MicroRNA signatures in tissues and plasma predict development and prognosis of computed tomography detected lung cancer

Abstract

The efficacy of computed tomography (CT) screening for early lung cancer detection in heavy smokers is currently being tested by a number of randomized trials. Critical issues remain the frequency of unnecessary treatments and impact on mortality, indicating the need for biomarkers of aggressive disease. We explored microRNA (miRNA) expression profiles of lung tumors, normal lung tissues and plasma samples from cases with variable prognosis identified in a completed spiral-CT screening trial with extensive follow-up. miRNA expression patterns significantly distinguished: (i) tumors from normal lung tissues, (ii) tumor histology and growth rate, (iii) clinical outcome, and (iv) year of lung cancer CT detection. Interestingly, miRNA profiles in normal lung tissues also displayed remarkable associations with clinical features, suggesting the influence of a permissive microenvironment for tumor development. miRNA expression analyses in plasma samples collected 1-2 y before the onset of disease, at the time of CT detection and in disease-free smokers enrolled in the screening trial, resulted in the generation of miRNA signatures with strong predictive, diagnostic, and prognostic potential (area under the ROC curve ≥ 0.85). These signatures were validated in an independent cohort from a second randomized spiral-CT trial. These results indicate a role for miRNAs in lung tissues and plasma as molecular predictors of lung cancer development and aggressiveness and have theoretical and clinical implication for lung cancer management.

Conflict of interest statement

The authors declare no conflict of interest.

Figures

Fig. 1.

Kaplan–Meier estimates of observed 5-y survival in CT-screening INT-IEO trial. (A) Data arranged according to the extent of disease: 92% for stage I (95% CI: 70.0–97.8) and 7% for stage II–IV (95% CI: 0.5–27.5, P < 0.001). (B) Data arranged according to the year of CT-detection: 77% for lung cancers detected in the first 2 y of the study (95% CI: 53.7–89.8) and 36% for lung cancers diagnosed from third to fifth years (95% CI: 13.7–58.7, P = 0.005)

Fig. 2.

Clustering analysis on 24 normal lung tissue samples using miRNAs differentially expressed between patients with tumors detected in the first 2 y and those of later years of screening. Clinical status of the patient (0 = alive, 1 = dead), tumor stage, and year of tumor detection are reported in columns A, B, and C, respectively.

Fig. 3.

Diagram of samples collection and analysis in the training set (INT-IEO trial; A) and in the validation set (MILD trial; B).

Fig. 4.

miRNA expression analyses in plasma samples collected before the onset and at the time of disease. The signatures of miRNA ratios and their direction in the analyses are listed in the tables. (A) miRNA signature of risk to develop lung cancer and (B) miRNA signature of lung cancer diagnosis. The ROC curves of samples belonging to the validation set are shown. (C) Kaplan–Meier survival curves of patients with miRNA signatures of risk of aggressive disease (RAD) in plasma samples collected 1–2 y before CT-detection of lung cancer. (D) Kaplan–Meier survival curves of patients with miRNA signatures of presence of aggressive disease (PAD) in plasma samples collected at the time of CT-detected lung cancer. The RAD- or PAD-positive patients show a significantly worse survival rate than RAD- or PAD-negative patients (P = 0.0006 and P = 0.0001, respectively).