely to be stable predictive methods for prognosis and response to chemotherapy in patients with advanced NSCLC.

CXCL16/CXCR6 Axis
CXCL16 is one of the most extensively studied transmembrane chemokines. It is lowly and highly expressed in lung tissues and specimens of bone metastasis, respectively (166). CXCR6 functions as the receptor of CXCL16, and the CXCL16/CXCR6 signaling system is involved in the progression of tumor growth. Ha et al. (166) demonstrated that high CXCL16/CXCR6 expression in prostate tumor tissue may be related to aggressive cancer behavior, as well as high CXCL16 expression in bone metastasis. Similarly, Na et al. (167) revealed that high mRNA expression levels and immunohistochemistry staining of CXCR6 and CXCL16 in Ewing sarcoma family tumors was associated with tumor progression and lung metastasis. Ajona et al. (108) found that CXCL16 participated in the blockage of the complement C5A/C5AR axis in lung cancer bone metastasis, suggesting that CXCL16 is also a potential biomarker for the prediction of lung cancer progression and bone metastasis.

OTHER POTENTIAL BIOMARKERS
The bone formation and resorption markers have been used clinically for several years, based on their direct correlation with the progression of bone metastasis and clear mechanism. However, the efficacy and mechanism of some potential biomarkers remain unknown. Numerous studies have suggested their usefulness in the early prediction of bone metastasis and their clinical therapeutic or prognostic effect.

Bone Morphogenic Protein 2 (BMP2)
BMP2 promotes the migration and invasion of NSCLC cells. Firstly, BMP2 expression is higher in tumor tissues than in adjacent normal lung tissues (168). Subsequently, Choi et al. (169) demonstrated significantly higher levels of serum BMP2 in patients with NSCLC versus the control group. In addition, the levels of BMP2 in the advanced stage group were significantly elevated compared with those measured in the localized stage group. Moreover, serum BMP2 has been strongly associated with the objective response to chemotherapy (170). Recently, Huang et al. (171) performed a series of in vivo experiments to certify that activation of BMP2 signaling can enhance bone metastasis of Lewis lung cancer through both osteolytic and osteoblastic mechanisms. Thus far, no retrospective research has directly analyzed the role of BMP2 in lung cancer bone metastasis. Nevertheless, it may be a useful noninvasive biomarker capable of prognostic utility.

Calcium Sensing Receptor (CASR)
CASR is a widely expressed G protein-coupled receptor, which is critical for maintaining metabolic balance between bone and calcium content (172). Its expression is low or absent in normal lung tissue, whereas it is significantly higher in lung cancer tissue (173). Moreover, Lian et al. (174) demonstrated that the expression of CASR in 120 cancer tissues with bone metastasis was significantly higher than that recorded in non-metastatic lung cancer tissues. Another study found that the increase of CASR was the result of PTHrP and NF-kB upregulation, which was also positively correlated with bone metastasis in lung cancer (173,175,176). This evidence attracted attention to CASR as a new biomarker for the prediction of lung cancer bone metastasis.
Sclerostin Domain Containing 1 (SOSTDC1) SOSTDC1, secreted by particular helper T cells and reticular cells, plays a critical role in the development and progression of multiple types of cancer (177). Immunohistochemistry staining has indicated that SOSTDC1 is downregulated in NSCLC bone metastatic lesions compared with primary tumors (178). Moreover, Chen et al. (179) showed that overexpression of SOSTDC1 suppressed NSCLC migration, invasion, and osteoclast activity, whereas its knockdown led to the opposite effect. In addition, several downstream genes related to bone metastasis were detected using RNA-sequencing and quantitative RT-PCR (qRT-PCR) assays. Mechanistically, SOSTDC1 inhibits tumor progression by blocking the Wnt-b-catenin axis and facilitating T cell differentiation This finding suggests that SOSTDC1 is a potential prognostic biomarker for NSCLC bone metastasis (177).

HECT and RLD Domain Containing E3 Ubiquitin Protein Ligase 5 (HERC5)
HERC5 is a gene fragment located in the chromosome 4q22 and associated with the early presence of disseminated tumor cells in the bone marrow. Accumulating evidence has shown that the HERC protein family is a key component of a wide range of cellular functions, including neurodevelopment, DNA damage repair, cell growth, and immune response (180). Wrage et al. (181) initially found the same loss of chromosome 4q12-q32 in brain metastases from lung cancer. Subsequently, they narrowed this loss to gene HERC5 using qRT-PCR. In addition, hypermethylation of the HERC5 promoter was associated with poor survival in patients with early-stage and metastatic lung cancer. Thus, HERC5 was thought to be a new metastasis suppressor gene, whose methylation and expression status may provide prognostic information for bone metastasis of NSCLC.

Cytokeratin 19 Fragment/ Carcinoembryonic Antigen (CYFRA/CEA)
CYFRA is one of the characteristic markers for the diagnosis of NSCLC. CEA is also widely used in clinical practice as a serum marker for NSCLC. Recently, Numata et al. concluded that elevated serum CEA and CYFRA levels were associated with the presence of bone and liver metastasis and also multiple-organ metastases; however, these high levels were not a powerful indicator of prognosis (38). Furthermore, serum CEA and CYFRA had predictive value with regard to response to therapy in NSCLC, and decreased levels of CYFRA are indicative of objective response (182). Hence, CEA and CYFRA were utilized in some clinical research studies as positive contrast to other markers (31,35). The use of CEA and CYFRA presents opportunities for risk stratification of patients and may aid in the clinical management of the disease.

Tissue Factor (TF)
TF, produced during the development of embryo and normal hemostasis, plays an essential role in regulating platelet activation, fibrin deposition, and the extrinsic coagulation cascade by binding to factor VIIa (183). It has been demonstrated that TF-induced fibrin deposition was positively associated with tumor progression by affecting complement activation and the recruitment of myeloid-derived suppressor cells (184). Xia et al. (185) reported that patients with high TF expression levels tended to have worse OS, and downregulation of TF inhibited the invasion and migration of NSCLC cells in vitro and in vivo. This evidence suggests that TF is an effective biomarker for predicting the prognosis of patients with NSCLC.

Cell-Free DNA (cfDNA)
Nonencapsulated extracellular DNA fragments, termed cfDNA, have been found in body fluids. These fragments could also be narrowly defined as ctDNA in patients with solid malignancies (186,187). cfDNA is thought to be generated during the process of cell apoptosis and necrosis, with a short half-life ranging 16 min-2.5 h. These features make it an ideal predictive biomarker for the early detection of lung cancer (188,189). The use of cfDNA is attracting attention, as studies have shown that it can identify and differentiate the heterogeneous nature of different metastatic sites (190).
Notably, higher concentrations of plasma cfDNA were detected in the peripheral blood of patients with bone metastasis. These higher levels are thought to be associated with an increased risk of distant migration and worse clinical outcome (191)(192)(193). Similarly, lung cancer patients without bone metastasis had significantly reduced urinary cfDNA and longer OS (194). Moreover, as the cfDNA turnaround time was significantly shorter than that of tissue-based genotyping, cfDNA was shown to be functional in the clinical management of the disease in the National Comprehensive Cancer Network (96).
In particular, cfDNA could be used as a non-invasive method to detect EGFR mutations in patients with NSCLC, with similar accuracy to that of tumor tissue biopsies. Dynamic changes in the cfDNA EGFR mutation status are associated with the clinical outcomes of treatment with EGFR-TKI (126,(195)(196)(197)(198)(199).
Long Non-Coding RNA (lncRNA) LncRNA are the products of DNA transcription with limited or no protein-coding capacities. However, lncRNA has a variety of functions in the regulation of biological processes, such as migration, proliferation, apoptosis, and invasion (200).
The expression levels of lncRNA in malignant tissues often differ significantly from those measured in normal tissues and correlate with tumor staging. Recently, studies utilizing high-throughput transcriptome analysis (RNA sequencing) indicated that lncRNA, including metastasis associated lung adenocarcinoma transcript 1 (MALAT1), PXN-AS1-L, and SUMO1 pseudogene 3 (SUMO1P3), etc., were significantly highly expressed in NSCLC tissues with bone migration. These findings suggested that lncRNA are of great predictive value in bone metastasis (201)(202)(203). However, the specific mechanisms involved in this relationship remain unknown and require further studies. Li et al. demonstrated that the plasma levels of lncRNA HOX transcript antisense RNA (HOTAIR) were higher in NSCLC samples versus normal specimens (204). Thus, lncRNAs can be utilized as biomarkers in NSCLC with or without bone metastasis.

MicroRNA (miRNA)
MiRNAs are a class of small, single-stranded, no-coding RNAs, which were recently regarded a research hotspot. They act as gene regulator by binding to mRNAs and inhibiting their functions (205). Plenty of studies revealed that miRNAs play significant roles, including both oncogenes and tumor suppressors, during the progression and metastasis of lung cancer (206)(207)(208).
Overexpression of the let-7 family, miR-486, miR-218, miR-34, and miR-200 were proved to inhibit the tumor cell proliferation, invasion and colony formation, though there is no specific article describing their relations with tumor cells bone migration (209)(210)(211)(212). In the opposite, some miRNAs take oncogenic roles, containing miR-196b, miR-221/222, miR-21, miR17/92, and miR-224 (213)(214)(215)(216)(217)(218). The genetic alterations or epigenetic changes of miRNAs are closely implicated in the metastasis of NSCLC, which identify more target therapies for lung cancer bone metastasis.  Table 5. Table 6 lists ongoing clinical trials assessing the efficacy of ALK inhibitors, including alectinib, brigatinib and lorlatinib which target advanced/metastatic ALK-positive NSCLC. Table 7 shows two ongoing clinical trials of AMG510 and MRTX849 targeting lung cancer patients with the KRAS G12C mutation. Table 8 summarizes information on ongoing clinical trials assessing the efficacy of MET inhibitors, such as cabozantinib, tepotinib, capmatinib, telisotuzumab, crizotinib, sacolitinib, and rilotumumab. Table 9 presents the ongoing clinical trials involving patients with bone metastatic NSCLC treated with the RANK-targeting denosumab. To some extent, these clinical trials offer opportunities for late-stage patients and those who cannot afford curative tumor resections. There is no approved therapy targeting c-MET, while several c-MET inhibitors (e.g., crizotinib, cabozantinib, tepotinib, telisotuzumab, etc.) are being investigated in clinical trials. Table 10 presents the details of these clinical trials. According  to a retrospective study, MET inhibitors are associated with prolonged OS of patients with metastatic METex14 NSCLC (130,131). Secondary c-MET amplification is the most frequent cause of EGFR-TKIs tolerance. Following the administration of osimertinib as first-line therapy, MET amplification (encountered in 15% of patients) was the most common mechanism of intrinsic resistance (99). Moreover, >50% of the EGFR T790M-positive patients retained the c-MET amplification, and c-MET mutations are likely to be associated with EGFR C797S, CDK6, and BRAF amplifications (131,132). These phenomenon provide opportunities for the combined treatment of lung cancer patients. For the future, therapies utilizing lncRNA is under development, which works by affecting transcriptome directly. Studies suggested that lncRNA was associated with resistance to both chemotherapy and targeted therapies (219). For example, lncRNA LINC00460 promotes EGFR expression by downregulating miR-769-5p, which results in the resistance of NSCLC cells to gefitinib (220). LncRNA MALAT1 upregulates multiple drug resistance-related protein (MDR1) and multidrug resistance protein (MDR1) by inducing STAT3 phosphorylation, which eventually enhance the resistance of lung cancer cells to treatment with cisplatin (221,222). Though there is no specific clinical trial targeting lung cancer, it's still a potential therapy for the comprehensive treatment of lung cancer.

CONCLUSION
Patients with lung cancer bone metastasis rarely receive effective targeted therapy at an early stage due to delayed diagnosis. Therefore, the discovery of more accurate diagnostic methods is warranted. The bone formation and resorption markers have already been applied to clinical practice, and greatly aid in improving clinical management. For example, liquid biopsy of circulating tumor cancer, circulating cell-free nucleic acids, and extracellular vesicles are a promising source of prognostic and therapeutic biomarkers for metastatic lung cancer (223)(224)(225). However, thus far, most of the signaling axes and other markers remain under investigation without a recognized standard due to inconsistent results and methodological differences. To combine the advantages of these markers, some research studies utilized meta-analysis to identify clinical criteria and construct a clinical model. However, the limited number of samples and marker types restrict the reliability of this model (32,52,74). Consequently, it is necessary to integrate clinical  resources and utilize larger sample sizes, as well as unified standards for the selection of patients and evaluation of bone metastasis and clinical outcomes. This approach would allow researchers to determine the ideal, inexpensive biomarkers for the evaluation of skeletal metastasis and skeletal-related events. Moreover, combination therapies targeting these biomarkers are currently under investigation in clinical trials. Based on the available evidence regarding these biomarkers, the prognosis of treatment using targeted therapies is promising. Further studies concerning the mechanisms of these biomarkers will contribute to the development of effective therapies against bone metastasis.

AUTHOR CONTRIBUTIONS
XC, EY and ZeW wrote the manuscript. ZhW, BL, and YZ provided expertise and advice. HZ, YWE, SW and WZ critically read the manuscript. ZY supervised the project. All authors contributed to the article and approved the submitted version.

FUNDING
Tumor tissue 120 CASR expression in lung cancer tissues was significantly higher than that measured in adjacent and normal lung tissues. The expression of CASR in lung cancer tissues with BM was higher than that observed in non-metastatic lung cancer tissues. BSP Bellahcene et al. (33) Primary lung tumor tissue 86 BSP was not specifically detected in normal lung tissue with the exception of cartilage associated with bronchi. Most adenocarcinoma (74%) and all squamous carcinoma of the lung samples examined exhibited detectable levels of BSP.
Zhang et al. (70) Primary lung tumor tissue 180 BSP protein expression in the primary resected NSCLC was strongly associated with BM and could be used to identify high-risk patients after primary tumor resection.
He et al.
Serum 146 The mean serum BSP levels in individuals with BM were significantly higher than those recorded in non-BM NSCLC and controls (p<0.001). The cut-off value was 33.56 ng/ml, and sensitivity and specificity values were 77.8% and 81.1%, respectively. BMP2 Bieniasz et al. (168) Tumor tissue The expression levels of VEGF, BMP2, and BMP4 mRNA were significantly higher (7.1-fold, 25.6-fold, and 2.3-fold, respectively) in lung cancer samples than those in adjacent normal lung tissues. Choi et al. (169) Serum 150 The NSCLC group demonstrated significantly higher levels of serum BMP2 than the control group. The median serum levels of BMP2 in the advanced stage group (stage IIIb or IV) were significantly elevated compared with those of the localized stage group (stages I, II, and IIIa). Elevated serum CEA and CYFRA levels appear to provide useful clinical information on the presence of BM and liver metastasis, as well as multiple-organ metastases, although they were not a powerful indicator of prognosis.

Tissue factor
Xia et al.

Serum 100
Patients with high tissue factor expression levels tended to have worse overall survival performance, and downregulation of tissue factor inhibited the invasion and metastasis of NSCLC cells in vitro and in vivo. Cell-free DNA (cfDNA) Pecuchet et al. (192) Serum 124 The presence of circulating tumor DNA at baseline was an independent marker of poor prognosis.