CT and clinical characteristics that predict risk of EGFR mutation in non-small cell lung cancer: a systematic review and meta-analysis.

INTRODUCTION

To systematically analyze CT and clinical characteristics to find out the risk factors of epidermal growth factor receptor (EGFR) mutation in non-small cell lung cancer (NSCLC). Then the significant characteristics were used to set up a mathematic model to predict EGFR mutation in NSCLC.

MATERIALS AND METHODS

PubMed, Web of Knowledge and EMBASE up to August 17, 2018 were systematically searched for relevant studies that investigated the evidence of association between CT and clinical characteristics and EGFR mutation in NSCLC. After study selection, data extraction, and quality assessment, the pooled odds ratios (ORs) were calculated. Then from May 2017 to August 2018, all NSCLC received EGFR mutation examination and CT examination in our hospital were chosen to test the prediction model by receiver operating characteristic (ROC) curves.

RESULTS

Seventeen original studies met the inclusion criteria. The results showed that the ORs of ground-glass opacity (GGO), air bronchogram, pleural retraction, vascular convergence, smoking history, female gender were, respectively, 1.93 (P = 0.003), 2.09 (P = 0.03), 1.59 (P < 0.01), 1.61 (P = 0.001), 0.28 (P < 0.01), 0.35 (P < 0.01). The result of speculation, cavitation/bubble-like lucency, lesion shape, margin, pathological stage were, respectively, 1.19 (P = 0.32), 0.99 (P = 0.97), 0.82 (P = 0.42), 1.02 (P = 0.90), 0.77 (P = 0.30). 121 NSCLC received EGFR mutation test were included to test the prediction model. The mathematical model based on the results of meta-analysis was: 0.74 × air bronchogram + 0.46 × pleural retraction + 0.48 × vascular convergence - 1.27 × non-smoking history - 1.05 × female. The area under the ROC curve was 0.68.

CONCLUSION

Based on the current evidence, GGO presence, air bronchogram, pleural retraction, vascular convergence were significant risk factors of EGFR mutation in NSCLC. And the prediction model can help to predict EGFR mutation status.