Sonographic appearance of focal liver lesions and likelihood of hepatocellular carcinoma in adult Thais with chronic hepatitis B virus infection

PURPOSE

The objective of our study was to study and compare the sonographic findings of hepatocellular carcinoma (HCC) and benign liver lesions, and apply these to an HCC surveillance program in patient with chronic hepatitis B virus (HBV).

METHODS

Sonographic findings of HCC and benign liver lesions were retrospectively reviewed following diagnosis based on either computer tomography or magnetic resonance imaging from July 2010 to December 2020. Multiple sonographic features were analyzed, including internal echogenicity, rim characteristics, and posterior acoustic enhancement. Associations between sonographic characteristics and HCC were assessed using uni- and multi-variate logistic regression analyses.

RESULTS

Of the focal liver lesions in 337 chronic HBV patients, there were 25 HCC and 410 benign lesions, with median sizes of 1.6 and 1.0 cm, respectively. Three ultrasound patterns, homogeneous hypoechogenicity, heterogeneous echogenicity, and hypoechoic rims were more frequently found in HCC than in benign lesions. Moreover, the hypoechoic rim feature was the only sonographic pattern independently associated with HCC (Odds ratio, 68.05; 95% confidence interval, 7.37-628.10; p-values < 0.001). In a subgroup analysis of the lesions sized 2 cm or smaller, no sonographic findings were associated with HCC.

CONCLUSION

A hypoechoic rim was a sonographic feature independently associated with HCC. These findings may aid in improving HCC detection and guiding management during HCC screening and surveillance with ultrasound.

Diagnosis of focal liver lesions from ultrasound images using a pretrained residual neural network

OBJECTIVE

This study aims to develop a ResNet50-based deep learning model for focal liver lesion (FLL) classification in ultrasound images, comparing its performance with other models and prior research.

METHODOLOGY

We retrospectively collected 581 ultrasound images from the Chulabhorn Hospital's HCC surveillance and screening project (2010-2018). The dataset comprised five classes: non-FLL, hepatic cyst (Cyst), hemangioma (HMG), focal fatty sparing (FFS), and hepatocellular carcinoma (HCC). We conducted 5-fold cross-validation after random dataset partitioning, enhancing training data with data augmentation. Our models used modified pre-trained ResNet50, GGN, ResNet18, and VGG16 architectures. Model performance, assessed via confusion matrices for sensitivity, specificity, and accuracy, was compared across models and with prior studies.

RESULTS

ResNet50 outperformed other models, achieving a 5-fold cross-validation accuracy of 87 ± 2.2%. While VGG16 showed similar performance, it exhibited higher uncertainty. In the testing phase, the pretrained ResNet50 excelled in classifying non-FLL, cysts, and FFS. To compare with other research, ResNet50 surpassed the prior methods like two-layered feed-forward neural networks (FFNN) and CNN+ReLU in FLL diagnosis.

CONCLUSION

ResNet50 exhibited good performance in FLL diagnosis, especially for HCC classification, suggesting its potential for developing computer-aided FLL diagnosis. However, further refinement is required for HCC and HMG classification in future studies.

Cost-Effectiveness Analysis of Ultrasound Surveillance for Cholangiocarcinoma in an Endemic Area of Thailand

OBJECTIVE

This research aimed to analyze the cost-effectiveness of ultrasound surveillance for Cholangiocarcinoma (CCA) in a high-risk population residing in the northern region of Thailand and compare these results with the non-surveillance patients who presented with CCA.

METHODS

This was a retrospective descriptive study of two groups monitored and treated for CCA in the same institute, but differing as to whether ultrasound was included in the surveillance. The ultrasound surveillance group comprised those who participated in the cohort study of CCA at Chulabhorn Hospital from 2011 to 2017. Whereas, the non-ultrasound surveillance group consisted of patients treated for CCA at Chulabhorn Hospital from 2009 to 2015 (and not part of the first group). Historical data and societal perspectives were analyzed and interpreted using a decision tree model. The results of one-way sensitivity analyses of probability and cost parameters affecting the change in incremental cost-effectiveness ratio (ICER) were analyzed and presented using tornado diagrams.

RESULT

The ultrasound surveillance group had the quality-adjusted life years gained (QALY gained) of 0.117 years, resulting in the incremental cost-effectiveness ratio (ICER) of 152,985 Thai Baht (THB) or equal to 4,222.6 US-dollars (USD) /QALY gained which is about when compared to the non-ultrasound surveillance group. Moreover, the ICER value was within the willingness to pay (WTP) calculated for Thailand of 4,416.2 USD (160,000 THB) /QALY gained.  The probability of non-diagnostic CCA in non-ultrasound surveillance group and the cost of abdominal ultrasound were found to be major factors affecting the ICER.

CONCLUSION

Ultrasound surveillance of CCA was cost-effective in Thailand. This information can be helpful for national policy planning in regard to Thai populations at high risk of CCA. Further study is recommended to assess cost-effectiveness in other areas with lower incidences of CCA.

Improvement of liver fibrosis, but not steatosis, after HCV eradication as assessment by MR-based imaging: Role of metabolic derangement and host genetic variants

Significant liver fibrosis regression occurs after hepatitis C virus (HCV) therapy. However, the impact of direct-acting antivirals (DAAs) on steatosis is less clear. This study was aimed at evaluating serial fibrosis and steatosis alterations in patients with HCV genotype 1, who achieved sustained virological response (SVR). We enrolled 55 HCV mono-infected and 28 HCV/HIV co-infected patients receiving elbasvir/grazoprevir from a clinical trial. Fibrosis and steatosis were assessed at baseline, follow-up week-24 (FUw24) and week-72 (FUw72) by magnetic resonance elastography (MRE) and proton density fat fraction (PDFF), respectively. Patatin-like phospholipase domain-containing protein 3 (PNPLA3) rs738409, transmembrane six superfamily member 2 (TM6SF2) rs58542926 and membrane bound O-acyltransferase domain-containing 7 (MBOAT7) rs641738 polymorphisms were determined by allelic discrimination. Overall, mean MRE decreased significantly from baseline to FUw24 and FUw72. At FUw72, patients with baseline F2-F4 had higher rate of ≥30% MRE decline compared with individuals with baseline F0-F1 (30.2%vs.3.3%, P = 0.004). In multivariate analysis, significant fibrosis was associated with MRE reduction. The prevalence of steatosis (PDFF≥5.2%) at baseline was 21.7%. Compared to baseline, there were 17 (20.5%) patients with decreased PDFF values at FUw72 (<30%), while 23 (27.7%) patients had increased PDFF values (≥30%). Regarding the overall cohort, mean PDFF significantly increased from baseline to FUw72, and displayed positive correlation with body mass index (BMI) alteration. In multivariate analysis, the presence of diabetes, PNPLA3 CG+GG genotypes and increased BMI at FUw72 were significantly associated with progressive steatosis after SVR. Other genetic variants were not related to fibrosis and steatosis alteration. This study concluded that HCV eradication was associated with fibrosis improvement. However, progressive steatosis was observed in a proportion of patients, particularly among individuals with metabolic derangement and PNPLA3 variants. The combined clinical parameters and host genetic factors might allow a better individualized strategy in this sub-group of patients to alleviate progressive steatosis after HCV cure.