Metabolic and multi-model intravoxel incoherent motion parameters based 18F-FDG PET/MRI for predicting subtypes of inoperable non-small cell lung cancer.

Background: To differentiate inoperable non-small cell lung cancer (NSCLC) subtypes by mono-exponential (MEM), bi-exponential (BEM), and stretched- exponential models (SEM) intravoxel incoherent motion (IVIM), and 18F-FDG PET parameters.

Methods: A total of 106 cases of NSCLC were included in this analysis, of which 68 cases were adenocarcinoma (AC) and 38 cases were squamous cell carcinoma (SCC). MEM derived parameter ADC; BEM derived parameters D, D*, and f, SEM derived parameters α, DDC; and 18F-FDG PET derived parameters MTV, SUVmax, and TLG were recorded and compared. Area under the receiver operating characteristic curve (AUC) was performed for diagnostic efficacy.

Results: SUVmax, MTV and TLG were lower and ADC, f, D and DDC were higher in AC than in SCC (p all < 0.001), whereas D* and α were not significantly different (p = 0.824, 0.152). Logistic regression analysis showed that the stage, ADC, and TLG were independent predictors for identification of SCC and AC, and when combined they showed best diagnostic result (AUC, 0.906; sensitivity, 79.41%; specificity, 94.74%), which was higher than any single clinical factor (maximum diameter, sex smoking, stage, and CT readout; AUC = 0.725, 0.686, 0.707, 0.721, and 0.666, respectively), IVIM (ADC, f, and D; AUC = 0.772, 0.686, and 0.696, respectively) or 18F-FDG PET-derived variable (SUVmax, MTV, and TLG; AUC = 0.693, 0.712, and 0.774, respectively).

Conclusions: The stage, ADC, and TLG were independent predictors for differentiating subtypes of inoperable NSCLC, and when combined they showed optimal diagnostic performance and could be a superior imaging marker.