A Radiomics Nomogram for Preoperative Prediction of Microvascular Invasion in Hepatocellular Carcinoma

Abstract

<b><i>Background:</i></b> Radiomics has emerged as a new approach that can help identify imaging information associated with tumor pathophysiology. We developed and validated a radiomics nomogram for preoperative prediction of microvascular invasion (MVI) in hepatocellular carcinoma (HCC). <b><i>Methods:</i></b> Two hundred and eight patients with pathologically confirmed HCC (training cohort: <i>n</i> = 146; validation cohort: <i>n</i> = 62) who underwent preoperative gadoxetic acid-enhanced magnetic resonance (MR) imaging were included. Least absolute shrinkage and selection operator logistic regression was applied to select features and construct signatures derived from MR images. Univariate and multivariate analyses were used to identify the significant clinicoradiological variables and radiomics signatures associated with MVI, which were then incorporated into the predictive nomogram. The performance of the radiomics nomogram was evaluated by its calibration, discrimination, and clinical utility. <b><i>Results:</i></b> Higher α-fetoprotein level (<i>p</i> = 0.046), nonsmooth tumor margin (<i>p</i> = 0.003), arterial peritumoral enhancement (<i>p <</i> 0.001), and the radiomics signatures of hepatobiliary phase (HBP) T1-weighted images (<i>p <</i> 0.001) and HBP T1 maps (<i>p <</i> 0.001) were independent risk factors of MVI. The predictive model that incorporated the clinicoradiological factors and the radiomic features derived from HBP images outperformed the combination of clinicoradiological factors in the training cohort (area under the curves [AUCs] 0.943 vs. 0.850; <i>p</i> = 0.002), though the validation did not have a statistical significance (AUCs 0.861 vs. 0.759; <i>p</i> = 0.111). The nomogram based on the model exhibited C-index of 0.936 (95% CI 0.895–0.976) and 0.864 (95% CI 0.761–0.967) in the training and validation cohort, fitting well in calibration curves (<i>p</i> > 0.05). Decision curve analysis further confirmed the clinical usefulness of the nomogram. <b><i>Conclusions:</i></b> The nomogram incorporating clinicoradiological risk factors and radiomic features derived from HBP images achieved satisfactory preoperative prediction of the individualized risk of MVI in patients with HCC.

Preoperative Prediction of Microvascular Invasion in Hepatocellular Carcinoma: Initial Application of a Radiomic Algorithm Based on Grayscale Ultrasound Images

“…Extensive studies have shown that radiomics have great potential in predicting tumor biology and in improving implementation of precision medicine (18,23,28,29). Previously, some studies have established radiomic signatures for detecting the presence of MVI based on CT and MRI (11)(12)(13)(14). Radiomic signatures based on arterial phase and delay phase of contrast-enhanced CT have yielded AUCs of 0.684 and 0.490, respectively (11).…”

“…Radiomic signatures based on arterial phase and delay phase of contrast-enhanced CT have yielded AUCs of 0.684 and 0.490, respectively (11). Additionally, radiomic signatures based on hepatobiliary-phase T1-weighted MRI have yielded an AUC of 0.705 in predicting MVI (14). A recent study incorporating clinical risk factors into ultrasound radiomic scores yielded efficacious performance in MVI prediction, with an AUC of 0.731 (18).…”

“…Extensive studies have been proposed to use various imaging methods to predict MVI in HCC. Current reports of MVI classification have been mainly based on computed tomography (CT) (11)(12)(13), magnetic resonance imaging (MRI) (14)(15)(16), and contrast-enhanced ultrasound (CEUS) (17,18). Several imaging features have been proposed as predictors of MVI, such as the status of tumor-internal arteries, hypodense halos on CT scans, arterial peritumoral enhancements, non-smooth tumor margins, and peritumoral hypointensities on gadoxetic-acid-enhanced MRI (16).…”

“…Ma et al, established radiomic signatures based on contrast-enhanced CT to predict the status of MVI (11). Yang et al, constructed radiomic signatures based on MRI for prediction of MVI (14). However, CT and MRI still have limitations, such as CT having a potential risk of radiation exposure, and MRI being relatively expensive and time consuming.…”

“…Grayscale ultrasound is the most commonly used first-line imaging method of HCC lesions before operation, which has unique advantages in terms of being a non-radiation, easy-toperform, and cost-effective imaging method. Recent studies have shown that radiomic analysis can also be applied to ultrasound images (11,14). Radiomic scores based on ultrasound images have potential to non-invasively predict the MVI status in HCC patients (18).…”