Skip to main content

Circulating microRNA-145 as a diagnostic biomarker for non-small-cell lung cancer: A systemic review and meta-analysis.

Tao S, Ju X, Zhou H, Zeng Q

Publication: Int J Biol Markers (2020) • Page 1724600820967124PMID: 33103527View on PubMed

Purpose

The purpose of this metanalysis was to use published evidence to evaluate circulating microRNA (miRNA, miR)-145 as a diagnostic marker for non-small cell lung cancer (NSCLC) and determine the effect of factors such as use of serum versus plasma, study sample size, time of publication, and blinding during the study on sensitivity and specificity.

Conclusion

The authors report that patient sampling in most cases was inadequately described and the interval between plasma sampling and diagnosis was not specified. Importantly, not all studies included cut-off values for miR-145 used to determine sensitivity and specificity. The sensitivity of miR-145 for the diagnosis of NSCLC in the nine studies ranged from 56% to 93% and the specificity ranged from 55% to 89%. Pooling of the studies found miR-145 had a sensitivity of 78% and specificity of 75% for the diagnosis of NSCLC with a diagnostic odds ratio of 11. The sensitivity and specificity were higher in papers published in the past 3 years than in those published more than 3 years ago. Importantly, selection of serum versus plasma was also identified as a source of variability with an area under the curve for serum of 0.90 versus 0.75 for plasma. While miR-145 in serum was found to have a sensitivity of 84% and specificity of 80% for the diagnosis of NSCLC, the specificity and sensitivity of miR-145 in plasma were 70% and 71%, respectively. miR-145 also displayed slightly higher sensitivity in the not early stage rather than early stage NSCLC, but specificity was comparable. There was no difference in sensitivity or specificity associated with the study sample size or blinding of the investigators.

Studies (1)

Purpose

The purpose of this metanalysis was to use published evidence to evaluate circulating miR-145 as a diagnostic marker for NSCLC and determine the effect of factors such as use of serum versus plasma, study sample size, time of publication, and blinding during the study on sensitivity and specificity. The authors identified nine papers investigating circulating miR-145 levels in plasma or serum of NSCLC by conducting targeted searches of five databases. Only papers with NSCLC diagnosis by the “gold standard” method (pathology or histopathology) and that included sufficient data for sensitivity, specificity, true positive (TP), false positive (FP), false negative (FN), and true negative (TN) to be calculated were included. All studies used real-time RT-PCR to quantify miR-145 but the cut-off values differed among the publications. Quality of each study was assessed using the Cochrane centers QUADAS-2 tool. Meta-regression analysis was used to verify sources of heterogeneity with significance evaluated using Cochrane’s Q statistic. Clinical utility was evaluated using the Fagan nomogram with pretest probabilities of 25%, 50%, and 75%.

Summary of Findings

The authors report that patient sampling in most cases was inadequately described and the interval between plasma sampling and diagnosis was not specified. Importantly, not all studies included cut-off values for miR-145 used to determine sensitivity and specificity. The sensitivity of miR-145 for the diagnosis of NSCLC in the nine studies ranged from 56% to 93% and the specificity ranged from 55% to 89%. Pooling of the studies found miR-145 had a sensitivity of 78% and specificity of 75% for the diagnosis of NSCLC with a diagnostic odds ratio of 11. The sensitivity and specificity were higher in papers published in the past 3 years than in those published more than 3 years ago (sensitivity of 81% versus 72%, P<0.001 and specificity of 84% versus 63%, P<0.001). Importantly, selection of serum versus plasma was also identified as a source of variability with an area under the curve for serum of 0.90 versus 0.75 for plasma. While miR-145 in serum was found to have a sensitivity of 84% and specificity of 80% for the diagnosis of NSCLC, the specificity and sensitivity of miR-145 in plasma were 70% and 71%, respectively (P<0.001, both). miR-145 also displayed slightly higher sensitivity in the not early stage rather than early stage NSCLC (79% versus 76%, P=0.02) but specificity was comparable. There was no difference in sensitivity or specificity associated with the study sample size or blinding of the investigators. The clinical utility of miR-145 was investigated by considering pretest and post-test probabilities. At a pretest probability of 25%, 50%, and 75% the post probability positive (PPP) were 51%, 76%, and 90%; respectively, and the post probability negative (PPN) were 9%, 23%, and 47%; respectively, which the authors state demonstrates good clinical performance.

  • Biospecimens

    FluidPlasma
    FluidSerum
  • Preservation Types

    None (Fresh)Frozen
  • Diagnoses

    NeoplasticCarcinoma
  • Platforms

    RNAReal-time qRT-PCR
  • Pre-analytical Factors

    ClassificationPre-Analytical FactorValue(s)
    PreaquisitionPrognostic factor
    Early stage disease
    Not early stage disease
    PreaquisitionStudy design
    Blinded vs not blinded compared
    Different study sample sizes compared
    Biospecimen AcquisitionTime of biospecimen collection
    >3 years ago
    Published less than 3 years ago
    Biospecimen Aliquots and ComponentsBlood and blood products
    Plasma
    Serum