MPM diagnosis generally occurs when tumour is only marginally responsive to therapies and pleuroectomy is not always feasible [3], the identification of sensitive and non –invasive detection tools, allowing an early therapeutic intervention, might improve therapy options and overall survival [33–35].
In the present work, we evaluated if EV-derived miRNAs could serve as reliable tumour biomarkers for early detection of MPM and to gain insight into the molecular mechanisms linked to the late carcinogenic process. By using a NGS approach, we profiled miRNAs expression enclosed in serum EVs of preclinical MPM individuals and cancer-free controls.
We observed a significant up-regulation of miR-11400, miR-148a-3p and miR-409-3p in serum EVs of 20 pre-diagnostic MPM samples collected up to five years before the clinical manifestation of the disease, compared with cancer-free individuals. In the validation phase, miR-11400 and miR-409-3p were found up-regulated, while a slight increase of miR-148a-3p expression was observed in preclinical cases.
So far, numerous cell-free miRNAs have been identified as non-invasive biomarkers for MPM, however few of them have been validated in large cohorts [14, 15, 17, 20, 36], and besides which, biomarker candidates need to be validated in longitudinal studies.
Santarelli et al. addressed their research in studying effective markers for the early detection of MPM [16]. The authors analysing miRNAs profiles in fresh MPM tissue compared with corresponding normal mesothelium, identified miR-126 as promising markers that may be potentially used to distinguish cancerous from normal tissue. Weber and colleagues analysed miR-126-3p with two additionally promising early detection biomarkers (miR-132-3p and miR-103a-3p) in 17 pre-diagnostic plasma samples collected about 8.9 months prior to clinical diagnosis, compared with 34 asbestos exposed controls, and concluded that this signature of miRNAs was not useful for the early detection of MPM. Their qPCR analysis revealed 0% sensitivity and a specificity of 98%. Likely, in pre-clinical cases the expression of these markers is not declined yet despite the promising results were obtained in cross-sectional studies from manifest MPM [37]. Indeed, we did not find any difference in expression levels for the three described miRNAs (Supplementary Table 3).
On the other hand, in EPIC preclinical MPM cases, a different three-miRNAs panel identified by us could predict MPM onset within five years, yielding an AUC of 0.81 with 75% sensitivity and 70% specificity. In the absence of a prospective validation cohort, we could examine the diagnostic performance of the model in a retrospective study, which confirmed the accuracy and reliability of the three-miRNA signature in distinguishing MPM cases from cancer-free controls even at the time of diagnosis (AUC = 0.86). Although the specificity was even higher than the prospective cohort at 95%, sensitivity dropped to 53% in the retrospective validation study, meaning almost half of the diagnosed cases tested as negative. This suggests that the present miRNA signature may weakens in a number of subjects overt time after diagnosis and warrants further investigation. However, it is unlikely to affect undiagnosed cases at earlier stages of MPM and the specificity/sensitivity balance can be adjusted to fine tune the score threshold.
The predictive role of miR-148a-3p has been described in previous cancer studies [38–40]. Increasing expression levels of miR-148a-3p have been described as a biomarker for early diagnosis of laryngeal squamous cell carcinoma, and high circulating levels of this miRNA were associated with poor overall and disease-free survival of hepatocellular carcinoma patients after liver transplantation [38, 39].
Josson et al. demonstrated that high expression of miR-409 delivered by EVs from prostate stromal fibroblasts promoted tumorigenesis through repression of tumour suppressor genes [41].
Limited data are available in the literature for miR-11400 [42]. The predicted target genes are highly enriched for key regulators of the lysosome, Hippo signalling and TGF-β pathways.
The long-term toxicity of asbestos fibres is due to their bio-persistence in cells and their accumulation in lysosomes, causing autophagy as well as lysosomal dysfunctions that result in oxidative stress and inflammation [43]. As it has been demonstrated for nanoparticles, also asbestos fibres could directly affect lysosomal stability by inducing lysosomal oxidative stress, alkalization, osmotic swelling, resulting in lysosomal membrane permeabilization and the subsequent activation of NLRP3 inflammasome [44]. Thereby, lysosomal dysfunction provides a mechanism for asbestos-mediated inflammation.
Significant enrichment for Hippo and TGF-β signalling has also been reported for miR-148a-3p-validated target genes. Previous studies have demonstrated a functional interaction between TGF-β and Hippo signalling pathways in MPM cell growth and proliferation [45]. In particular, the inactivation of Hippo signalling components can play a remarkable role in the development and progression of MPM [46]. The activated LATS kinase, in association with the tumour-suppressor MOB, predicted target of miR-11400, phosphorylates and inhibits the transcription coactivators TAZ and/or YAP [47]. Therefore, dysregulation of LATS1/2 has been shown to result in constitutive activation of the YAP1/TAZ transcriptional coactivators, conferring malignant phenotypes to mesothelial cells.
Amongst predicted miR-11400-target genes, the tumour suppressor RASSF1 and the G1-phase cell-cycle regulator CDKN2B, are both frequently lost in MPM tissue and implicated in the mechanisms of carcinogenesis [48–50]. Although experimental evidence is needed to validate the functional relationship between miR-11400 and its target genes, these findings suggest its role as a mediator of inflammation and malignant transformation.
Conversely, further evidence is available for miR-148a-3p. According to pathway enrichment analysis, validated miR-148a-3p target genes were enriched in several MPM-related pathways. We obtained a significant enrichment for FOXO signalling pathway, given by downstream genes mediating the cell cycle and apoptosis. Suzuki and colleagues demonstrated that the up-regulation of miR-182 and miR-183 suppresses the expression of FOXO1, which in turn promotes cell proliferation by acting on its downstream targets CDKN1A, and CDKN1B [51]. Likewise, miR-148a-3p targeted and suppressed cell cycle and apoptotic regulators (BCL2L11, CDKN1A, CDKN1B, S1PR1) likely leading to uncontrolled proliferation [52].
It is well demonstrated that EVs represent an ideal tool used by cells of primary neoplastic lesions to shape local and distant microenvironments promoting optimal conditions for tumour growth and proliferation [53]. Greening and colleagues, showed that MPM-derived exosomes are enriched with selective oncogenic molecular cargo and may contribute to the progressive changes of MPM by tumour angiogenesis and malignancy [54].
The levels of serum EV miRNA might reflect the profiles in cancer tissue. We tested this hypothesis using dbDEMC, representing a database for collection of differentially expressed miRNAs in human cancers obtained from microarray data, among which MPM. We reported that miR-148a-3p and miR-409-3p were up-regulated in MPM tissues in agreement with the results in serum EVs, supporting their potential mechanisms in tumour-related biological processes and potentially reflecting the MPM progression [31, 32]. The expression changes in miR-11400 levels were not detected since the authors used a microarray not including miR-11400 among the probes.
Our study presents several limitations. First, the small number of prospectively collected cases limits the robustness of our results. In addition, EPIC did not include multiple sampling during follow-up; as a result, a serum sample preceding diagnosis no more than five years (i.e. presumably during MPM pre-clinical phase) was available for only part of all EPIC cases, entailing further limitation of the study dimension. MPM is a rare malignancy and obtaining pre-clinical samples is challenging. Settings where members of high-risk groups might be invited to participate to studies of pre-clinical MPM markers are medical surveillance programs for former asbestos workers; no such program, however, is active in Italy. A validation prospective cohort was not, thus, available to us; we tested, therefore, the identified miRNAs panel in a case-control setting, which had the disadvantage of evaluating sensitivity and specificity at the time of MPM diagnosis, rather than during its pre-clinical phase.
More importantly, it might be questioned whether research into pre-clinical MPM markers is justified on clinical as well as public health grounds, given the current limitations of the options that might be offered to suspected pre-clinical MPM cases: the diagnostic work-up is invasive and associated with non-negligible risks, while treatment offers only limited hopes of disease eradication. However, both diagnostic approaches and treatment might evolve if a shift might be obtained towards a larger proportion of MPM cases in potentially resectable disease stages. It has to be mentioned, also, the fact that the same markers may prove useful as ancillary diagnostic tools for a non-negligible proportion of cases, such as those for whom thoracoscopy is contraindicated or whose microscopic samples turn out to be difficult to interpret.
Besides the potential role of miRNAs, it would be worthy to identify and validate other biomarkers of different molecular sources (e.g. proteins, DNA methylation) to improve the diagnostic performance of MPM detection. As in other tumours, it is well known that the use of circulating biomarkers in pre-diagnostic samples may be a meaningful approach for detection at early stages [55, 56]. So far, the combination of calretinin and mesothelin has been showed to have the better diagnosis performance in plasma of preclinical asbestos exposed individuals, up to about one year before diagnosis [57]. However, several confounding factors may influence their plasma concentration and the study of additional markers that can decrease the false positive tests is warranted [4, 58].