Liver elastography malignancy prediction score for noninvasive characterization of focal liver lesions.

ABSTRACT

BACKGROUND & AIMS: To analyse elastographic characteristics of focal liver lesions (FLL)s and diagnostic performance of real-time two-dimensional shear-wave elastography (RT-2D-SWE) in order to differentiate benign and malignant FLLs. METHODS: Consecutive patients diagnosed with FLL by abdominal ultrasound (US) underwent RT-2D-SWE of FLL and non-infiltrated liver by intercostal approach over the right liver lobe. The nature of FLL was determined by diagnostic work-up, including at least one contrast-enhanced imaging modality (MDCT/MRI), check-up of target organs when metastatic disease was suspected and FLL biopsy in inconclusive cases. RESULTS: We analysed 196 patients (median age 60 [range 50-68], 50.5% males) with 259 FLLs (57 hepatocellular carcinomas, 17 cholangiocarcinomas, 94 metastases, 71 haemangiomas, 20 focal nodular hyperplasia) of which 70 (27%) were in cirrhotic liver. Malignant lesions were stiffer (P < .001) with higher variability in intralesional stiffness (P = .001). The best performing cut-off of lesion stiffness was 22.3 kPa (sensitivity 83%; specificity 86%; positive predictive value [PPV] 91.5%; negative predictive value [NPV] 73%) for malignancy. Lesion stiffness <14 kPa had NPV of 96%, while values >32.5 kPa had PPV of 96% for malignancy. Lesion stiffness, lesion/liver stiffness ratio and lesion stiffness variability significantly predicted malignancy in stepwise logistic regression (P < .05), and were used to construct a new Liver Elastography Malignancy Prediction (LEMP) score with accuracy of 96.1% in validation cohort (online calculator available at http//bit.do/lemps). CONCLUSION: The comprehensive approach demonstrated in this study enables correct differentiation of benign and malignant FLL in 96% of patients by using RT-2D-SWE.