Globally, it has been estimated that asbestos is still the leading cause of morbidity, disability, and mortality for occupational cancer accounting for 4120 (3060–5240) thousand DALYs (Disability-Adjusted Life Years) and 236 (176–296) thousand deaths. In relation to the specific cancer type, lung cancer (LC) is the most frequent, with 199 (140–257) thousand deaths, and malignant pleural mesothelioma (MPM) is the rarest, with 26.8 (24.3–28.6) thousand deaths, but it is virtually only caused by asbestos.1 The International Agency for Research on Cancer (IARC) has classified asbestos as a known human carcinogen not only for LC and MPM, but also for all mesothelioma types, larynx, and ovarian cancer, and, with weaker evidence, for throat, stomach, and colorectum malignancies.2 Moreover, chronic degenerative pleural and interstitial lung diseases, such as asbestosis are caused by asbestos exposure,3 so increasing the associated global morbidity and mortality burden.4 Of note, asbestos exposure is not only occupational, but it may occur in the home, and in surroundings of contaminated worksites with potential exposure of the most vulnerable, such as children and pregnant women.5,6 Regrettably, only 69 of the world's 195 countries have banned asbestos.7 The World Health Organization (WHO) estimated that 125 million people worldwide are still exposed to asbestos, and considering the long cancer latency (up to 60 years for MPM) the expected associated cancer burden won't decrease in the near future.8 Despite important advances in cancer therapy, both LC and MPM have poor prognosis, especially if diagnosed at a late stage. In addition, there is no agreed standard treatment for MPM whose median survival is less than one year from diagnosis.9 Cancer screening programs among ex-exposed to asbestos using low-dose chest CT-scans have been proposed,10 but so far none has started yet due to uncertain cost-benefits and challenges in risk stratification to identify which subgroup of subjects would benefit the most. Therefore, non-invasive biomarkers for risk stratification, and earlier cancer detection are urgently needed to improve overall survival and quality of life, as was recently recommended by the European MPM guidelines.11 MicroRNAs (miRNAs) are short, endogenous, non-coding ribonucleic acids that have been suggested as potential candidates. MiRNAs regulate key processes in cells and signaling pathways involved in lung tumorigenesis, such as cell proliferation, differentiation, angiogenesis, apoptosis, invasion, and metastasis by regulating gene expression at the post-transcriptional level,12 and are potential molecular targets for cancer therapies. Also, the availability of miRNAs in several accessible biological fluids and exhaled breath condensate (EBC) make them ideal candidates for liquid biopsies. Changes in miRNAs expression have been associated with diagnosis and prognosis of several chronic diseases and cancers,13 but the evidence for asbestos-related LC and MPM is still scarce and inconsistent. Therefore, the aim of our systematic literature review is to evaluate the role of miRNAs as diagnostic and prognostic biomarkers of asbestos-related LC and MPM.

We performed the systematic review according to the Cochrane Handbook for Systematic Reviews14 and the Preferred Reporting Items for Systematic Reviews and Meta-analyses guidelines.15 MeSH and free text terms for ‘asbestos’, ‘occupational exposure’, ‘lung cancer’, ‘mesothelioma’ and ‘miRNAs’ were used to search the literature in electronic databases: Medical Literature Analysis and Retrieval System Online (MEDLINE), Excerpta Medica Database (EMBASE) via Ovid platform, PubMed, and Cochrane Library (search period: January 1990 to April 2023) (see detailed search strategy in supplementary Table 1s). A grey literature search was conducted using the Google search engine platform to identify relevant studies not captured through database searches. Our systematic review protocol is registered in the PROSPERO database (registration Number: CRD42023414412 accessible at crd.york.ac.uk/prospero/display_record.php?ID=CRD42023414412). The studies retrieved from our electronic search were reviewed following the Population (P), Interventions (I), Comparators (C), Outcomes (O), Study Design (S), and Time Frame (T) model and screened for suitability according to our inclusion criteria (see PICOS-T criteria in supplementary Table 2s) by two reviewers independently (DM and SDM). A third reviewer was available in case of any disagreement (PC). Relevant data from each included study were extracted into an ad hoc Microsoft Excel (Microsoft Corp., Redmond, WA) table. Each study was appraised for quality using the Newcastle-Ottawa Scale (NOS) for observational studies.15 Studies of very poor quality were not included. For the miRNAs reported at least by two similar studies in association with diagnosis and/or prognosis of asbestos-related LC and/or MPM, a meta-analysis was performed to estimate a pooled quantitative diagnostic and/or prognostic accuracy using the AUC (Area Under the Curve) and its 95 % Confidence Intervals (CIs). We extracted the AUCs and their standard errors (s.e.) from the studies. We displayed graphically in a forest plot the pooled AUCs and their asymptotic intervals (AUC +/- 1.96*s.e.) on a non-transformed scale. Random effect methods16 were applied in case of high heterogeneity (I2>50 %). A potential small study effect bias was evaluated using Egger test and visualised using a funnel plot. The STATA v.17 (Stata Corp LP, College Station, TX) software was used for all analyses.

The literature search retrieved a total of 331 citations. After duplicate removal and application of our selection criteria, 28 articles were included in the review. The quality of the included studies was scored between fair (n = 14), and high (n = 13); only one study17 was removed due to poor quality (see the New-Castle Ottawa quality of studies assessment scale in Table 1). Among the 27 studies included in the final review, only seven were selected as suitable for the meta-analysis (see PRISMA flow chart14 in Fig. 1). The characteristics of the 27 included studies are summarized in Table 2. Most studies were conducted in Europe, specifically in Italy (n = 13).18-30 Regarding the study design, all studies were hospital-based case-controls, but one study31 was a hospital-based retrospective cohort study. Most studies (n = 18) had low sample size. MiRNAs expression was evaluated mostly in blood serum or plasma19,20,22,26-30,32-37 and serum or plasma extracellular vesicles (EV).18,23,38,39 Some studies evaluated miRNA expression directly in the lung tumour tissue,23-25,31,40-42 and one in the EBC.21 The most frequently evaluated disease outcome was MPM, followed by non-small cell lung cancer (NSCLC)22,29,30 and lung adenocarcinoma.21,33 In most studies (n = 15), asbestos-exposed cancer-free hospital patients were used as controls. All but two studies19,26 were only among men. The age range was 52 – 73 years for MPM cases, 61 – 69 for LC cases, 58 – 74 years for non-asbestos-exposed controls and 55 – 76 for asbestos-exposed controls. Ten studies18-20,22,23,27,29,33,36,37 reported the asbestos exposure assessment method, either via personal interview or self-administered questionnaire and six studies20,22,27,36,41,43 calculated the asbestos exposure duration in years. The most frequent type of asbestos exposure was occupational. The majority of studies reported smoking status and some studies19,20,22-25,29,30,32-40 managed to include never smokers only; however seven studies18,26,28,31,41,42,44 did not report the smoking status. All included studies reported miRNAs diagnostic accuracy, and two studies reported the prognostic accuracy19,31 (Table 3). The miRNAs most frequently reported in association with MPM diagnosis were miR-126,22,25,27,29,40 miR-103a-3p,20,36-38 and miR-132–3p.35,37 Of note, miR-126 was confirmed alone,29 and combined with miR-222 22 also for LC diagnosis. Other miRNAs resulted associated with LC, but only by single studies, were miR-1254 and miR-574–5p,30 and let-7f-5p, miR-518f-3p, miR-597–5p, miR-1260a.21 In relation to types of miRNAs perturbations, several miRNAs associated with MPM diagnosis resulted up-regulated,18,19,22,26,30,32-34,41,42 others down-regulated.20,23-25,27,29,31,35-37,39 For LC, all miRNAs resulted up-regulated,21,30 but let-7f-5p.21 Most studies (n = 16) 18,20-24,26,31-33,37,40-44 estimated miRNA expression difference between cases and controls as fold change (FC) using different cut-off thresholds. Some studies managed to adjust tests’ p-values for multiple testing for MPM diagnosis 18,19,23,30,31,40,41,44 and LC diagnosis.21 Eight studies.19,25,32,35-37,39,43 matched cases and controls by potential confounders (e.g. age, sex, smoking status, or asbestos exposure) while nine studies 18,20-24,27,31,40 controlled for them in the statistical analysis. Ten studies 20,21,25,28,30,32-34,40,41 reported also cancer stages and the most frequent was stage I. Twenty-two studies 19-25,27,28,30,31,33-38,40-44 reported the histological subtypes; the most frequent type for asbestos-related LC was adenocarcinoma and for MPM was epithelioid. In relation to diagnostic accuracy (Table 4), the miRNAs with the highest values for MPM were: miR-103a-3p 38 with 86 % sensitivity, 63 % specificity and AUC of 0.76; miR-126 29 with 80 % sensitivity, 60 % specificity and AUC of 0.75(0.62–0.89) and miR-132–3p 35 with 86 % sensitivity, 61 % specificity and AUC of 0.91(0.8–1.0). In relation to MPM prognosis, miR-197‑3p 19 was associated with an increased survival for epithelioid type of 13.5 (±0.6) months, for sarcomatoid of 7.9 (±0.7) months, and for biphasic of 12.4 (±0.6) months. Also, six miRNAs combined (miR-21–5p, miR-23a-3p, miR-30e-5p, miR-221–3p, miR-222–3p, miR-31–5p) 31 resulted associated to 57.2 (45.83–90.48) months to 6.4 (1.94–8.28) months increased survival. In relation to asbestos-related LC diagnosis, miR-126 was confirmed associated also to LC diagnosis, alone,29 and in combination with miR-222.22 Also, miR-1254 and miR-574–5p30 were found associated with early-stage NSCLC even months before clinical diagnosis. No miRNAs were reported in association to survival for asbestos-related LC.

Fig. 1.

PRISMA flow diagram showing screening and selection of articles related to miRNAs and asbestos-related lung cancer outcomes resulting from the search in electronic bibliographic databases.

(0.65MB).

Meta-analysis

We managed to perform a meta-analysis for miRNAs diagnostic accuracy by pooling the AUC reported by seven similar studies that found the same miRNAs associated to MPM diagnosis among men only.25,27,29,35-37,40 One study,37 contributed twice to the meta-analysis for two different mi-RNAs. Random effect methods were used given the high heterogeneity detected (>50 %). We estimated for miR-126, miR-132–3p, and miR-103a-3p pooled diagnostic AUCs of 85 %, 73 %, and 50 %, respectively. The overall pooled accuracy resulted 73 % (Fig. 2). No small study effect bias was shown in the funnel plot (see supplementary file Fig. 1s) as confirmed by the Egger test (p-value= 0.450).

Fig. 2.

Meta-analysis of selected studies evaluating the diagnostic accuracy expressed as Area Under the Curve (AUC) and 95 % Confidence Intervals (CIs) of specific miRNAs for MPM among men only.

Footnote: AUC: Area Under the Curve; s.e.: standard error.

(0.27MB).

In our systematic review and meta-analysis to evaluate the role of miRNAs as potential diagnostic and/or prognostic biomarkers of asbestos-related LC and MPM, we found that several miRNAs are promising candidates especially for MPM diagnosis. In particular, we managed to estimate for the three top miRNAs (miR-126, miR-132–3p, and miR-103a-3p) associated to MPM diagnosis a pooled accuracy of 73 % with the highest performance of 85 % for miR-126. In relation to MPM survival, miR-197‑3p19 and six combined miRNAs (miR-21–5p, miR-23a-3p, miR-30e-5p, miR-221–3p, miR-222–3p, miR-31–5p)31 appeared associated to a better prognosis. In relation to asbestos-related LC, only single miRNAs were associated with diagnosis and/or survival, so we were unable to perform a meta-analysis among the most frequently reported. Of note, miR-126, was confirmed also for LC diagnosis, alone,29 and in association with miR-222.22 Also, miR-1254 and miR-574–5p30 were found associated with early-stage NSCLC samples even months before clinical diagnosis, so potentially useful for LC screening programs. We did not find any promising miRNAs in relation to asbestos-related LC survival.

The strength of our systematic review is the comprehensive search strategy including also the so-called grey literature, as also confirmed by the absence of publication bias in the Egger test for the studies included in the meta-analysis. Also, we used a standard tool to assess study quality. Further, we evaluated both diagnosis and prognosis for both asbestos-related LC and MPM. Finally, we managed to quantify the diagnostic performance of the top miRNAs associated with MPM using a meta-analytic approach, so quantifying their potential role as cancer biomarkers.

We acknowledge several limitations. Almost all studies were case-controls, so subject to reverse causation bias. Almost all studies included men only, and had small sample size, so preventing generalizability of the findings to women. Also, asbestos exposure was usually self-reported and without information on type and duration of exposure, so vulnerable to exposure misclassification; there is even no evidence this was a differential between cases and hospital controls. The studies were also heterogeneous in methodology, and this allowed us to perform a meta-analysis only on a small selected sub-sample of more comparable studies. In addition, given that we pooled AUCs estimated also from multivariable models or studies matched by design, we cannot rule out a certain degree of overestimation.

Several knowledge gaps on this topic remain to be addressed: the accuracy of miRNAs to predict asbestos-related LC and MPM in prospective longitudinal studies of healthy ex-asbestos-exposed subjects, especially among women; the exposure-response relationship between asbestos and miRNA changes; the underlying biological mechanisms, and the influence of internal (e.g. age, sex, genes) and external (e.g. smoking, co-exposure to other occupational agents, therapies) factors on miRNA expression perturbations. Future research addressing these shortcomings in large prospective studies is warranted to shed some light on these issues.

To conclude, in this comprehensive systematic review and meta-analysis we identified some promising miRNA candidates to predict diagnosis and survival of asbestos-related LC and MPM. Future large longitudinal standardized validation studies are needed to confirm these findings, assess their clinical relevance, and address present knowledge gaps. The current poor survival and quality of life for patients affected by asbestos-related lung cancers, especially MPM, urge the identification of accurate and reliable, non-invasive, early diagnostic biomarkers to be included in cancer screening protocols among ex asbestos-exposed subjects, as well as to provide molecular targets for new therapies. However, the best prevention remains to ban asbestos globally to avoid the associated important death toll for the long-term future.