Computed tomography texture analysis to facilitate therapeutic decision making in hepatocellular carcinoma

Models predicting future macrovascular invasion in hepatocellular carcinoma are constructed to assist timely interventions. A total of 366 HCC cases were retrospectively collected from five Chinese hospitals between April 2007 and November 2016: the training dataset comprised 281 patients from four hospitals; the external validation dataset comprised 85 patients from another hospital. Multi-task deep learning network-based models were constructed to predict future macrovascular invasion. The discrimination, calibration, and decision curves were compared to identify the best model. We compared the time to macrovascular invasion and overall survival using the best model and related image heterogeneity scores (H-score). Then, we determined the need for a segmentation subnet or the replacement deep learning algorithm by logistic regression in screening clinical/radiological factors. Finally, an applet was constructed for future application. The best model combined clinical/radiological factors and radiomic features. It achieved best discrimination (areas under the curve: 0·877 in the training dataset and 0·836 in the validation dataset), calibration, and decision curve. Its performance was not affected by the treatments and disease stages. The subgroups had statistical significance for time to macrovascular invasion (training: hazard ratio [HR] = 0·073, 95% confidence interval [CI]: 0·032–0·167, p < 0·001 and validation: HR = 0·090, 95%CI: 0·022–0·366, p < 0·001) and overall survival (training: HR = 0·344, 95%CI: 0·246–0·547, p < 0·001 and validation: HR = 0·489, 95%CI: 0·279 – 0·859, p = 0·003). Similar results were achieved when the patients were subdivided by the H-score. The subnet for segmentation and end-to-end deep learning algorithms improved the performance of the model. Our multi-task deep learning network-based model successfully predicted future macrovascular invasion. In high-risk populations, besides the current first-line treatments, more therapies may be explored for macrovascular invasion.

Background/purpose: Overt hepatic encephalopathy (HE) risk should be preoperatively predicted to identify patients suitable for curative transjugular intrahepatic portosystemic shunt (TIPS) instead of palliative treatments.Methods: A total of 185 patients who underwent TIPS procedure were randomised (130 in the training dataset and 55 in the validation dataset). Clinical factors and imaging characteristics were assessed. Three different models were established by logistic regression analyses based on clinical factors (Model^C), imaging characteristics (Model^I), and a combination of both (Model^CI). Their discrimination, calibration, and decision curves were compared, to identify the best model. Subgroup analysis was performed for the best model.Results: Model^CI, which contained two clinical factors and two imaging characteristics, was identified as the best model. The areas under the curve of Model^C, Model^I, and Model^CIwere 0.870, 0.963, and 0.978 for the training dataset and 0.831, 0.971, and 0.969 for the validation dataset. The combined model outperformed the clinical and imaging models in terms of calibration and decision curves. The performance of Model^CIwas not influenced by total bilirubin, Child–Pugh stages, model of end-stage liver disease score, or ammonia. The subgroup with a risk score ≥ 0.88 exhibited a higher proportion of overt HE (training dataset: 13.3% vs. 97.4%,p < 0.001; validation dataset: 0.0% vs. 87.5%,p < 0.001).Conclusion: Our combination model can successfully predict the risk of overt HE post-TIPS. For the low-risk subgroup, TIPS can be performed safely; however, for the high-risk subgroup, it should be considered more carefully.Graphic

Texture analysis of intermediate-advanced hepatocellular carcinoma: prognosis and patients’ selection of transcatheter arterial chemoembolization and sorafenib

Liver cancer stem cells (LCSCs) have important roles in the occurrence, development, recurrence, therapy resistance and metastasis of hepatocellular carcinoma (HCC). Therefore, intensive studies are undergoing to identify the mechanisms by which LCSCs contribute to HCC invasion and metastasis, and to design more efficient treatments for this disease. With continuous efforts in LCSC research over the years, therapies targeting LCSCs are thought to have great potential for the clinical treatment and prognosis of liver cancer. Novel LCSC surface markers are continuously discovered and several have been used in targeted therapies to reduce HCC recurrence, metastasis, and drug resistance following tumor resection. The present review describes the surface markers characterizing LCSCs and the recent progress in therapies targeting these markers, including antibodies and polypeptides.

Preoperative prediction of overt hepatic encephalopathy (OHE) should be performed in patients with variceal bleeding treated using the transjugular intrahepatic portosystemic shunt (TIPS) procedure. A reliable prediction tool is therefore required. Patients with cirrhosis-related variceal bleeding treated using the TIPS procedure were screened at two hospitals. Patients classified as Child-Pugh Class B were identified. The least absolute shrinkage and selection operator method and the backward stepwise selection method were used to screen the clinical and radiological characteristics of participants. Then, models were constructed accordingly to predict OHE. Area under the receiver operating characteristic curves, calibration curves, and decision curves were performed to discover the optimal model. Finally, whether clinical factors influenced the performance of our optimal model was tested. A total of 191 patients were included (training cohort: 127 cases; validation cohort: 64 cases). Three novel radiological independent risk factors were found. The combined model outperformed the models containing clinical factors or radiological characteristics alone. The areas under the curve for the training and validation cohorts were 0.901 and 0.903, respectively, with satisfactory calibration and decision curves. The Model for End-Stage Liver Disease score, serum sodium, albumin, total bilirubin, and age exhibited limited influence on the performance of the combined model. These radiological characteristics are also independent risk factors for post-TIPS OHE. Combining clinical factors and radiological characteristics was an effective means of predicting OHE. This study's model could be used for preoperative selection of appropriate patients before the TIPS procedure is performed.