Abstract
Lung cancer is one of the most common neoplasms and the leading cause of cancer-related deaths worldwide. Despite recent progress in understanding the pathomechanisms of lung cancer, it is frequently associated with late diagnosis, high incidence of metastases and poor response to treatment. Thus, there is extensive research in the field of biomarkers that aims to optimize management of lung cancer. The aim of this study was to review the current perspectives of a wide spectrum of circulating molecules that seem promising as new potential biomarkers of lung cancer. Among these, biochemical (active proteins), immunological (immunocompetent cells, cytokines, chemokines, and antibodies) and genetic (circulating tumor DNA, cell-free DNA and microRNA) markers are presented and discussed. The use of these markers would support the early detection of lung cancer and might be used for predicting disease progression, response of the disease to targeted therapies, monitoring the course of treatment, and developing individualized diagnostic and therapeutic strategies. Special attention was given to potential markers of nervous system involvement in the course of lung cancer, due to its prevalence and devastating impact. Limitations of the potential biomarkers are also outlined and future directions of investigations in this field highlighted, with the aim of improving the accuracy and practical utility of these biomarkers.
Key words: lung cancer, biomarkers, molecular, biochemical, immune
Introduction
According to the GLOBOCAN estimates,1 lung cancer was the 2nd most common type of cancer in terms of incidence in 2020, accounting for 11.4% of all newly diagnosed cancer cases. It was also by far the leading cause of death due to malignancies, accounting for 18% of all cancer mortality, which is almost double that of the 2nd most common cause, colorectal cancer. Moreover, it is estimated that lung cancer will remain at the top of both of these categories by 2040, with an expected growth of 58.8% in the number of cases and 63.8% in mortality.2 From a histological point of view, lung cancer is typically divided into subtypes: small cell lung cancer (SCLC), adenocarcinoma, squamous cell carcinoma (SCC), and large cell carcinoma, the latter 3 usually being jointly referred to as non-small cell lung cancer (NSCLC).3 The main symptoms of lung cancer, that can occur separately or in combination, are cough, dyspnea, pain, hemoptysis, aphonia or hoarseness, weight loss or asthenia, and superior vena cava syndrome, although an asymptomatic course at the time of diagnosis is not unusual.4 Moreover, distant metastases are frequent, especially to the central nervous system. In the early stages of NSCLC, brain metastases are present in 0.6–3% of patients5 and this increases up to 50% in the course of the disease.6 In SCLC, brain metastases occur in about 10% of patients at the time of diagnosis and in additional 40–50% at later stages.7
Another type of nervous system involvement, resulting from immune-mediated responses to the presence of lung cancer antigens, are paraneoplastic neurological syndromes (PNS). Lung cancer, predominantly SCLC, is considered to be the most common malignancy associated with PNS.8, 9 From a clinical perspective, PNS can involve both the central and peripheral nervous systems, with the most commonly reported syndromes being peripheral neuropathy, followed by limbic encephalitis, subacute cerebellar degeneration, Lambert–Eaton syndrome, myopathy, encephalomyelitis,8, 9, 10 and neuromyelitis optica.11, 12 Paraneoplastic neurological syndromes may develop a few years before the detection of cancer,13, 14 which highlights the potential of using such syndromes for early diagnosis.15, 16
Treatment options for lung cancer can be applied alone or in combination and include surgery (for early stage disease), chemotherapy, radiotherapy, targeted therapy, and immunotherapy.17, 18, 19, 20, 21 Despite recent advances in the diagnosis and treatment of lung cancer, the prognosis is still unfavorable. According to the tumor-node-metastasis (TNM)-based staging of lung cancer, the 5-year survival rate for NSCLC varies between 50% in clinical stage IA and 2% in clinical stage IV.22 Small cell lung cancer is associated with even worse outcomes, such as a 5-year survival of 10% in the early stage of disease, with only 4.6% of patients diagnosed in the extensive stage surviving 2 years.23 As a consequence of high mortality rates and frequency of metastases present at diagnosis, much of the recent research has focused on early diagnosis and identification of potential markers of disease progression, local infiltration and metastatic activity, as well as treatment response. The early diagnosis of lung cancer is based mainly on computed tomography imaging, confirmed using cytological and histopathological examination of specimens obtained during bronchoscopy or other invasive procedures.24 However, the diagnostic process and prognosis may be complemented by additional biomarkers.
By definition, biomarkers are molecules or abnormal parameters that distinguish an individual with a particular disease from the studied population. Biomarkers can be detected in bodily fluids such as blood, serum, urine, sputum, pleural effusion, or cerebrospinal fluid.25, 26 Recently, biochemical, immune and molecular biomarkers have been recognized as the most promising and clinically relevant with regard to lung cancer, and they are being extensively investigated to evaluate their sensitivity and specificity.27 An early detection of the dissemination of neoplastic processes and the establishment of risk factors for its occurrence are particularly important in terms of prognosis and therapeutic possibilities. Given the significant impact of nervous system involvement on disease burden, morbidity and mortality, the identification of its presence and selection of patients at increased risk of this complication are of great importance.
Objectives
The aim of this study was to review the current data on the role of new biochemical, immune and molecular markers in the diagnosis of lung cancer, and to evaluate its progression, with a focus on the involvement of nervous system in the course of disease. Ongoing research and its future directions in this field have been reviewed in view of potential implications for early detection of cancer, tailoring treatment plans based on prognosis, and monitoring the course of disease.
Materials and methods
A literature search was performed using the PubMed and Embase databases, covering the period from the beginning of 2010 until February 28, 2022, with a combination of the search terms: “lung cancer”, “NSCLC”, “SCLC”, “biomarker”, “biochemical”, and “molecular”. After excluding papers written in a language other than English, conference abstracts and duplicates from further screening, a total of 2745 original studies and review articles were retrieved. Full texts of eligible papers were analyzed for their relevance to the topic, as well as several further potentially relevant papers that were identified in reference lists from the texts. Initially, the literature search was conducted by the lead author, with the results reviewed and verified by the other authors. This led to the identification and inclusion of 217 published studies that were considered the most relevant to the topic. The preparation of the study was conducted by following the Enhancing Transparency in Reporting the Synthesis of Qualitative Research (ENTREQ) checklist,28 selected according to the Enhancing the QUAlity and Transparency of Health Research (EQUATOR) Network guidelines (https://www.equator-network.org).
Biochemical markers
of lung cancer
Several biochemical biomarkers have already been implemented into lung cancer diagnostics and management, including carcinoembryonic antigen (CEA), cytokeratin 19 fragment marker (CYFRA 21-1), neuron-specific enolase (NSE), and cancer antigen 125 (CA-125).29, 30 However, the sensitivity and specificity of these markers are disputable, as their levels can be elevated in other diseases. As such, new candidate biomarkers have been proposed that are thought to have better accessibility and clinical utility, such as soluble intercellular cell adhesion molecule-1 (sICAM-1), which plays an important role in adhesion between host cells and cancer cells in the promotion of tumor growth. The overexpression of sICAM-1 was reported in lung cancer patients with lymph node and distant metastases, and was linked to shorter overall survival (OS) and progression-free survival (PFS).31 Similarly, high levels of angiopoietin-2, an important factor involved in angiogenesis, were associated with lymph node metastases and a poorer prognosis.32 Transforming growth factor beta (TGF-β),33 glucose transporter 1 (GLUT1), which enhances the supply of glucose to tumor cells,34 and urinary GM2 activator protein (GM2AP), a molecule involved in the induction of cancer invasion,35 are other recently proposed predictors of poor outcome. Podoplanin, a potential inhibitor of tumor cell growth and self-renewal, was identified as a marker of lower malignancy in SCC and better prognosis in patients with this type of lung cancer.36 In another meta-analysis, high serum levels of amyloid A, a protein correlated with an acute inflammatory response, were suggested as a discriminative marker, especially for the detection of SCC.37 Among other potential biomarkers, tumor necrosis factor receptor-associated protein 1 (TRAP1) was overexpressed in patients with higher pathological TNM stage and lymph node metastases, and was correlated with a shorter disease-free survival.38
Immune markers of lung cancer
The development of lung cancer is associated with a changing profile of immune system activity, with a shift from type 1 T helper cell-derived signaling to type 2 T helper cell pathways. Furthermore, dendritic cell, natural killer (NK) cell and T helper cell activity has been shown to decrease, whilst regulatory T cell (Treg) activity has been seen to increase. Additionally, programmed cell death protein 1 (PD-1) and cytotoxic T-lymphocyte-associated protein 4 (CTLA-4), which are markers of checkpoint inhibition, have also been shown to increase. Meanwhile, tumor-specific antigens and tumor-associated antigens (TAAs) are expressed by neoplastic cells and evoke an immune response, such as the induction of antibody production. Therefore, these immune-mediated processes might be used as markers for detection and monitoring of lung cancer, or for predicting its activity.39, 40
A range of immunocompetent cells have been identified within lung tumors, with their type and distribution in the nest or stroma of the tumor found to have prognostic significance. The prevalence of Tregs, M2 macrophages and immature dendritic cells was associated with poor survival, while the presence of CD8+ T cells, CD4+ T cells, M1 macrophages, and NK cells was linked to better outcomes.41 Some of these findings were specific for particular types of lung cancer, with high intratumoral neutrophil density being correlated with poor prognosis in patients with adenocarcinoma, contrary to the patients with SCC.42 A similar analysis of bronchoalveolar lavage fluid demonstrated material obtained from the affected lung that contained an increased number of neutrophils and a predominance of CD8+ T cells, Tregs and M2 macrophages.43
Measures of the systemic inflammatory response from peripheral blood, including monocyte count and neutrophil-to-lymphocyte ratio (NLR), might also serve as predictors of tumor development, especially its propensity to metastasize.44 Indeed, a meta-analysis of 14 studies revealed that a high NLR was associated with shorter OS in NSCLC and SCLC patients.45 Additionally, flow cytometry studies have demonstrated that CD3+, CD4+ and CD4+/CD8+ ratio, and NK cells were all decreased, and inversely correlated with the progression of clinical stage in NSCLC, while Tregs increased parallel to cancer progression.46
Emphasis has been put on measuring serum cytokine and chemokine levels, which reflect the inflammatory processes related to the development of the cancer, in both NSCLC47 and SCLC.48 Higher serum levels of interleukin (IL)-6 and IL-8 predicted a risk of lung cancer up to several years before the diagnosis,49 whilst the expressions of IL-8 with IL-6 and IL-6 with IL-17 were shown to be negative prognostic factors for early-stage lung cancer.50 In addition, elevated levels of IL-17 in the serum of SCLC patients correlated with a propensity to metastasize and a shorter OS.51 Angiogenesis inhibitors IL-20, and IL-22, which promotes tumor growth, were also found to be prognostic factors of lung cancer outcomes. High serum levels of IL-20 adversely correlated with time to cancer progression, and lower levels of IL-22 in the bronchoalveolar lavage of NSCLC patients were associated with worse rates of survival.52 Studies on the prognostic value of chemokines have demonstrated that high levels of C-C motif chemokine ligand 2 (CCL2), CCL19 and C-X-C motif chemokine ligand 16 (CXCL16), and low levels of CCL5 were linked with better survival. In contrast, high levels of CXCL8 and C-X-C motif chemokine receptor 4 (CXCR4) were associated with worse survival rates.53
Autoantibodies to TAAs may be detectable in the asymptomatic stage of cancer and persist in high levels in serum, which indicates their potential use as biomarkers for early detection of lung cancer.54 The Early CDT®-Lung is a panel test for the presence of 7 autoantibodies against TAAs (p53, NY-ESO-1, CAGE, GBU4−5, SOX2, HuD, and MAGE A4) that is currently used in patients with a high risk of lung cancer. The test was validated in large cohorts of NSCLC and SCLC patients and demonstrated high overall specificity, but a rather low sensitivity in SCLC, which was even lower in NSCLC.40 The presence of relevant autoantibodies may also have a prognostic value in lung cancer, with 1 study reporting on patients with NSCLC who were positive for antineural and antinuclear antibodies, and showed better rates of survival.55 Another study reported a panel of 4 antibodies (MAGEA1, PGP9.5, SOX2, and TP53) that were overexpressed in NSCLC and correlated negatively with OS.56 In another report, levels of an antibody against human DNA-topoisomerase I were significantly higher in the NSCLC group than in the controls, though the prognosis was worse in the antibody-negative group.57
Studies on the tumor-related immune environment have identified antibodies against PD-1 and its ligand (PD-L1) as a potential therapeutic option in lung cancer. Thus, the value of PD-L1 was investigated as a predictive biomarker for the efectiveness of therapy with anti-PD1/PD-L1 agents. Higher levels of PD-L1 expression were shown to be associated with a more effective treatment and longer survival, although results have not been consistent across studies.58 At the same time, a high expression of CTLA-4 predicted worse survival in NSCLC but was not validated as a predictive marker of the response to the anti-CTLA-4 treatment.41 Other markers have been sought, with Zhou et al. constructing a panel of 5 tumor-associated autoantibodies (p53, BRCA2, HuD, TRIM21, and NY-ESO-1) designed to predict the response to immune checkpoint inhibitors.59 Panel positivity was found to be indicative of a better response and longer PFS. In the previous work, the same association was established for NY-ESO-1 and XAGE1 serum antibodies,60 anti-nuclear antigens, extractable nuclear antigens, and anti-smooth cell antigens.61 In another study, SIX2 autoantibodies were consistently upregulated in the non-responder group.62 A panel comprising 13 antibodies showed high accuracy in predicting poor outcome in pre-operative samples of NSCLC patients (stage I–IIIa).63 Concordant prognostic utility was confirmed for an antibody against cyclin Y.64 Moreover, autoantibody status was suggested to be helpful not only in predicting clinical outcome, but also in assessing the risk of immune-related adverse events during treatment.65
Different types of biomarkers can be combined to further improve their diagnostic and predictive value, with 1 report establishing a panel of markers that could identify patients at risk of lymph node metastases.66 The panel included tumor necrosis factor alpha (TNF-α), tumor necrosis factor-receptor I (TNFR1) and macrophage inflammatory protein-1α (MIP-1α), along with 3 autoantibodies that target ubiquilin-1, hydroxysteroid-(17-β)-dehydrogenase and triosephosphate isomerase. Validation of the panel using a classification algorithm revealed a sensitivity of 94% and specificity of 97%. A meta-analysis on advanced lung cancer inflammation index, which is a prognostic score that considers body mass index (BMI), serum albumin and NLR, revealed a significant correlation between the score and OS and PFS.67 The prediction of survival in SCC patients was proposed on the basis of 4 immunological markers, including monocyte ratio, NLR, PD-L1 immunostaining score, and PD-1-positive stained tumor-infiltrating lymphocyte counts.44 A large study was performed comprising patients with NSCLC, treated with PD-1/PD-L1 inhibitors, in order to establish the potential efficacy of 2 combined biomarkers, defined as Lung Immune Prognostic Index (LIPI). This index measured derived neutrophil/(leukocyte minus neutrophil) ratio and lactate dehydrogenase levels in order to predict the resistance to immune checkpoint inhibitors.68 Both of these factors were independently associated with worse OS and PFS in patients treated with immune checkpoint inhibitors, while no such correlation was observed in a group treated with chemotherapy only. Neutrophil-to-lymphocyte ratio was shown to correlate not only with shorter OS but also with the presence of Kirsten rat sarcoma viral oncogene homologue (KRAS) and epidermal growth factor receptor (EGFR) mutations.69 Lung Immune Prognostic Index70 and NLR71 have also been investigated for their prognostic value in patients with SCLC. The aforementioned potential biochemical and immune biomarkers of lung cancer are summarized in Table 1.
Circulating tumor DNA
and circulating tumor cells
as lung cancer markers
Circulating tumor DNA
Circulating tumor DNA (ctDNA) enters the bloodstream predominantly as a result of necrosis and apoptosis of tumor cells, although there is also evidence that it can be actively released by viable cells and several other processes.72 It usually constitutes a small fraction (0.1–1%) of cell-free DNA (cfDNA) in plasma73; however, its level reflects tumor activity and expansion and can be much higher in patients with a more advanced disease.74 There is increasing interest in using ctDNA in the diagnosis of various types of neoplasms, including lung cancer, and for monitoring the course of disease.75 The method of obtaining ctDNA from plasma, known as liquid biopsy,76 is considered a promising alternative to standard tissue biopsy. This noninvasive and safe technique may be easily implemented in all patients, even those for whom a traditional biopsy is not possible, and it enables avoiding complications such as pneumothorax, hemorrhage and air leaks.77, 78
Rapid advances in molecular techniques that used to detect cancer-specific mutations in cfDNA, such as polymerase chain reaction (PCR) or next-generation sequencing, have offered new perspectives of on implementing liquid biopsies into clinical practice.79 In a large analysis of data from over 8000 lung cancer patients, ctDNA profiling revealed somatic alterations in 86%, and identified driver oncogene mutations in 48.4% of them.80 Furthermore, ctDNA profiling has been used to distinguish between benign and malignant lung tumors, and to detect lung cancer at an early stage. Indeed, the assay based on deep sequencing detected 63% of stage I and 83% of stage II lung cancers, respectively.81
In a study by Liang et al., the analysis of DNA methylation patterns was performed using tissue samples from patients with lung nodules in order to distinguish between malignant and benign tumors.82 A predictive model based on 9 methylation markers for ctDNA was then applied to plasma samples, with a sensitivity of 79.5% and specificity of 85.2% for detecting lung cancer. Regarding its subtypes, the sensitivity was 73.9% for adenocarcinoma and 100% for SCC. This difference may be explained by higher intensity of necrotic processes observed in SCC tissue, which results in a greater release of ctDNA into the bloodstream, therefore being eligible for analysis. Existing data support the utility of ctDNA methylation analysis in detecting early-stage lung cancer,83 and a subsequent study on a large group of lung cancer patients is being conducted to develop a ctDNA methylation classifier for incidental lung nodules.84 Longitudinal methylation profiling along with somatic mutation analysis in patients with NSCLC have also shown prognostic potential in assessing the risk of recurrence.85
With regard to its prognostic value, the level of ctDNA was found indicative of lymph node involvement in resectable NSCLC.86 Other investigators collected tissue and plasma samples from NSCLC patients before and after surgery in order to identify driver mutations in genes, including EGFR, KRAS, TP53, BRAF, PIK3CA, and ERBB2.87 Out of 46.3% of plasma samples which were positive for ctDNA before tumor resection, a significant decrease in mutation frequency was noticed, from 8.88% before surgery to 0.28% after the procedure. Furthermore, ctDNA was more prominent in stage Ia and Ib cancers than in more advanced stages. In a follow-up study of surgically treated lung cancer patients, targeted mutations were present in 93% of patients before surgery and in 54% at some point after surgical resection. Interestingly, all of the patients with ctDNA still detectable after surgery experienced progression of the disease, while those without ctDNA remained disease-free.88
Use of ctDNA in detecting minimal residual disease was demonstrated in a study where multiplex-PCR assay panels were used to screen for ctDNA in plasma samples of early-NSCLC patients, pre- and postoperatively.89 A sample was considered ctDNA-positive if at least 2 pre-established single nucleotide variants were detected. Circulating tumor DNA was found in 48% of pre-operative samples, and the detection rate was substantially higher for SCC (97%) than for adenocarcinoma (19%). Again, this discrepancy may be due to less extensive necrotic processes in the latter. Moreover, significant correlations were observed between the results of postoperative ctDNA profiling and the occurrence of clinical relapse or resistance to chemotherapy.90, 91, 92, 93, 94 The use of ctDNA profiling has also been researched in SCLC, although to a lesser extent. In a Chinese study, SCLC patients with higher ctDNA levels had significantly shorter PFS and OS.95 This relationship between ctDNA detection and poor prognosis has been also been observed in other research.96 The potential role of ctDNA in SCLC detection and progression monitoring was further strengthened by a large ctDNA analysis in over 10,000 cancer patients. In this group, the highest detectability of ctDNA in all cancer types was in SCLC, reaching 91.1%.97
The role of ctDNA profiling is also gaining attention in tailoring and monitoring of lung cancer treatment, and several liquid biopsy tests have been developed for this purpose.98, 99, 100 This method can be used before applying adjuvant chemotherapy, which is considered an option in NSCLC, to identify eligible patients.101 Based on recent understanding of the mechanisms of resistance to tyrosine kinase inhibitors (TKIs),102 ctDNA analysis may be a promising tool in this area. Circulating tumor DNA analysis has also been used to investigate resistance mechanisms in patients with NSCLC treated with rociletinib, a 3rd generation EGFR inhibitor.103 Multiple resistance mechanisms to the drug were present in 46% of patients, while at least 1 such mechanism was found in 65% of them, with MET copy number gain being the most common, as it was found in 26% of the patients. In another experiment, researchers were able to identify driver and resistance mutations through next generation sequencing of ctDNA, even when tissue sequencing was not successful.104 Furthermore, there is also some evidence for the detection of T790M mutation in ctDNA profiling in patients with T970M-negative tissue.105 These observations support the potential of ctDNA not only as a supplementary method, but also as an independent screening tool that could be applied in the planning of individualized treatment strategies.
Detectable EGFR mutations in cfDNA were associated with a longer PFS in response to treatment with erlotinib, a TKI, while its persistence in a follow-up plasma analysis resulted in shorter PFS and OS.106 In another study comprising patients treated with erlotinib, EGFR T790M mutations linked to TKI treatment resistance were detectable in cfDNA even before disease progression.107 Several other studies have also underlined the potential role of ctDNA profiling in the detection of resistance mutations as a part of disease monitoring.108, 109, 110, 111, 112, 113 Changes in ctDNA profile demonstrated a good predictive value in a study by Nabet et al., where plasma samples were analyzed in patients with advanced lung cancer treated with immune checkpoint inhibitors.114 A significant (at least 50%) drop in detectable ctDNA levels at 4 weeks after the initial treatment was considered a molecular response and helped identify patients with durable clinical benefit, defined as PFS of at least 6 months. Similar results were found by other authors, underlining the association between ctDNA decrease and better PFS and OS.115, 116, 117, 118 Accordingly, baseline and post-treatment ctDNA indicated worse clinical outcomes.119, 120, 121, 122 Circulating tumor DNA has also been investigated in the evaluation of tumor mutation burden (TMB), a novel predictive marker reflecting the total number of existing mutations, which is thought to be predictive of the response to PD-1 and PD-L1 inhibitors. It was hypothesized that patients with a higher burden of somatic mutations would benefit from immune checkpoint inhibitors due to a better recognition of neoantigens. This beneficial effect was confirmed with tumor tissue analyses of NSCLC patients treated with pembrolizumab,123 nivolumab or ipilimumab.124 The evaluation of blood-based TMB, assessed with ctDNA genetic profiling, revealed complementary findings. Non-small cell lung cancer patients with high blood-based TMB treated with atezolizumab showed a better response to the therapy in 1 study,125 while in another report,126 higher TMB was found to correlate with shorter PFS and OS in NSCLC. These diverse results point to potential limitations of ctDNA analysis, such as a small possible range of mutations that can be detected using liquid biopsy. Future improvements to the method should include establishing validated sequencing panels and cut points.127
As a potential marker of lung cancer diagnosis and progression, ctDNA was also compared to previously known biomarkers and showed a higher detection rate and positive predictive value than CYFRA21-1, CEA, NSE, SCC, CA-125, and CA19-9.87 Concordant results were obtained in a similar study, where plasma samples were taken before, during and after surgery.128 In this study, the sensitivity of ctDNA detection was higher than for protein tumor markers (63.2% compared to 49.3%), and a significant drop in the average ctDNA mutation frequency after surgery was also reported.
Circulating tumor cells
Apart from ctDNA, the so-called “liquid biopsy” techniques may also reveal circulating tumor cells (CTCs) that originate from primary or metastatic tumors.129 As CTC numbers in plasma are very low, they may be detected by means of various methods, including immunomagnetic separation with EpCAM- or CD45-based assays, PCR or telomerase-based assays, as well as cellular isolation with size-dependent filters.130 A recently published meta-analysis, including 21 studies with almost 4000 participants, demonstrated high pooled sensitivity and specificity of CTCs in lung cancer detection.131 There is also some evidence of CTC role as a potential marker of lung cancer progression and dissemination, as a higher abundance of detectable CTCs before the commencement of treatment resulted in shorter OS and PFS in NSCLC patients.132 In another meta-analysis,133 the presence of CTCs was shown to be associated with response to chemotherapy and prognosis. Patients who were CTC-positive at baseline or who converted to CTC-positive during treatment, presented with lower rates of disease control, as well as worse OS and PFS. Irrespective of their correlation with survival rates, CTCs were also associated with lymph node metastasis.134
The analysis of CTC number at baseline and at different time points in the course of SCLC was referred to for the prediction and monitoring of the response to chemotherapy.135 Circulating tumor cells obtained from plasma samples may be also used for the detection of specific mutations related to lung cancer, such as EGFR136 and KRAS,137 with a higher sensitivity than ctDNA. Moreover, specific gene rearrangements can be detected in CTCs with promising acurracy. In patients with lung adenocarcinoma, anaplastic large-cell lymphoma (ALK) gene rearrangement and ALK protein expression in CTCs were concordant with findings from tumor tissue,138 which has been confirmed by other researchers.139, 140 A rearrangement of repressor of silencing 1 (ROS1) is another example of chromosomal aberrations detectable in CTCs, with biopsy-confirmed gene fusion in NSCLC patients.141 Dynamic changes in the number of CTCs with aberrant ALK-fluorescence in situ hybridization patterns, such as ALK copy number gain, might serve as predictive markers of the response to treatment, as these aberrations are considered to be one of the mechanisms underlying acquired resistance to crizotinib (an ALK and ROS1 inhibitor). A decrease in CTCs with ALK copy number gain during treatment with crizotinib was linked to a longer PFS.142
Apart from the lack of standardized methods of analysis, CTCs appear to have other limitations similar to ctDNA evaluation. These include low detection rate, especially in patients with an early stage of the disease, and an unclear influence of tumor heterogeneity and its localization on liquid biopsy findings. Table 2 and Table 3 summarize the results of studies concerning ctDNA and CTCs in lung cancer, respectively.
MicroRNA as a lung cancer marker
MicroRNAs (miRNAs) are noncoding small molecules, comprising approx. 21 nucleotides. They are considered post-transcriptional regulators of gene expression. They achieve this by binding to the 3′-UTR of target messenger RNA, which results in repressing translation or promoting messenger RNA deadenylation and degradation.143 Due to their biological role, miRNAs are thought to be important in cancer initiation and progression, as they can influence both oncogenes and tumor suppressor genes.144 Furthermore, a potential role of miRNA in the diagnostics and treatment of lung cancer has been recently highlighted.145 Considering the availability of miRNA expression in cancer tissue and bodily fluids, especially in serum, it can be easily measured using liquid biopsy.145 A growing popularity of miRNA research has led to the development of diagnostic panels which may be used complementarily in the early detection of malignant lung lesions and are constantly being improved.146, 147
A signature panel of 15 miRNAs was able to differentiate between patients with lung cancer and those with non-tumor lung disease, other systemic diseases, and healthy controls, with a sensitivity of 82.8% and a specificity of 93.5%.148 Other authors used a panel of 2 miRNAs (miRs-31-5p and 210-3p) detected in sputum and 1 miRNA (miR-21-5p) from plasma, that reached sensitivity and specificity in the detection of lung cancer of 85.5% and 91.7%, respectively.149 Even greater sensitivity and specificity (99% for both) was achieved by a combination of miR-1268b and miR-6075, and was validated in a group of over 3000 participants and maintained its performance regardless of TNM stage or histological type of tumor.150 A number of specific miRNAs have also proven to be valuable in the early diagnosis of NSCLC,151 distinguishing NSCLC from SCLC152 and specific types of NSCLC.153
Numerous miRNAs are efficient in the prognosis of disease progression and resistance to treatment. The analysis of miRNA expression in advanced NSCLC cases revealed 17 miRNAs significantly associated with 2-year survival rate.154 At the same time, the downregulation of miR-590-5p was linked to lower median survival rates in a cohort of NSCLC patients,155 while the upregulation of miR-25 was higher in NSCLC patients compared to the control group, but also correlated negatively with OS and relapse-free survival.156 In a separate analysis, patients with adenocarcinoma and SCC with high expression of miR-25-3p had shorter OS, regardless of tumor histology.157 A meta-analysis on the prognostic value of the downregulation of miR-126 highlighted its relationship to unfavorable outcomes of NSCLC.158 Others reported an association between miR-153,159 miR-494,160 miR-519d161 and more advanced clinical stage, presence of lymph node metastases, and worse OS in NSCLC patients. Similar results regarding a poor prognosis in NSCLC patients were reported for the downregulation of miR-184,162 miR-185,163 miR-770,164 and miR-30a-5p,165, and the upregulation of miR-23b-3p, miR-10b-5p and miR-21-5p,166 miR-31,167 miR-378,168 miR-942, and miR-601.169 On the contrary, a high expression of miR-3195 resulted in longer OS,170 while miR-21 and miR-4257 were established as predictors of NSCLC recurrence.171 In patients with SCLC, the upregulation of miR-92b and miR-375 was related to chemotherapy resistance and shorter PFS.172 At the same time, miR-422a and miR-135a showed a strong association with metastases to lymph nodes in lung cancer patients,173, 174 lower expression of miR-139-5p was found in NSCLC patients with bone metastases,175 and miR-375-3p was also proposed to be a possible biomarker of SCLC metastatic activity.176
Expression profiles of miRNA may also serve as markers for treatment response,170 with higher expression of miR-1249-3p observed in individuals who responded well to chemotherapy. Changes in serum levels of various miRNA panels have been used to predict worse sensitivity to chemotherapy.177, 178 Additionally, in a cohort of early-stage NSCLC patients, the expression of miR-216b was significantly increased after a successful tumor resection.179 Profiling of miRNA may also be indicative of the response to radiotherapy180, 181 or immunotherapy, with patients who significantly overexpressed miR-320b-d before the treatment with PD-1/PD-L1 inhibitors not responding well to the therapy. In the same group, a decrease in miR-125b-5p was observed in those who presented with only a partial response.182 In NSCLC patients, miR-504 expression differed significantly depending on EGFR mutation status.183 An experimental miRNA panel was also tested for discrimination between ALK-positive and ALK-negative lung cancers,184 which is relevant to immunotherapy treatment options. The miRNAs were also proposed as markers of resistance to EGFR-TKI therapy,185 as shorter OS was reported in patients with high serum levels of miR-30b and miR-30c treated with erlotinib.186 Furthermore, miR-30c expression patterns showed utility in predicting cardiotoxicity in patients treated with bevacizumab,187 which indicates the potential of this method for stratifying the risk of adverse events for particular therapies. Further attempts to improve diagnostic and prognostic accuracy of miRNA in lung cancer patients include combining this method with other commonly used biomarkers, such as CEA and CYFRA21-1.188, 189
Although miRNA profiling has gained much interest in recent years, its application in clinical practice still has some limitations. Methodological discrepancies within study design and technological details of tools applied can be seen throughout the studies on miRNA in lung cancer, which prevents consistent conclusions.190 Another issue to be addressed is a lack of specificity of candidate miRNAs, as there is a large number of these being examined in various types of cancer, and these miRNAs are involved in the regulation of multiple biological pathways. Furthermore, miRNA expression can be affected by disease stage and the treatment used,191, 192 which has to be considered in the clinical interpretation of research findings. Therefore, there is a need for further studies on representative groups of patients with the use of consistent methodology, in order to ensure reproducibility and generalizability of results. Studies investigating miRNA in lung cancer are presented in Table 4.
Markers of nervous system involvement in the course
of lung cancer
Markers of brain metastases
Calcium binding protein B (S100B), synthesized in astrocytic terminal processes,193 is an established marker of blood–brain barrier disruption. The detection of S100B protein along with anti-S100B autoantibody allows to distinguish between lung cancer patients with or without brain metastases, with a sensitivity of 89% and a specificity of 58%.194 Furthermore, the evidence of an association between serum S100B level and brain metastases with subsequently worse prognosis has been shown by other researchers.195 However, low specificity for S100B is a main barrier to its implementation, as its abnormal expression has also been reported in patients with cerebrovascular disease. Nonetheless, pro-apolipoprotein A-1 levels, measured using proteomic techniques, appear to be a more specific marker as it was increased in lung cancer patients with brain metastases, regardless of cerebrovascular disease.196
Among other potential markers, high NLR, platelet-to-lymphocyte ratio and C-reactive protein (CRP) levels were suggested to indicate the development of brain metastases in lung cancer patients after definitive radiotherapy or radiotherapy combined with chemotherapy.197 High NLR (>4.95) and lower mean platelet volume were associated with an increased risk of brain metastases in patients with NSCLC.198, 199 At the same time, high plasma fibrinogen concentration and platelet count correlated with shorter OS in NSCLC patients already diagnosed with brain metastases.200
The miRNA has also emerged as a relevant marker of brain metastases in lung cancer, with the overexpression of miR-330-3p noted in the serum of NSCLC patients with brain metastases, when compared to those without dissemination to the central nervous system.201 Significantly lower expression of miR-330 was found in lung cancer patients who had undergone whole-brain radiation therapy, and these patients proved to be radiation-sensitive.202 Furthermore, serum levels of miR-21 before and after radiotherapy in lung cancer patients with brain metastases were significantly correlated with OS.203 The expression of miR-483-5p and miR-342-5p in serum and cerebrospinal fluid differed between patients with leptomeningeal and brain parenchymal metastases.204
The detection of ctDNA in the cerebrospinal fluid of patients with lung cancer brain metastases displayed higher mutation detection rates than peripheral blood samples.205 Sensitivity of detecting EGFR mutations in ctDNA from plasma or cerebrospinal fluid was comparable, while T790M mutations were more prevalent in plasma samples.206 With regard to the location of metastases, EGFR, KRAS, BRAF, or ERBB2 mutations in plasma ctDNA could be detected in half of the patients with isolated brain metastases,207, 208 and ctDNA positivity was associated with a higher risk of extra central nervous system dissemination.
Markers of paraneoplastic neurological syndromes
Specific autoantibodies related to the development of PNS can be regarded as markers of the involvement of nervous system in the course of lung cancer.14 Onconeural antibodies, directed against intracellular antigens, are already well recognized and commonly used in practice. Most prevalent among these antibodies are anti-Hu, anti-Yo, anti-Ri, anti-CV2, anti-Tr, anti-amphiphysin, and anti-Ma/Ta,8 with anti-Hu, anti-Ri, anti-CV2, and anti-amphiphysin being closely linked to lung cancer.209 Recent years have also seen advances in establishing the role of antibodies against cell-surface or synaptic antigens, such as antibodies against N-methyl-D-aspartate receptors, the α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor, the γ-aminobutyric acid receptor-B, leucine-rich glioma-inactivated protein 1, and contactin-associated protein-like 2.210
Several other autoreactive antibodies and related antigens were investigated for their use in the detection of PNS in lung cancer patients, including phosphodiesterase 10A and Purkinje cell cytoplasmic antibody type 2 antibodies.211, 212 Apart from early detection of lung cancer, some of these antibodies were investigated in terms of their prognostic value. Indeed, some reports have indicated that the presence of anti-Hu antibodies in patients with SCLC was associated with limited stage of the disease and satisfactory response to therapy213 or longer median survival,214 although this correlation was not clear.215 Despite their clinical utility, onconeural and cell-surface antibodies have limited sensitivity and specificity, they are not detectable in every patient with PNS,216 and may be associated with different kinds of neoplasms.217 Thus, further exploration in this field seems warranted.
Conclusions
In this review, we outlined the recent development in the research on of potential biochemical, immunological and molecular markers of lung cancer that have shown promising sensitivity and/or specificity. Molecular markers are associated with improved understanding of complex tumor genetics, while immunological markers have provided a more thorough insight into the tumor-related immune environment, thus opening new perspectives in diagnosis and effective management of lung cancer. Diagnostic markers that enable an early detection of the disease may be valuable in supporting existing screening methods, while markers with predictive potential could contribute to the identification of patients at high risk of recurrence or with the propensity for metastases, with scope for the development of individualized monitoring and treatment strategies. With the advent of new treatment options such as immune checkpoint and kinase inhibitors, the recognition of mechanisms of resistance to targeted therapy and emerging decisions about personalized treatment strategies appear as key elements of lung cancer management, with the use of relevant biomarkers being indispensable. Further investigation should be aimed at improving the accuracy and specificity of these markers, perhaps by combining them into panels.
Circulating biomarkers seem particularly promising for practical use as their detection is non-invasive, safe and easily repeatable. However, the majority of molecules investigated still need thorough validation, with standardization of techniques and assays used, and established cutoff values. Furthermore, a systematic comparative analysis of efficacy should be performed for findings from liquid biopsy and from tumor tissue studies. Finally, the implementation of markers into clinical practice needs more supportive evidence, preferably from clinical trials.