Introduction

Malignant pleural mesothelioma (MPM) is a tumor strictly connected with occupational and/or environmental exposure to asbestos [1]. Epithelioid subtype occurs more frequently (60%), whereas approximately 10% of MPMs are sarcomatoid and the remainder biphasic. Prognosis remains dismal as it commonly diagnosed in advanced stage and response to available medical treatments is poor. Moreover, paradigm shift of precision medicine has not yet translated into new systemic approaches [2].

Regarding mechanisms of tumorigenesis, chronic inflammation has been linked to the initiation and progression of MPM. Inhalation of asbestos fibers results in a response characterized by recruitment of macrophages and neutrophils, production of pro-inflammatory cytokines, cell proliferation and angiogenesis [3]. As result, MPM is characterized by a low mutational burden and a genomic landscape dominated by inactivation of several tumor suppressors genes, such as cyclin-dependent kinase inhibitor 2A (CDKN2A), BRCA1 associated protein (BAP1) and neurofibromatosis 2 (NF2) [4].

So far, radiological screenings have not proven to be effective for detecting early MPM among asbestos-exposed subjects [5]. Also differential diagnosis of MPM is sometimes very difficult. A wide tissue immunohistochemical staining, including mesothelioma-associated markers as Wilms tumor-1 (WT-1), calretinin, cytokeratin5/6 (CK5/6) and podoplanin is required for diagnosis and differentiation from pulmonary carcinoma or metastasis from other tumors [6, 7]. Since sarcomatoid histotype often does not express these markers and most commonly only stains for pan-Cytokeratin, GATA binding protein 3 (GATA-3) has been proposed as sensitive and specific biomarker for distinguishing sarcomatoid mesothelioma from sarcomatoid carcinoma of the lung [8]. Finally, BAP1 immunostaining and CDKN2A/p16 fluorescence in-situ hybridization (FISH) in effusion cytology specimens are useful in distinguishing malignant from benign mesothelial proliferation, although BAP1 and p16 are not deleted or lost in all MPM tumors [9]. In particular, BAP1 is the most commonly mutated gene in MPM and loss of nuclear staining occurs in more than 60% of tumors [10], while homozygous deletion of CDKN2A is seen in 70% of epithelial MPM [11, 12].

In this scenario, use of biological fluids, such as blood and pleural effusion, represents an ideal source of non-invasive biomarkers. Tumor cells can release into bloodstream different molecules (proteins and nucleic acids, also epigenetically modulated); circulating tumor cells (CTCs) and extracellular vesicles (EV) can be detected too (Figure 1).

Display full size

Figure 1. 

Circulating biomarkers analysis in malignant pleural mesothelioma

So, the search for biomarkers, useful for early detection, prognosis and in the prediction of response to treatments appears as a promising field of investigation [13]. This review provides an overview on current understanding and possible developments of liquid biopsy in MPM, mostly focused on its utility as diagnostic and prognostic test.

Proteins

A number of proteins have been evaluated in MPM over this time as diagnostic or screening markers. At the moment, mesothelin remains the most extensively studied and the only blood-based test approved by Food and Drug Administration (FDA) for detecting MPM recurrence in patients following surgery or assess response to therapies. Further candidate protein biomarkers include osteopontin (OPN), fibulin-3 (FBLN3), high-mobility group box 1 (HMGB1) protein, angiogenic factors, syndecan-1 (CD138), whereas proteomics represents a promising new approach (Table 1).

MicroRNAs (miRNAs)

miRNAs are non-coding RNA, 20–25 nucleotides long, that regulate gene expression at post-transcriptional level by binding the 3′-untranslated regions of their messenger RNA (mRNA) targets and inhibiting translation or inducing cleavage [69].

As expected, miRNAs are involved in main human diseases, ranging from cardiovascular disorders to cancer. In tumorigenesis miRNAs can act as either tumour suppressors or oncogenes, as several targets correspond to a single miRNA. Genome-wide profiling demonstrates that miRNA expression signatures are associated with tumor type and clinical outcomes. So, miRNAs could be potential candidates for diagnostic and prognostic biomarkers, or therapeutic targets tools [70].

miRNAs are secreted into the liquid fraction of peripheral blood or other body fluids inside exosomes or in vesicle-free forms, associated with high density lipoproteins (HDL) or Argonaute proteins [71]. Therefore, these molecules can be extracted and profiled.

During the discovery phase high throughput techniques, as microarray hybridization, enable the identification of a wide number of miRNAs. Quantitative techniques, as real-time reverse transcription quantitative polymerase chain reaction (RT-qPCR) or next generation sequencing (NGS) are used during the following validation phase [72]. The choice of an appropriate normalization method, able to reduce analytical variability, represents another important aspect in miRNA quantification for clinical implementation [73].

Regarding MPM patients, many studies investigated circulating miRNA profile with the aim of identifying markers for early detection, differential diagnosis and prognosis [74] (Table 2).

In 2009 Guled et al. [75], were among the first to compare miRNA profile in MPM cells and normal human pericardium using microarrays. They identified 12 over-expressed (let-7b, miR-30b, miR-32, miR-195, miR-345, miR-483-3p, miR-584, miR-595, miR-615-3p, and miR-1228) and 9 down-regulated miRNAs (let-7e, miR-7-1, miR-9, miR-34a, miR-144, miR-203, miR-340 and miR-423, miR-582) in neoplastic cells. Among those over-expressed, miR-30b, miR-32, miR-483-3p, miR-584, and miR-885-3p regulate the tumor suppressor genes CDKN2A and NF2, whereas downregulated miRNAs such as miR-9, miR-7-1 and miR-203 have as targets the oncogenes hepatocyte growth factor (HGF), platelet derived growth factor subunit A (PDGFA), epidermal growth factor (EGF) and Jun oncogene (JUN). Of interest, the majority of these miRNAs are located in chromosomal regions already known as altered in MPM, such as 8q24, 1p36, and 14q32 [76].

As reported below, among those studied in MPM patients, more evidence emerged about the role as possible circulating biomarkers of miR-126-3p, miR-103a-3p and miR-625.

In 2011 Santarelli et al. [77] obtained a miRNA signature by analyzing biopsies collected at diagnosis from MPM patients, compared to healthy human mesothelium. Among eight significantly downregulated miRNAs, miR-126 acts as tumor suppressor by decreasing the translation of vascular endothelial growth factor-A (VEGF-A) mRNA. Evaluation of serum samples obtained from 44 MPM patients, 196 asbestos-exposed and 55 HS confirmed higher concentrations of SMRP and VEGF in MPM patients. On the other hand, low levels of MiR-126 correlated with high concentration of VEGF and discriminated high risk individuals from HS (sensitivity 60%, specificity 74%) and MPM patients (sensitivity 73%, specificity 74%) [77].

In another contribution the same group demonstrated that a “3-biomarker classifier”, based on serum levels of SMRP, miR-126 and methylated thrombomodulin promoter (Met-TM), improved the differential diagnosis of MPM from asbestos-exposed subjects and healthy controls, when compared to SMRP alone [78].

Under expression of miR-126 in serum was confirmed in the study of Tomasetti et al. [79], capable to discriminate MPM patients from those with NSCLC or HS, using relative qRT-PCR. Conversely, miR-126 levels did not differentiate NSCLC patients from controls. In this study, results were normalized to small nuclear RNA U6 as endogenous control and the cel-miR-39 as exogenous control. The low expression of miR-126 was also associated with a worse prognosis, corroborating the clinical meaning of this biomarker. However, dysregulation of U6 levels in different conditions limits its potential use for normalization of circulating miRNA analysis. For instance, U6 levels were significantly upregulated in sera of patients with chronic inflammatory diseases [80].

Weber et al. [81], showed different expression levels of circulating miR-132 between mesothelioma patients and asbestos-exposed controls. For discrimination, sensitivity of 86% and specificity of 61% were calculated. Combining miR-132 with the previously described miR-126, sensitivity of 77% and specificity of 86% were obtained.

More recently, Bononi et al. [82], found three miRNAs (miR-197-3p, miR-32-3p and miR-1281) to be upregulated in MPM patients compared to HS and workers ex-exposed to asbestos (WEA), and the latter also upregulated in WEA compared to HS. Instead, the down regulation of miR-126 in serum was not confirmed. In this investigation MiR-3665, an endogenous stable microRNA (MiR-3665), was also proposed as suitable normalizer. Forkhead box O3 (FOXO3), a key gene promoting apoptosis, is reported as a target of miR-197. It has also been published that miR-32-3p acts as oncogene down-regulating phosphatase and tensin homolog (PTEN) and targeted B-cell translocation gene 2 (BTG2) [83].

Kirschner et al. [84], firstly reported elevated levels of a circulating miRNA with potential value as biomarker, describing, through a microarray analysis, 15 miRNAs upregulated in MPM patients compared with controls. In fact, through qRT-PCR, they demonstrated higher levels of miR-625-3p in the serum of MPM patients (n = 30) compared to AES (n = 10, sensitivity 70%, specificity 90%). However, these data were normalized against miR-16, which is known to be altered in MPM and influenced by sample haemolysis [85]. In a study aimed to assess the effects of miR-625-3p on thyroid cancer cells it appears to promote proliferation, migration and invasion by enhancing the expression of astrocyte elevated gene-1 (AEG-1) and activating downstream Wnt/β-catenin and Janus kinase (JNK) pathways [86]. Finally, diagnostic approaches different from those described above were proposed in other works.

In 2017 Cavalleri et al. [87], using an OpenArray method, identified a new miRNA signature in exosomes isolated from plasma. More specifically, a combination of miR-103a-3p and miR-30e-3p discriminated MPM and WEA (sensitivity 95.8%, specificity 80%).

In two consecutive works Weber et al. [88], quantified miRNAs in cellular fraction of peripheral blood. In the first study, downregulated miR-103 showed sensitivity of 83% with specificity of 71%, and sensitivity of 78% with specificity of 76% for discrimination of MPM from AES and the general population, respectively. In the second investigation, to enhance the performance of this biomarker the authors suggested a combination of miR-103a-3p and mesothelin to differentiate MPM and WEA (sensitivity 95%, specificity 81%) [89].

Circulating tumor DNA (ctDNA)

Circulating free DNA (cfDNA) is released from healthy and cancerous tissues undergoing apoptosis or necrosis. ctDNA originates from tumor cells and carries somatic mutations, representing only a small fraction of cfDNA.

Notably, cfDNA demonstrates a wide potential as biomarker in oncology for multiple indications including staging and prognosis, monitoring response to treatments or minimal residual disease and identification of acquired resistance mechanisms to drugs [90].

The degree of cfDNA integrity, called integrity index, is based on the ratio between long and short cfDNA fragments and has been investigated as biomarker. Sriram et al. [91], found that pleural fluid DNA integrity index was higher in MPM (n = 52) than in benign effusions (n = 23). In particular, an high pleural fluid DNA integrity index exhibited a PPV of 81% in detecting MPM in patients with cytology-negative pleural effusion, suggesting its usefulness to guide more invasive procedures in this population (Table 3).

New perspectives have been opened afterwards the demonstration of the presence of ctDNA in treatment naïve MPM patients. Hylebos et al. [92], performed whole exome sequencing (WES) on paired germline and tumor DNA of ten MPM patients to identify cancer-specific variants of interest. Hence, using droplet digital PCR (ddPCR), they confirmed the presence of selected mutations in serum samples of three out of five treatment naive patients (detection rate 60%) at allelic fractions ranging from 0.28 to 0.9%. On the other hand, the absence of detectable tumor specific alterations in cfDNA of patients who received chemotherapy is interesting in view of the potential use of ctDNA as a biomarker for treatment response assessment.

However, broad implementation of ctDNA analysis in daily clinical practice requires validation of these data as well as use of accessible and cost-effectiveness technique.

CTCs

CTCs are cells shed by solid tumors and circulating in the bloodstream. As CTCs play a key role in development of distant metastases, they could provide useful information in a variety of malignancies. However, their identification requires highly sensitive techniques, as they are present in a very small numbers [93].

Among various systems developed, CellSearch is the only FDA-approved for clinical purpose. It detects cells expressing epithelial cell adhesion molecule (EpCAM) through magnetic particles coated with specific antibodies [94]. As important limitation, CTCs without EpCAM expression, such as those exhibiting EMT or of non-epithelial origin, may not be recognized [95]. Accordingly, CellSearch presents a low sensibility for identification of CTCs for MPM.

To overcome this issue a CTC-chip system coated with an anti-podoplanin antibody has been developed [96]. In peripheral blood drawn from 15 MPM patients, the CTC-chip showed better performance to detect CTCs in all pathological subtypes over CellSearch (68.8% vs. 6.3%; P < 0.001), and the capability to predict unresectable disease (area under the ROC curve, 0.851). Setting a cut-off value of 2 cells/mL, an higher CTC count was also significantly associated with worse prognosis, and count decrease was demonstrated after surgical resection or tumor shrinkage [97] (Table 3).

Epigenetic biomarkers

Epigenetic modifications occurring during tumor evolution can be detected in fluids and serve as potential biomarkers [98]. In asbestos-induced carcinogenesis, reactive oxygen species (ROS) formation is responsible of gene promoter methylation, through poly(ADP-ribose) polymerase 1 (PARP1) and DNA (cytosine-5) methyltransferase 1 (DNMT1) recruitment [99]. Interestingly, epigenetic biomarkers could represent novel therapy targets.

Nocchi L et al. [100], combined two epigenetically regulated markers (miR-126 and TM) with SMPR. In fact, it has been reported that, in MPM tissue, TM gene expression can be silenced through epigenetic mechanisms and, similarly, hypermethylation of miR-126 promoter region is responsible for its downregulation [101]. Despite the poor sensitivity (60%), the authors reported that circulating methylated TM DNA significantly discriminated MPM patients from controls with high specificity (82%), thus complementing the performance of miR-126 and SMRP alone for MPM detection [78] (Table 3).

More recently, Guarrera et al. [102], using a genome-wide methylation array, found differential methylation patterns of selected CpGs in DNA extracted from white blood cells (WBC) between MPM patients (n = 163) and controls (n = 137). The most hypo-methylated single-CpG is located in a gene coding for a transcription factor, namely Forkhead-box K1 (FOXK1), which directly interacts with BAP1. Significantly, FOXK1 maps to chromosome 7p22.2, one of the genomic areas described as associated to MPM [103], whereas somatic and germline mutation of BAP1 is reported in MPM [10, 104].

Exosomes

Exosomes are small nano-EV derived from the endosomal pathway and released from normal and cancer cells. Their presence in body fluids makes them potential novel biomarkers for early diagnosis and prognostication, as well as therapeutic targets [105].

The protein composition of in vitro produced exosomes has been studied using MS, demonstrating that exosomes derived from MPM tumour cells contain select miRNAs or cargo proteins known to be associated with angiogenesis, cell migration, metastasis, and immunoregulation [106]. Exosomes present in pleural fluid may have various cellular origins, as highlighted through proteomic analysis of malignant effusion [107].

A current limitation to adopt them as biomarkers is represented from the lack of standardization of techniques required for isolation (based on differential centrifugation) and characterization (including flow cytometry, proteomic and MS analysis).

Immunophenotype of circulating cells

A great advantage of multiplex assay for flow cytometry is ability of measuring a large number of analytes at the same time reducing time of analysis, costs and sample volume [108].

A comprehensive study of the immune signature of MPM may contribute to understanding the immune-evasion mechanisms related to tumor growth, concurring to differential diagnosis and prognostic stratification.

Using multiparametric flow cytometry, Salaroglio et al. [109], analyzed simultaneously immune infiltrate of either pleural fluid and biopsy specimens of pleural tissue from 275 patients with pleural effusion of unknown cause. So, they described an immune signature discriminating MPM from pleuritis and from pleural metastases, identifying higher CD8 T cells and Treg cells within pleural fluid as specific biomarkers of MPM. The immune phenotype of pleural fluid cells had no prognostic significance, whereas expression of programmed death 1 (PD-1), lymphocyte activating 3 (LAG-3), and T cell immunoglobulin and mucin domain 3 (TIM-3) in tumor-infiltrating lymphocytes (TILs), but not in T cells in pleural effusion, correlated with lower OS [109].

Discussion

MPM is a challenging malignancy to treat as early diagnosis is difficult and effective treatments are lacking. Recently, The Cancer Genome Atlas (TCGA) Research Network conducted a comprehensive integrated genomic study of MPM. A novel subtype characterized by extensive loss of heterozygosity, expression of the immune checkpoint regulator V-domain Ig suppressor of T cell activation (VISTA) and a comprehensive analysis of BAP1 mutations emerged as key findings [12].

Disappointingly, the deeper knowledge of MPM biology has not yet translated into the validation of non-invasive biomarkers that can assist in diagnosis (identifying patients with all subtypes of MPM and predicting its development in asbestos-exposed subjects), definition of prognosis and monitor treatments response. The search for circulating biomarkers has been ongoing from 1990s. As several candidates have been investigated only mesothelin obtained FDA approval, but its poor sensitivity limits the utility as diagnostic marker. OPN is a marker of the duration of asbestos exposure and has a potential prognostic role but lacks specificity for MPM. Other studies have shown promising results for HMGB1 (and its hyperacetylated form) and CD138 but their diagnostic role require further validation. A variety of deregulated miRNAs, as well as different new approaches based on proteomics technology, have been proposed as suitable biomarkers and are currently under investigation. Despite technical challenges, also exosomes provide a novel means of uncovering biomarkers in MPM.

As clearly emerges from this work, and from reviews published elsewhere [110–113], none of single biomarkers has ever reached adequate accuracy as to ensure its use in clinical practice. Since MPM is an extremely heterogeneous tumor, also combination of different markers together could contribute to improve diagnostic accuracy. On the other hand, clinical utility of biomarkers is dependent on how they modify subsequent clinical decisions and influence outcomes. At moment, no evidence suggests that early detection of MPM can improve survival and none of these biomarkers is recommended for screening.

More of biomarkers were studied retrospectively in small cohorts of patients, whereas analytical and pre-analytical heterogeneity affect strength of conclusions. Improvement of diagnostic accuracy, standardization of the different assays for reproducibility, choice of appropriate cut-off for each biomarker represent the main critical issues to ensure translation to daily clinical practice. Broad application of a next generation of high-performance technologies based on machine learning algorithms is expected to have a significant impact on the implementation of liquid biopsy [114]. Therefore, integration of multidisciplinary skills and increased international cooperation appears essential for improving validity of results and ensure translation to clinical practice.