To explore the value and efficacy of the risk model based on the metabolic parameters of ¹⁸F-FDG PET-CT in predicting epidermal growth factor receptor (EGFR) gene mutations in non-small cell lung cancer (NSCLC).

This retrospectives study reviewed 105 NSCLC patients who were tested for EGFR gene expression and underwent ¹⁸F-FDG PET-CT exam prior to treatment from Jan 2017 to June 2018 in our hospital. The patients were divided into EGFR mutations group (n=40) and EGFR wild type group (n=65). The differences between the different groups were analyzed in several clinical characteristics and three metabolic parameters based on ¹⁸F-FDG PET-CT, including the maximal standard uptake value (SUV_max), metabolic tumor volume (MTV), total lesion glycolysis (TLG) of the primary tumor. Multivariate logistic regression analysis was performed to identify predictors of EGFR mutations, and the risk prediction model and nomogram graph were constructed. Diagnostic efficiency of the model was done by the receiver operating characteristics (ROC) curve analysis, and the Calibration plot was performed by Hosmer-Lemeshow (H-L) test to evaluate the calibration scale of the model.

There were statistically significant differences in gender, smoking status, serum CEA level, length of tumor, pathological types, TTF-1 and NapsinA expression between the EGFR mutant groups and EGFR wild-type groups (all P<0.05). The MTV and TLG of EGFR mutation group were 4.4 (4.5，37.1) cm³ and 46.6 (21.2，118.2), respectively. The MTV and TLG of EGFR wild type group were 7.4 (3.2，13.5) cm³ and 95.4 (26.4，345.1), respectively. The MTV and TLG of EGFR mutation group were significantly lower than those of EGFR wild type group (Z=-2.452, P=0.014; Z=-2.379, P=0.017). ROC curve analysis showed area under the curve (AUC) predicted by SUV_max, MTV and TLG for EGFR mutations was 0.597, 0.643 and 0.639, respectively. Multivariate analysis demonstrated that gender, length of tumor, SUV_max and MTV were independent predictors of EGFR mutations, with the odds ratio (OR) values (95%CI) as 3.811 (1.508-9.629), 1.679 (0.899-3.136), 0.928 (0.848-1.015) and 0.924 (0.865-0.986), respectively. The predictive model and nomogram graph was established, with the sensitivity, specificity, positive predictive value, negative predictive value and AUC of 80.0%, 66.2%, 68.8%, 75.3% and 0.775 (0.687-0.864), respectively. The H-L test showed the model had excellent accuracy (χ^²=3.872, P=0.869).

The risk model based on the metabolic parameters of ¹⁸F-FDG PET-CT has a good performance in predicting the mutations of EGFR gene in patients with NSCLC.

Receptor，epidermal growth factor; Carcinoma，non-small-cell lung; Positron-emission tomography