Preoperative prediction of hepatocellular carcinoma tumour grade and micro-vascular invasion by means of artificial neural network: a pilot study

Survival Prediction in Intrahepatic Cholangiocarcinoma: A Proof of Concept Study Using Artificial Intelligence for Risk Assessment

Several scoring systems have been devised to objectively predict survival for patients with intrahepatic cholangiocellular carcinoma (ICC) and support treatment stratification, but they have failed external validation. The aim of the present study was to improve prognostication using an artificial intelligence-based approach. We retrospectively identified 417 patients with ICC who were referred to our tertiary care center between 1997 and 2018. Of these, 293 met the inclusion criteria. Established risk factors served as input nodes for an artificial neural network (ANN). We compared the performance of the trained model to the most widely used conventional scoring system, the Fudan score. Predicting 1-year survival, the ANN reached an area under the ROC curve (AUC) of 0.89 for the training set and 0.80 for the validation set. The AUC of the Fudan score was significantly lower in the validation set (0.77, p < 0.001). In the training set, the Fudan score yielded a lower AUC (0.74) without reaching significance (p = 0.24). Thus, ANNs incorporating a multitude of known risk factors can outperform conventional risk scores, which typically consist of a limited number of parameters. In the future, such artificial intelligence-based approaches have the potential to improve treatment stratification when models trained on large multicenter data are openly available.

Clinical value of artificial intelligence in hepatocellular carcinoma: Current status and prospect

Hepatocellular carcinoma (HCC) is the most commonly diagnosed type of liver cancer and the fourth leading cause of cancer-related mortality worldwide. The early identification of HCC and effective treatments for it have been challenging. Due to the sufficient compensatory ability of early patients and its nonspecific symptoms, HCC is more likely to escape diagnosis in the incipient stage, during which patients can achieve a more satisfying overall survival if they undergo resection or liver transplantation. Patients at advanced stages can profit from radical therapies in a limited way. In order to improve the unfavorable prognosis of HCC, diagnostic ability and treatment efficiency must be improved. The past decade has seen rapid advancements in artificial intelligence, underlying its unique usefulness in almost every field, including that of medicine. Herein, we sought and reviewed studies that put emphasis on artificial intelligence and HCC.

Role of Machine Learning and Artificial Intelligence in Interventional Oncology

PURPOSE OF REVIEW

The purpose of this review is to highlight the current role of machine learning and artificial intelligence and in the field of interventional oncology.

RECENT FINDINGS

With advancements in technology, there is a significant amount of research regarding the application of artificial intelligence and machine learning in medicine. Interventional oncology is a field that can benefit greatly from this research through enhanced image analysis and intraprocedural guidance. These software developments can increase detection of cancers through routine screening and improve diagnostic accuracy in classifying tumors. They may also aid in selecting the most effective treatment for the patient by predicting outcomes based on a combination of both clinical and radiologic factors. Furthermore, machine learning and artificial intelligence can advance intraprocedural guidance for the interventional oncologist through more accurate needle tracking and image fusion technology. This minimizes damage to nearby healthy tissue and maximizes treatment of the tumor. While there are several exciting developments, this review also discusses limitations before incorporating machine learning and artificial intelligence in the field of interventional oncology. These include data capture and processing, lack of transparency among developers, validating models, integrating workflow, and ethical challenged. In summary, machine learning and artificial intelligence have the potential to positively impact interventional oncologists and how they provide cancer care treatments.

Nomogram development and validation to predict hepatocellular carcinoma tumor behavior by preoperative gadoxetic acid-enhanced MRI

OBJECTIVES

Pretreatment evaluation of tumor biology and microenvironment is important to predict prognosis and plan treatment. We aimed to develop nomograms based on gadoxetic acid-enhanced MRI to predict microvascular invasion (MVI), tumor differentiation, and immunoscore.

METHODS

This retrospective study included 273 patients with HCC who underwent preoperative gadoxetic acid-enhanced MRI. Patients were assigned to two groups: training (N = 191) and validation (N = 82). Univariable and multivariable logistic regression analyses were performed to investigate clinical variables and MRI features' associations with MVI, tumor differentiation, and immunoscore. Nomograms were developed based on features associated with these three histopathological features in the training cohort, then validated, and evaluated.

RESULTS

Predictors of MVI included tumor size, rim enhancement, capsule, percent decrease in T1 images (T1_D%), standard deviation of apparent diffusion coefficient, and alanine aminotransferase levels, while capsule, peritumoral enhancement, mean relaxation time on the hepatobiliary phase (T1_E), and alpha-fetoprotein levels predicted tumor differentiation. Predictors of immunoscore included the radiologic score constructed by tumor number, intratumoral vessel, margin, capsule, rim enhancement, T1_D%, relaxation time on plain scan (T1_P), and alpha-fetoprotein and alanine aminotransferase levels. Three nomograms achieved good concordance indexes in predicting MVI (0.754, 0.746), tumor differentiation (0.758, 0.699), and immunoscore (0.737, 0.726) in the training and validation cohorts, respectively.

CONCLUSION

MRI-based nomograms effectively predict tumor behaviors in HCC and may assist clinicians in prognosis prediction and pretreatment decisions.

KEY POINTS

• This study developed and validated three nomograms based on gadoxetic acid-enhanced MRI to predict MVI, tumor differentiation, and immunoscore in patients with HCC. • The pretreatment prediction of tumor microenvironment may be useful to guide accurate prognosis and planning of surgical and immunological therapies for individual patients with HCC.

Micronvasive behaviour of single small hepatocellular carcinoma: which treatment?

BACKGROUND

Microinvasion (MI), defined as infiltration of the portal or hepatic vein or bile duct and intrahepatic metastasis are accurate indicators of a poor prognosis for mall hepatocellular carcinomas (HCC). A previous study showed that intraoperative ultrasound (IOUS) definition of MI-HCC had a high concordance with histological findings. Aim of this study is to evaluate overall survival and recurrence patterns of patients with MI-HCC submitted to hepatic resection (HR) or laparoscopic ablation therapies (LAT).

METHODS

A total of 171 consecutive patients (78 h; 93 LAT) with single, small HCC (< 3 cm) with a MI pattern at IOUS examination were compared analyzing overall survival and recurrence patterns using univariate and multivariate analysis and weighting by propensity score.

RESULTS

Overall recurrences were similar in the 2 groups (HR: 51 patients (65%); LAT: 66 patients (71%)). The rate of local tumor progression in the HR group was very low (5 pts; 6%) in comparison to LAT group (22 pts; 24%; p = 0.002). The overall survival curves of HR are significantly better than that of the LAT group (p = 0.0039). On the propensity score Cox model, overall mortality was predicted by the surgical treatment with a Hazard ratio 1.68 (1.08-2.623) (p = 0.022).

CONCLUSIONS

If technically feasible and in patients fit for surgery, HR with an adequate tumor margin should be preferred to LAT in patients with MI-HCC at IOUS evaluation, to eradicate MI features near the main nodule, which are relatively frequent even in small HCC (< 3 cm).

Microvascular Invasion Status and Its Survival Impact in Hepatocellular Carcinoma Depend on Tissue Sampling Protocol

BACKGROUND

The aim of this work is to explore the impact of the number of sampling sites (NuSS) and sampling location on microvascular invasion (MVI) detection rate and long-term survival of hepatocellular carcinoma (HCC), and determine the minimum NuSS for sufficient MVI detection.

PATIENTS AND METHODS

From January 2008 to March 2017, 1144 HCC patients who underwent hepatectomy were retrospectively enrolled. Associations between NuSS and MVI positive rates and overall survival were investigated. NuSS thresholds were determined by Chow test and confirmed prospectively in 305 patients from April 2017 to February 2019. In the prospective cohort, the distribution of MVI in different sampling locations and its prognostic effect was evaluated.

RESULTS

MVI positive rates increased as NuSS increased, steadily reaching a plateau when NuSS reached a threshold. A threshold of four, six, eight, and eight sampling sites within paracancerous parenchyma ≤ 1 cm from tumor was required for detecting MVI in solitary tumors measuring 1.0-3.0, 3.1-4.9, and ≥ 5.0 cm and multiple tumors. Patients with adequate NuSS achieved longer survival than those with inadequate NuSS [hazard ratio (HR) = 0.75, P = 0.043]. For all MVI-positive patients, MVI could be detected positive in paracancerous parenchyma ≤ 1 cm from tumor. Patients with MVI positive in paracancerous parenchyma > 1 cm had higher recurrence risk than those with MVI positive only in parenchyma ≤ 1 cm (HR = 6.05, P < 0.001).

CONCLUSIONS

Adequate NuSS is associated with higher MVI detection rate and better survival of HCC patients. We recommend four, six, eight, and eight as the cut-points for evaluating MVI sampling quality and patients' prognostic stratification in the subgroups of solitary tumors measuring 1.0-3.0 cm, 3.1-4.9 cm and ≥ 5.0 cm and multiple tumors, respectively.

Indication of Liver Transplantation for Hepatocellular Carcinoma Should Be Reconsidered in Case of Microvascular Invasion and Multilocular Tumor Occurrence

Liver transplantation (LT) is routinely performed for hepatocellular carcinoma (HCC) in cirrhosis without major vascular invasion. Although the adverse influence of microvascular invasion is recognized, its occurrence does not contraindicate LT. We retrospectively analyzed in our LT cohort the significance of microvascular invasion on survival and demonstrate bridging procedures. At our hospital, 346 patients were diagnosed with HCC, 171 patients were evaluated for LT, and 153 were listed at Eurotransplant during a period of 11 years. Among these, 112 patients received LT and were included in this study. Overall survival after 1, 3 and 5 years was 86.3%, 73.9%, and 67.9%, respectively. Microvascular invasion led to significantly reduced overall (p = 0.030) and disease-free survival (p = 0.002). Five-year disease-free survival with microvascular invasion was 10.5%. Multilocular tumor occurrence with simultaneous microvascular invasion revealed the worst prognosis. In our LT cohort, predominant bridging treatment was transarterial chemoembolization (TACE) and the number of TACE significantly correlated with poorer overall survival after LT (p = 0.028), which was confirmed in multiple Cox regression analysis for overall and disease-free survival (p = 0.015 and p = 0.011). Microvascular tumor invasion is significantly associated with reduced prognosis after LT, which is aggravated by simultaneous occurrence of multiple lesions. Therefore, indication strategies for LT should be reconsidered.

Preoperative identification of microvascular invasion in hepatocellular carcinoma by XGBoost and deep learning

PURPOSE

Microvascular invasion (MVI) is a valuable predictor of survival in hepatocellular carcinoma (HCC) patients. This study developed predictive models using eXtreme Gradient Boosting (XGBoost) and deep learning based on CT images to predict MVI preoperatively.

METHODS

In total, 405 patients were included. A total of 7302 radiomic features and 17 radiological features were extracted by a radiomics feature extraction package and radiologists, respectively. We developed a XGBoost model based on radiomics features, radiological features and clinical variables and a three-dimensional convolutional neural network (3D-CNN) to predict MVI status. Next, we compared the efficacy of the two models.

RESULTS

Of the 405 patients, 220 (54.3%) were MVI positive, and 185 (45.7%) were MVI negative. The areas under the receiver operating characteristic curves (AUROCs) of the Radiomics-Radiological-Clinical (RRC) Model and 3D-CNN Model in the training set were 0.952 (95% confidence interval (CI) 0.923-0.973) and 0.980 (95% CI 0.959-0.993), respectively (p = 0.14). The AUROCs of the RRC Model and 3D-CNN Model in the validation set were 0.887 (95% CI 0.797-0.947) and 0.906 (95% CI 0.821-0.960), respectively (p = 0.83). Based on the MVI status predicted by the RRC and 3D-CNN Models, the mean recurrence-free survival (RFS) was significantly better in the predicted MVI-negative group than that in the predicted MVI-positive group (RRC Model: 69.95 vs. 24.80 months, p < 0.001; 3D-CNN Model: 64.06 vs. 31.05 months, p = 0.027).

CONCLUSION

The RRC Model and 3D-CNN models showed considerable efficacy in identifying MVI preoperatively. These machine learning models may facilitate decision-making in HCC treatment but requires further validation.

Anatomical resection is useful for the treatment of primary solitary hepatocellular carcinoma with predicted microscopic vessel invasion and/or intrahepatic metastasis

PURPOSE

The aim of this study was to evaluate anatomical resection (AR) versus non-AR for primary solitary hepatocellular carcinoma (HCC) with predicted microscopic vessel invasion (MVI) and/or microscopic intrahepatic metastasis (MIM).

METHODS

This retrospective study included 358 patients who underwent hepatectomy and had no evidence of MVI and/or MIM on preoperative imaging. The predictors of MVI and/or MIM were identified. The AR group (n = 222) and the non-AR group (n = 136) were classified by number of risk factor, and the survival rates were compared.

RESULTS

Microscopic vessel invasion and/or MIM were identified in 81 (22.6%) patients. A multivariate analysis showed that high des-gamma-carboxy prothrombin concentration [odds ratio (OR) 3.35], large tumor size (OR 3.16), and high aspartate aminotransferase concentration (OR 2.13) were significant predictors. The 5-year overall survival (OS) in the patients with zero, one, two, and three risk factors were 97.4%, 73.5%, 71.5%, and 65.5%, respectively. The OS of AR is superior to that of non-AR only in patients with one or two risk factors.

CONCLUSION

The present findings suggest that AR should be performed for patients with one or two risk factors, and that AR may prevent recurrence, as these patients are at risk of having MVI and/or MIM.

New insights into the pathophysiology and clinical care of rare primary liver cancers

Hepatocholangiocarcinoma, fibrolamellar carcinoma, hepatic haemangioendothelioma and hepatic angiosarcoma represent less than 5% of primary liver cancers. Fibrolamellar carcinoma and hepatic haemangioendothelioma are driven by unique somatic genetic alterations (DNAJB1-PRKCA and CAMTA1-WWTR1 fusions, respectively), while the pathogenesis of hepatocholangiocarcinoma remains more complex, as suggested by its histological diversity. Histology is the gold standard for diagnosis, which remains challenging even in an expert centre because of the low incidences of these liver cancers. Resection, when feasible, is the cornerstone of treatment, together with liver transplantation for hepatic haemangioendothelioma. The role of locoregional therapies and systemic treatments remains poorly studied. In this review, we aim to describe the recent advances in terms of diagnosis and clinical management of these rare primary liver cancers.

Artificial intelligence in transplantation (machine-learning classifiers and transplant oncology)

PURPOSE OF REVIEW

To highlight recent efforts in the development and implementation of machine learning in transplant oncology - a field that uses liver transplantation for the treatment of hepatobiliary malignancies - and particularly in hepatocellular carcinoma, the most commonly treated diagnosis in transplant oncology.

RECENT FINDINGS

The development of machine learning has occurred within three domains related to hepatocellular carcinoma: identification of key clinicopathological variables, genomics, and image processing.

SUMMARY

Machine-learning classifiers can be effectively applied for more accurate clinical prediction and handling of data, such as genetics and imaging in transplant oncology. This has allowed for the identification of factors that most significantly influence recurrence and survival in disease, such as hepatocellular carcinoma, and thus help in prognosticating patients who may benefit from a liver transplant. Although progress has been made in using these methods to analyse clinicopathological information, genomic profiles, and image processed data (both histopathological and radiomic), future progress relies on integrating data across these domains.

Integration of pre-surgical blood test results predict microvascular invasion risk in hepatocellular carcinoma

Microvascular invasion (MVI) is one of the most important factors leading to poor prognosis for hepatocellular carcinoma (HCC) patients, and detection of MVI prior to surgical operation could great benefit patient's prognosis and survival. Since it is still lacking effective non-invasive strategy for MVI detection before surgery, novel MVI determination approaches were in urgent need. In this study, complete blood count, blood test and AFP test results are utilized to perform preoperative prediction of MVI based on a novel interpretable deep learning method to quantify the risk of MVI. The proposed method termed as "Interpretation based Risk Prediction" can estimate the MVI risk precisely and achieve better performance compared with the state-of-art MVI risk estimation methods with concordance indexes of 0.9341 and 0.9052 on the training cohort and the independent validation cohort, respectively. Moreover, further analyses of the model outputs demonstrate that the quantified risk of MVI from our model could serve as an independent preoperative risk factor for both recurrence-free survival and overall survival of HCC patients. Thus, our model showed great potential in quantification of MVI risk and prediction of prognosis for HCC patients.

A novel model for predicting fatty liver disease by means of an artificial neural network

BACKGROUND

The artificial neural network (ANN) emerged recently as a potent diagnostic tool, especially for complicated systemic diseases. This study aimed to establish a diagnostic model for the recognition of fatty liver disease (FLD) by virtue of the ANN.

METHODS

A total of 7,396 pairs of gender- and age-matched subjects who underwent health check-ups at the First Affiliated Hospital, College of Medicine, Zhejiang University (Hangzhou, China) were enrolled to establish the ANN model. Indices available in health check-up reports were utilized as potential input variables. The performance of our model was evaluated through a receiver-operating characteristic (ROC) curve analysis. Other outcome measures included diagnostic accuracy, sensitivity, specificity, Cohen's k coefficient, Brier score, and Hosmer-Lemeshow test. The Fatty Liver Index (FLI) and the Hepatic Steatosis Index (HSI), retrained using our training-group data with its original designated input variables, were used as comparisons in the capability of FLD diagnosis.

RESULTS

Eight variables (age, gender, body mass index, alanine aminotransferase, aspartate aminotransferase, uric acid, total triglyceride, and fasting plasma glucose) were eventually adopted as input nodes of the ANN model. By applying a cut-off point of 0.51, the area under ROC curves of our ANN model in predicting FLD in the testing group was 0.908 [95% confidence interval (CI), 0.901-0.915]-significantly higher (P < 0.05) than that of the FLI model (0.881, 95% CI, 0.872-0.891) and that of the HSI model (0.885; 95% CI, 0.877-0.893). Our ANN model exhibited higher diagnostic accuracy, better concordance with ultrasonography results, and superior capability of calibration than the FLI model and the HSI model.

CONCLUSIONS

Our ANN system showed good capability in the diagnosis of FLD. It is anticipated that our ANN model will be of both clinical and epidemiological use in the future.

Prognostic role of artificial intelligence among patients with hepatocellular cancer: A systematic review

BACKGROUND

Prediction of survival after the treatment of hepatocellular carcinoma (HCC) has been widely investigated, yet remains inadequate. The application of artificial intelligence (AI) is emerging as a valid adjunct to traditional statistics due to the ability to process vast amounts of data and find hidden interconnections between variables. AI and deep learning are increasingly employed in several topics of liver cancer research, including diagnosis, pathology, and prognosis.

AIM

To assess the role of AI in the prediction of survival following HCC treatment.

METHODS

A web-based literature search was performed according to the Preferred Reporting Items for Systemic Reviews and Meta-Analysis guidelines using the keywords "artificial intelligence", "deep learning" and "hepatocellular carcinoma" (and synonyms). The specific research question was formulated following the patient (patients with HCC), intervention (evaluation of HCC treatment using AI), comparison (evaluation without using AI), and outcome (patient death and/or tumor recurrence) structure. English language articles were retrieved, screened, and reviewed by the authors. The quality of the papers was assessed using the Risk of Bias In Non-randomized Studies of Interventions tool. Data were extracted and collected in a database.

RESULTS

Among the 598 articles screened, nine papers met the inclusion criteria, six of which had low-risk rates of bias. Eight articles were published in the last decade; all came from eastern countries. Patient sample size was extremely heterogenous (n = 11-22926). AI methodologies employed included artificial neural networks (ANN) in six studies, as well as support vector machine, artificial plant optimization, and peritumoral radiomics in the remaining three studies. All the studies testing the role of ANN compared the performance of ANN with traditional statistics. Training cohorts were used to train the neural networks that were then applied to validation cohorts. In all cases, the AI models demonstrated superior predictive performance compared with traditional statistics with significantly improved areas under the curve.

CONCLUSION

AI applied to survival prediction after HCC treatment provided enhanced accuracy compared with conventional linear systems of analysis. Improved transferability and reproducibility will facilitate the widespread use of AI methodologies.

Can LI-RADS imaging features at gadoxetic acid-enhanced MRI predict aggressive features on pathology of single hepatocellular carcinoma?

PURPOSE

To investigate whether Liver Imaging Reporting and Data System (LI-RADS) imaging features at preoperative gadoxetic acid-enhanced MRI can predict microvascular invasion (MVI) and histologic grade of hepatocellular carcinoma (HCC) and to evaluate their associations with recurrence after curative resection of single HCC.

MATERIALS AND METHODS

From July 2015 to September 2018, 111 consecutive patients with pathologically confirmed HCC who underwent gadoxetic acid-enhanced MRI within 1 month before surgery were included in this retrospective study. Significant MRI findings and clinical parameters for predicting MVI, high-grade HCCs and postoperative recurrence were identified by logistic regression model and Cox proportional hazards model.

RESULTS

Twenty-six of 111 (23.4 %) patients had MVI and 36 of 111 (32.4 %) patients had high-grade HCCs, whereas 44 of 95 (46.3 %) patients experienced recurrence. Tumor size > 5 cm (OR = 9.852; p < 0.001) and absence of nodule-in-nodule architecture (OR = 8.302; p = 0.001) were independent predictors of MVI. Enhancing capsule (OR = 4.396; p = 0.004) and corona enhancement (OR = 3.765; p = 0.021) were independent predictors of high-grade HCCs. Blood products in mass (HR = 2.275; p = 0.009), corona enhancement (HR = 4.332; p < 0.001), and serum AFP level > 400 ng/mL (HR = 2.071; p = 0.023) were independent predictors of recurrence.

CONCLUSION

LI-RADS imaging features can be used as potential biomarkers for predicting aggressive pathologic features and recurrence of HCC. The identification of prognostic LI-RADS imaging features may facilitate the selection of surgical candidates and optimize the management of HCC patients.

Texture analysis on preoperative contrast-enhanced magnetic resonance imaging identifies microvascular invasion in hepatocellular carcinoma

BACKGROUND

Radiomic texture analysis quantifies tumor heterogeneity. The aim of this study is to determine if radiomics can predict biologic aggressiveness in HCC and identify tumors with MVI.

METHODS

Single-center, retrospective review of HCC patients undergoing resection/ablation with curative intent from 2009 to 2017. DICOM images from preoperative MRIs were analyzed with texture analysis software. Texture analysis parameters extracted on T1, T2, hepatic arterial phase (HAP) and portal venous phase (PVP) images. Multivariate logistic regression analysis evaluated factors associated with MVI.

RESULTS

MVI was present in 52.2% (n = 133) of HCCs. On multivariate analysis only T1 mean (OR = 0.97, 95%CI 0.95-0.99, p = 0.043) and PVP entropy (OR = 4.7, 95%CI 1.37-16.3, p = 0.014) were associated with tumor MVI. Area under ROC curve was 0.83 for this final model. Empirical optimal cutpoint for PVP tumor entropy and T1 tumor mean were 5.73 and 23.41, respectively. At these cutpoint values, sensitivity was 0.68 and 0.5, respectively and specificity was 0.64 and 0.86. When both criteria were met, the probability of MVI in the tumor was 87%.

CONCLUSION

Tumor entropy and mean are both associated with MVI. Texture analysis on preoperative imaging correlates with microscopic features of HCC and can be used to predict patients with high-risk tumors.

Development of machine learning-based clinical decision support system for hepatocellular carcinoma

There is a significant discrepancy between the actual choice for initial treatment option for hepatocellular carcinoma (HCC) and recommendations from the currently used BCLC staging system. We develop a machine learning-based clinical decision support system (CDSS) for recommending initial treatment option in HCC and predicting overall survival (OS). From hospital records of 1,021 consecutive patients with HCC treated at a single centre in Korea between January 2010 and October 2010, we collected information on 61 pretreatment variables, initial treatment, and survival status. Twenty pretreatment key variables were finally selected. We developed the CDSS from the derivation set (N = 813) using random forest method and validated it in the validation set (N = 208). Among the 1,021 patients (mean age: 56.9 years), 81.8% were male and 77.0% had positive hepatitis B BCLC stages 0, A, B, C, and D were observed in 13.4%, 26.0%, 18.0%, 36.6%, and 6.3% of patients, respectively. The six multi-step classifier model was developed for treatment decision in a hierarchical manner, and showed good performance with 81.0% of accuracy for radiofrequency ablation (RFA) or resection versus not, 88.4% for RFA versus resection, and 76.8% for TACE or not. We also developed seven survival prediction models for each treatment option. Our newly developed HCC-CDSS model showed good performance in terms of treatment recommendation and OS prediction and may be used as a guidance in deciding the initial treatment option for HCC.

A-G Score Associated With Outcomes in Solitary Hepatocellular Carcinoma Patients After Hepatectomy

Aim: The study aimed to investigate the clinical significance of preoperative alpha-fetoprotein (AFP) and gamma-glutamyl transferase (GGT) (A-G score) on hepatocellular carcinoma (HCC) patients.

Methods: A total of 474 solitary HCC patients were included. Survival analysis was evaluated by Kaplan-Meier method. Prognostic factors were analyzed in a multivariate model. The comparison of the predictive value of AFP, GGT, and A-G score was performed by receiver operating characteristic curve (ROC) analysis and decision curve analysis (DCA).

Results: Of the 474 patients, 137(28.9%), 241(50.8%), and 96(20.3%) patients were assigned to A-G score 0, 1, and 2, respectively. In multivariate analysis, A-G score, tumor size, microvascular invasion, tumor differentiation, satellite lesion, and state of HBV infection were independently predictive factors for RFS of solitary HCC patients. The A-G score could significantly stratify solitary HCC patients with a distinguished prognosis. The 1-, 3-, and 5-year RFS and OS among patients with A-G score 1 was better than that of patients with A-G score 2 and worse than that of patients with A-G score 0(all p < 0.05). Based on the result from the ROC analysis and DCA analysis, the A-G score appeared to be superior to either AFP or GGT alone in the prediction of prognosis of solitary HCC patients. In the subgroup analysis, the A-G score could accurately predict the prognosis of solitary HCC patients without MVI or with liver cirrhosis.

Conclusions: Preoperative A-G score could effectively and simply predict prognosis of solitary HCC patients after hepatectomy, especially for those with non-MVI solitary HCC or those with liver cirrhosis.

Autoantibody signature in hepatocellular carcinoma using seromics

Background

Alpha-fetoprotein (AFP) is a widely used biomarker for hepatocellular carcinoma (HCC) early detection. However, low sensitivity and false negativity of AFP raise the requirement of more effective early diagnostic approaches for HCC.

Methods

We employed a three-phase strategy to identify serum autoantibody (AAb) signature for HCC early diagnosis using protein array-based approach. A total of 1253 serum samples from HCC, liver cirrhosis, and healthy controls were prospectively collected from three liver cancer centers in China. The Human Proteome Microarray, comprising 21,154 unique proteins, was first applied to identify AAb candidates in discovery phase (n = 100) and to further fabricate HCC-focused arrays. Then, an artificial neural network (ANN) model was used to discover AAbs for HCC detection in a test phase (n = 576) and a validation phase (n = 577), respectively.

Results

Using HCC-focused array, we identified and validated a novel 7-AAb panel containing CIAPIN1, EGFR, MAS1, SLC44A3, ASAH1, UBL7, and ZNF428 for effective HCC detection. The ANN model of this panel showed improvement of sensitivity (61.6–77.7%) compared to AFP (cutoff 400 ng/mL, 28.4–30.7%). Notably, it was able to detect AFP-negative HCC with AUC values of 0.841–0.948. For early-stage HCC (BCLC 0/A) detection, it outperformed AFP (cutoff 400 ng/mL) with approximately 10% increase in AUC.

Conclusions

The 7-AAb panel provides potentially clinical value for non-invasive early detection of HCC, and brings new clues on understanding the immune response against hepatocarcinogenesis.

Preoperative Prediction Power of Imaging Methods for Microvascular Invasion in Hepatocellular Carcinoma: A Systemic Review and Meta-Analysis

Background: To compare the predictive power between radiomics and non-radiomics (conventional imaging and functional imaging methods) for preoperative evaluation of microvascular invasion (MVI) in hepatocellular carcinoma (HCC).

Methods: Comprehensive publications were screened in PubMed, Embase, and Cochrane Library. Studies focusing on the discrimination values of imaging methods, including radiomics and non-radiomics methods, for MVI evaluation were included in our meta-analysis.

Results: Thirty-three imaging studies with 5,462 cases, focusing on preoperative evaluation of MVI status in HCC, were included. The sensitivity and specificity of MVI prediction in HCC were 0.78 [95% confidence interval (CI): 0.75–0.80; I² = 70.7%] and 0.78 (95% CI: 0.76–0.81; I² = 0.0%) for radiomics, respectively, and were 0.73 (95% CI: 0.71–0.75; I² = 83.7%) and 0.82 (95% CI: 0.80–0.83; I² = 86.5%) for non-radiomics, respectively. The areas under the receiver operation curves for radiomics and non-radiomics to predict MVI status in HCC were 0.8550 and 0.8601, respectively, showing no significant difference.

Conclusion: The imaging method is feasible to predict the MVI state of HCC. Radiomics method based on medical image data is a promising application in clinical practice and can provide quantifiable image features. With the help of these features, highly consistent prediction performance will be achieved in anticipation.

Impact of Surgical Margin on the Prognosis of Early Hepatocellular Carcinoma (≤5 cm): A Propensity Score Matching Analysis

Aim: The influence of surgical margin on the prognosis of patients with early solitary hepatocellular carcinoma (HCC) (≤5 cm) is undetermined.

Methods: The data of 904 patients with early solitary HCC who underwent liver resection were collected for recurrence-free survival (RFS) and overall survival (OS). Propensity score matching (PSM) was performed to balance the potential bias.

Results: Log-rank tests showed that 2 mm was the best cutoff value to discriminate the prognosis of early HCC. Liver resection with a >2 mm surgical margin distance (wide-margin group) led to better 5-year RFS and OS rate compared with liver resection with a ≤2 mm surgical margin distance (narrow-margin group) among patients both before (RFS: 59.1% vs. 39.6%, P < 0.001; OS: 85.3% vs. 73.7%, P < 0.001) and after PSM (RFS: 56.3% vs. 41.0%, P < 0.001; OS: 83.0% vs. 75.0%, P = 0.010). Subgroup analysis showed that a wide-margin resection significantly improved the prognosis of patients with microvascular invasion (RFS: P < 0.001; OS: P = 0.001) and patients without liver cirrhosis (RFS: P < 0.001; OS: P = 0.001) after PSM. Multivariable Cox regression analysis revealed that narrow-margin resection is associated with poorer RFS [hazard ratio (HR) = 1.781, P < 0.001), OS (HR = 1.935, P < 0.001], and early recurrence (HR = 1.925, P < 0.001).

Conclusions: A wide-margin resection resulted in better clinical outcomes than a narrow-margin resection among patients with early solitary HCC, especially for those with microvascular invasion and without cirrhosis. An individual strategy of surgical margin should be formulated preoperation according to both tumor factors and background liver factors.

Development, Validation and Comparison of Artificial Neural Network Models and Logistic Regression Models Predicting Survival of Unresectable Pancreatic Cancer

Background: Prediction models for the overall survival of pancreatic cancer remain unsatisfactory. We aimed to explore artificial neural networks (ANNs) modeling to predict the survival of unresectable pancreatic cancer patients.

Methods: Thirty-two clinical parameters were collected from 221 unresectable pancreatic cancer patients, and their prognostic ability was evaluated using univariate and multivariate logistic regression. ANN and logistic regression (LR) models were developed on a training group (168 patients), and the area under the ROC curve (AUC) was used for comparison of the ANN and LR models. The models were further tested on the testing group (53 patients), and k-statistics were used for accuracy comparison.

Results: We built three ANN models, based on 3, 7, and 32 basic features, to predict 8 month survival. All 3 ANN models showed better performance, with AUCs significantly higher than those from the respective LR models (0.811 vs. 0.680, 0.844 vs. 0.722, 0.921 vs. 0.849, all p < 0.05). The ability of the ANN models to discriminate 8 month survival with higher accuracy than the respective LR models was further confirmed in 53 consecutive patients.

Conclusion: We developed ANN models predicting the 8 month survival of unresectable pancreatic cancer patients. These models may help to optimize personalized patient management.

Predicting survival after transarterial chemoembolization for hepatocellular carcinoma using a neural network: A Pilot Study

BACKGROUND AND AIMS

Deciding when to repeat and when to stop transarterial chemoembolization (TACE) in patients with hepatocellular carcinoma (HCC) can be difficult even for experienced investigators. Our aim was to develop a survival prediction model for such patients undergoing TACE using novel machine learning algorithms and to compare it to conventional prediction scores, ART, ABCR and SNACOR.

METHODS

For this retrospective analysis, 282 patients who underwent TACE for HCC at our tertiary referral centre between January 2005 and December 2017 were included in the final analysis. We built an artificial neural network (ANN) including all parameters used by the aforementioned risk scores and other clinically meaningful parameters. Following an 80:20 split, the first 225 patients were used for training; the more recently treated 20% were used for validation.

RESULTS

The ANN had a promising performance at predicting 1-year survival, with an area under the ROC curve (AUC) of 0.77 ± 0.13. Internal validation yielded an AUC of 0.83 ± 0.06, a positive predictive value of 87.5% and a negative predictive value of 68.0%. The sensitivity was 77.8% and specificity 81.0%. In a head-to-head comparison, the ANN outperformed the aforementioned scoring systems, which yielded lower AUCs (SNACOR 0.73 ± 0.07, ABCR 0.70 ± 0.07 and ART 0.54 ± 0.08). This difference reached significance for ART (P < .001); for ABCR and SNACOR significance was not reached (P = .143 and P = .201).

CONCLUSIONS

Artificial neural networks could be better at predicting patient survival after TACE for HCC than traditional scoring systems. Once established, such prediction models could easily be deployed in clinical routine and help determine optimal patient care.

Predictive models for patients with lung carcinomas to identify EGFR mutation status via an artificial neural network based on multiple clinical information

PURPOSE

Epidermal growth factor receptor (EGFR) mutation testing has several limitations. Therefore, we built predictive models to determine the EGFR mutation status of patients and guide therapeutic decision-making.

METHODS

We collected data from 320 patients with lung carcinoma, including sex, age, smoking history, serum tumour marker levels, maximum standardized uptake value, pathological results, computed tomography images, and EGFR mutation status. Artificial neural network (ANN) models based on multiple clinical characteristics were proposed to predict EGFR mutation status.

RESULTS

A training set (n = 200) was used to develop predictive models of the EGFR mutation status (Model 1: area under the receiver operating characteristic curve [AUROC] = 0.910, 95% CI 0.861-0.945; Model 2: AUROC = 0.859, 95% CI 0.803-0.904; Model 3: AUROC = 0.711, 95% CI 0.643-0.773). A testing set (n = 50) and temporal validation data set (n = 70) were used to evaluate the generalisation performance of the established models (testing set: Model 1, AUROC = 0.845, 95% CI 0.715-0.932; Model 2, AUROC = 0.882, 95% CI 0.759-0.956; Model 3, AUROC = 0.817, 95% CI 0.682-0.912; temporal validation dataset: Model 1, AUROC = 0.909, 95% CI 0.816-0.964; Model 2, AUROC = 0.855, 95% CI 0.751-0.928; Model 3, AUROC = 0.831, 95% CI 0.723-0.910). The predictive abilities of the three ANN models were superior to that of a previous logistic regression model (P < 0.001, 0.027, and 0.050, respectively).

CONCLUSIONS

ANN models provide a non-invasive and readily available method for EGFR mutation status prediction.

Correlation Analysis Between Preoperative Serum Iron Level and Prognosis as Well as Recurrence of HCC After Radical Resection

Background

The purpose of this retrospective study was to investigate the relationship between serum iron levels and the prognosis and risk of recurrence in patients with hepatocellular carcinoma (HCC).

Methods

A total of 253 HCC patients who underwent radical resection were involved in this study.

Results

According to the receiver operating characteristic (ROC) curve, the optimal cut-off value for preoperative serum iron in the assessment of HCC postoperative prognosis was 94 ug/dL. The overall survival (OS) of patients in the high iron group was significantly better than that in the low iron group (p < 0.001). The recurrence rate of patients in the low iron group was higher than that in the high iron group (p = 0.011). Correlation analysis showed that preoperative serum iron level was correlated with tumor size >5 cm (χ ² = 11.590, p < 0.001), recurrence (χ ² = 5.714, p = 0.017) and microvascular invasion (χ ² = 5.087, p = 0.024). In addition, univariate analysis showed that OS and disease-free survival (DFS) of HCC patients with high iron level were better than those with low iron level. Furthermore, multivariate COX proportional hazards regression analysis showed that serum iron ≤94 μg/dL, tumor size >5 cm, and microvascular invasion were independent predictors for shorter OS and DFS in HCC patients after operation, while recurrence was for shorter OS.

Conclusion

Patients with low preoperative serum iron level had worse postoperative survival and higher recurrence rate in HCC. Preoperative serum iron is an independent predictor of HCC patients. For HCC patients with low iron levels, prognosis of patients may be improved if appropriate iron is supplemented.

Nomogram to Assist in Surgical Plan for Hepatocellular Carcinoma: a Prediction Model for Microvascular Invasion

BACKGROUND

Microvascular invasion (MVI) relates to poor survival in hepatocellular carcinoma (HCC) patients. In this study, we aim at developing a nomogram for MVI prediction and potential assistance in surgical planning.

METHODS

A total of 357 patients were assigned to training (n = 257) and validation (n = 100) cohort. Univariate and multivariate analyses were used to reveal preoperative predictors for MVI. A nomogram incorporating independent predictors was constructed and validated. Disease-free survival was compared between patients, and the potential of the predicted MVI in making surgical procedure was also explored.

RESULTS

Pathological examination confirmed MVI in 140 (39.2%) patients. Imaging features including larger tumor, intra-tumoral artery, tumor type, and higher serum AFP independently correlated with MVI. The nomogram showed desirable performance with an AUROC of 0.803 (95% CI, 0.746-0.860) and 0.814 (95% CI, 0.720-0.908) in the training and validation cohorts, respectively. Good calibration were also revealed by calibration curve in both cohorts. The decision curve analysis indicated that the prediction nomogram was of promising usefulness in clinical work. In addition, survival analysis revealed that patients with positive-predicted MVI suffered a higher risk of early recurrence (P < 0.01). There was no difference in disease-free survival between anatomic or non-anatomic resection in large HCC or small HCC without nomogram-predicted MVI. However, anatomic resection improved disease-free survival in small HCC with nomogram-predicted MVI.

CONCLUSIONS

The nomogram obtained desirable results in predicting MVI. Patients with predicted MVI were associated with early recurrence and anatomic resection was recommended for small HCC patients with predicted MVI.

Integrated Nomograms for Preoperative Prediction of Microvascular Invasion and Lymph Node Metastasis Risk in Hepatocellular Carcinoma Patients

BACKGROUND

The aim of the present work is to develop and validate accurate preoperative nomograms to predict microvascular invasion (MVI) and lymph node metastasis (LNM) in hepatocellular carcinoma.

PATIENTS AND METHODS

A total of 268 patients with resected hepatocellular carcinoma (HCC) were divided into a training set (n = 180), in an earlier period, and a validation set (n = 88), thereafter. Risk factors for MVI and LNM were assessed based on logistic regression. Blood signatures were established using the least absolute shrinkage and selection operator algorithm. Nomograms were constructed by combining risk factors and blood signatures. Performance was evaluated using the training set and validated using the validation set. The clinical values of the nomograms were measured by decision curve analysis.

RESULTS

The risk factors for MVI were hepatitis B virus (HBV) DNA loading, portal hypertension, Barcelona liver clinic (BCLC) stage, and three computerized tomography (CT) imaging features, namely tumor number, size, and encapsulation, while only BCLC stage, Child-Pugh classification, and tumor encapsulation were associated with LNM. The nomogram incorporating both risk factors and blood signatures achieved better performance in predicting MVI in the training and validation sets (C-indexes of 0.828 and 0.804) than the LNM nomogram (C-indexes of 0.765 and 0.717). Calibration curves also demonstrated a good fit. The decision curves indicate significant clinical usefulness.

CONCLUSIONS

The novel validated nomograms for HCC patients presented herein are noninvasive preoperative tools that can effectively predict the individualized risk of MVI and LNM, and this predictive power can aid doctors in explaining the illness for patient counseling.

Application of preoperative artificial neural network based on blood biomarkers and clinicopathological parameters for predicting long-term survival of patients with gastric cancer

BACKGROUND

Because of the powerful abilities of self-learning and handling complex biological information, artificial neural network (ANN) models have been widely applied to disease diagnosis, imaging analysis, and prognosis prediction. However, there has been no trained preoperative ANN (preope-ANN) model to preoperatively predict the prognosis of patients with gastric cancer (GC).

AIM

To establish a neural network model that can predict long-term survival of GC patients before surgery to evaluate the tumor condition before the operation.

METHODS

The clinicopathological data of 1608 GC patients treated from January 2011 to April 2015 at the Department of Gastric Surgery, Fujian Medical University Union Hospital were analyzed retrospectively. The patients were randomly divided into a training set (70%) for establishing a preope-ANN model and a testing set (30%). The prognostic evaluation ability of the preope-ANN model was compared with that of the American Joint Commission on Cancer (8th edition) clinical TNM (cTNM) and pathological TNM (pTNM) staging through the receiver operating characteristic curve, Akaike information criterion index, Harrell's C index, and likelihood ratio chi-square.

RESULTS

We used the variables that were statistically significant factors for the 3-year overall survival as input-layer variables to develop a preope-ANN in the training set. The survival curves within each score of the preope-ANN had good discrimination (P < 0.05). Comparing the preope-ANN model, cTNM, and pTNM in both the training and testing sets, the preope-ANN model was superior to cTNM in predictive discrimination (C index), predictive homogeneity (likelihood ratio chi-square), and prediction accuracy (area under the curve). The prediction efficiency of the preope-ANN model is similar to that of pTNM.

CONCLUSION

The preope-ANN model can accurately predict the long-term survival of GC patients, and its predictive efficiency is not inferior to that of pTNM stage.

Serum milk fat globule-EGF factor 8 (MFG-E8) as a diagnostic and prognostic biomarker in patients with hepatocellular carcinoma

Current serum hepatocellular carcinoma (HCC) biomarkers are insufficient for early diagnosis. We aimed to clarify whether serum MFG-E8 can serve as a diagnostic or prognostic biomarker of HCC. Serum MFG-E8 levels of 282 HCC patients, who underwent primary hepatectomy, were examined by ELISA. We also quantified serum MFG-E8 levels in patients with chronic hepatitis (CH), liver cirrhosis (LC), as well as in healthy volunteers (HVs). Serum MFG-E8 levels were significantly lower in HCC patients than in HVs regardless of the etiology of liver disease (3.6 ± 0.1 vs 5.8 ± 0.2 ng/mL, p < 0.0001), and recovered after treatment of HCC. Serum MFG-E8 levels in CH and LC patients were comparable to those in HVs. Serum MFG-E8 could detect HCCs, even α-fetoprotein (AFP)-negative or des-γ-carboxy prothrombin (DCP)-negative HCCs, in CH and LC patients. Our new HCC prediction model using MFG-E8 and DCP (Logit(p) = 2.619 − 0.809 × serum MFG-E8 + 0.0226 × serum DCP) distinguished HCC patients from CH and LC patients with an area under the curve of 0.923, a sensitivity of 81.1%, and a specificity of 89.8%. Futhermore, low preoperative serum MFG-E8 was an independent predictor of poor overall survival. Thus, serum MFG-E8 could serve as a feasible diagnostic and prognostic biomarker for HCC.

Nomogram For Preoperative Prediction Of Microvascular Invasion Risk In Hepatocellular Carcinoma

Objective

To preoperatively predict the microvascular invasion (MVI) risk in hepatocellular carcinoma (HCC) using nomogram.

Methods

A retrospective cohort of 513 patients with HCC hospitalized at Xiangya Hospital between January 2014 and December 2018 was included in the study. Univariate and multivariate analysis was performed to identify the independent risk factors for MVI. Based on the independent risk factors, nomogram was established to preoperatively predict the MVI risk in HCC. The accuracy of nomogram was evaluated by using receiver operating characteristic (ROC) curve, calibration curve and decision curve analysis (DCA).

Results

Tumor size (OR=1.17, 95% CI: 1.11–1.23, p<0.001), preoperative AFP level greater than 155 ng/mL (OR=1.65, 95% CI: 1.13–2.39, p=0.008) and NLR (OR=1.14, 95% CI: 1.00–1.29, p=0.042) were the independent risk factors for MVI. Incorporating these 3 factors, nomogram was established with the concordance index of 0.71 (95% CI, 0.66–0.75) and well-fitted calibration curves. DCA confirmed that using this nomogram added more benefit compared with the measures that treat all patients or treat none patients. At the cutoff value of predicted probability ≥0.44, the model demonstrated sensitivity of 61.64%, specificity of 71.53%, positive predictive value (PPV) of 64.13%, and negative predictive value (NPV) of 69.31%.

Conclusion

Nomogram was established for preoperative prediction of the MVI risk in HCC patients, and better therapeutic choice will be made if it was applied in clinical practice.

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

BACKGROUND

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).

METHODS

Two hundred and eight patients with pathologically confirmed HCC (training cohort: n = 146; validation cohort: n = 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.

RESULTS

Higher α-fetoprotein level (p = 0.046), nonsmooth tumor margin (p = 0.003), arterial peritumoral enhancement (p < 0.001), and the radiomics signatures of hepatobiliary phase (HBP) T1-weighted images (p < 0.001) and HBP T1 maps (p < 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; p = 0.002), though the validation did not have a statistical significance (AUCs 0.861 vs. 0.759; p = 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 (p > 0.05). Decision curve analysis further confirmed the clinical usefulness of the nomogram.

CONCLUSIONS

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.

Prognostic Value and Predication Model of Microvascular Invasion in Patients with Intrahepatic Cholangiocarcinoma

Background: Whether microvascular invasion (MVI) adversely influences oncological outcomes for intrahepatic cholangiocarcinoma (ICC) patients remains unclear. The purpose of this study was to determine the impact of MVI on postoperative survival and establish a new predictive model for MVI before surgical intervention in patients with ICC.

Methods: In this two-center retrospective study, 556 and 31 consecutive patients who underwent curative liver resection for ICC at ZSH and XJFH were analyzed, respectively. Propensity score matching (PSM) and Cox regression analyses were used to explore the prognostic role of MVI on the OS and DFS. Multivariate logistic regression was used to identify the relative risk factors of MVI, which were incorporated into the nomogram.

Results: After PSM, 50 MVI cases matched with 172 non-MVI cases, and no bias was observed between the two groups (propensity score, 0.118 (0.099, 0.203) vs. 0.115 (0.059, 0.174), p=0.251). The multivariate Cox analysis showed that MVI was negatively associated with OS (HR 1.635, 95% CI 1.405-1.993, p=0.04) and DFS (HR 1.596, 95% CI 1.077-2.366, p=0.02). The independent factors associated with MVI were ALT, AFP, tumor maximal diameter, and tumor capsule. The nomogram that incorporated these variables achieved good concordance indexes for predicting MVI. Patients with a cutoff score of 168 were considered to have different risks of the presence of MVI preoperatively.

Conclusions: The presence of MVI was an adverse prognostic factor for ICC patients. Using the nomogram model, the risk of an individual patient harboring MVI was determined, which led to a rational therapeutic choice.

Preoperative prediction of microvascular invasion in hepatocellular cancer: a radiomics model using Gd-EOB-DTPA-enhanced MRI

OBJECTIVES

Preoperative prediction of microvascular invasion (MVI) in patients with hepatocellular cancer (HCC) is important for surgery strategy making. We aimed to develop and validate a combined intratumoural and peritumoural radiomics model based on gadolinium-ethoxybenzyl-diethylenetriamine (Gd-EOB-DTPA)-enhanced magnetic resonance imaging (MRI) for preoperative prediction of MVI in primary HCC patients.

METHODS

This study included a training cohort of 110 HCC patients and a validating cohort of 50 HCC patients. All the patients underwent preoperative Gd-EOB-DTPA-enhanced MRI examination and curative hepatectomy. The volumes of interest (VOIs) around the hepatic lesions including intratumoural and peritumoural regions were manually delineated in the hepatobiliary phase of MRI images, from which quantitative features were extracted and analysed. In the training cohort, machine-learning method was applied for dimensionality reduction and selection of the extracted features.

RESULTS

The proportion of MVI-positive patients was 38.2% and 40.0% in the training and validation cohort, respectively. Supervised machine learning selected ten features to establish a predictive model for MVI. The area under the receiver operating characteristic curve (AUC), sensitivity, specificity of the combined intratumoural and peritumoural radiomics model in the training and validation cohort were 0.85 (95% confidence interval (CI), 0.77-0.93), 88.2%, 76.2%, and 0.83 (95% CI, 0.71-0.95), 90.0%, 75.0%, respectively.

CONCLUSIONS

We evaluate quantitative Gd-EOB-DTPA-enhanced MRI image features of both intratumoural and peritumoural regions and provide an effective radiomics-based model for the prediction of MVI in HCC patients, and may therefore help clinicians make precise decisions regarding treatment before the surgery.

KEY POINTS

• An effective radiomics model for prediction of microvascular invasion in HCC patients is established. • The radiomics model is superior to the radiologist in prediction of MVI. • The radiomics model can help clinicians in pretreatment decision making.

The impact of resection margin and microvascular invasion on long-term prognosis after curative resection of hepatocellular carcinoma: a multi-institutional study

BACKGROUND

The resection margin (RM) status and microscopic vascular invasion (MVI) are known prognostic factors for hepatocellular carcinoma (HCC). An enhanced understanding of their impact on long-term prognosis is required to improve oncological outcomes.

METHODS

Using multi-institutional data, the different impact of the RM status (narrow, <1 cm, or wide, ≥1 cm) and MVI (positive or negative) on overall survival (OS) and recurrence-free survival (RFS) after curative liver resection of solitary HCC without macrovascular invasion was analyzed.

RESULTS

In 801 patients, 306 (38%) had a narrow RM and 352 (44%) had positive MVI. The median OS and RFS were 109.8 and 74.8 months in patients with wide RM & negative MVI, 93.5 and 53.1 months with wide RM & positive MVI, 79.2 and 41.6 months with narrow RM & negative MVI, and 69.2 and 37.5 months with narrow RM & positive MVI (both P < 0.01). On multivariable analyses, narrow RM & positive MVI had the highest hazard ratio with reduced OS and RFS (HR 2.96, 95% CI 2.11-4.17, and HR 3.15, 95% CI, 2.09-4.67, respectively).

CONCLUSIONS

Concomitant having narrow RM and positive MVI increases the risks of postoperative death and recurrence by about 2-fold in patients with solitary HCC.

Safety and Efficacy of Radiofrequency Ablation for Solitary Hepatocellular Carcinoma (3-5 cm): a Propensity Score Matching Cohort Study

BACKGROUND

We aim to investigate the safety and efficacy of radiofrequency ablation in the treatment of solitary hepatocellular carcinoma (3-5 cm) in comparison with surgical resection.

METHODS

Included in this study were 388 patients with primary and solitary hepatocellular carcinoma, of whom 196 patients underwent surgical resection and the other 192 patients received radiofrequency ablation. Clinicopathological characteristics, prognosis, post-treatment complications, hospital stay, and financial expenditures between the two groups were compared retrospectively.

RESULTS

The result of propensity score matching and subgroup analysis showed that the 1-, 3-, and 5-year overall survival and disease-free survival were comparable in patients with tumors of 3-4 cm in diameter between surgical resection and radiofrequency ablation groups. However, when the tumor size exceeded 4 cm in diameter, surgical resection exhibited a superior long-term prognosis compared with radiofrequency ablation. Nevertheless, hepatectomy was associated with high occurrences of postoperative complications, long hospital stay, and high hospitalization cost as compared with radiofrequency ablation. Further analysis of the relationship between tumor size and pathological features of hepatocellular carcinoma showed that tumors larger than 4 cm were positively correlated with a high rate of microvascular invasion and satellite nodule formation.

CONCLUSION

For solitary hepatocellular carcinoma of 3-4 cm in diameter, radiofrequency ablation could achieve a comparable prognosis with a low incidence of post-treatment complications and low hospitalization costs, while surgical resection is recommended for solitary hepatocellular carcinoma tumors of 4-5 cm in diameter when long-term prognosis is considered.

Signatures of gene expression, DNA methylation and microRNAs of hepatocellular carcinoma with vascular invasion

Aim: Micro and macro vascular invasion (VI) are known as independent predictors of tumor recurrence and poor survival after surgical treatment of hepatocellular carcinoma (HCC). Here, we aimed to re-analyze The Cancer Genome Atlas of liver hepatocellular carcinoma datasets to identify the VI-expression signatures. Materials & methods: We filtered The Cancer Genome Atlas liver hepatocellular carcinoma (LIHC) datasets into three groups: no VI (NVI = 198); micro VI (MIVI = 89) and macro VI (MAVI = 16). We performed differential gene expression, methylation and microRNA analyses. Results & conclusion: We identified 12 differentially expressed genes and 55 differentially methylated genes in MAVI compared with no VI. The GPD1L gene appeared in all of the comparative analyses. Higher GPD1L expression was associated with VI and poor outcomes in the HCC patients.

Grading of hepatocellular carcinoma based on diffusion weighted images with multiple b-values using convolutional neural networks

PURPOSE

To effectively grade hepatocellular carcinoma (HCC) based on deep features derived from diffusion weighted images (DWI) with multiple b-values using convolutional neural networks (CNN).

MATERIALS AND METHODS

Ninety-eight subjects with 100 pathologically confirmed HCC lesions from July 2012 to October 2018 were included in this retrospective study, including 47 low-grade and 53 high-grade HCCs. DWI was performed for each subject with a 3.0T MR scanner in a breath-hold routine with three b-values (0,100, and 600 s/mm² ). First, logarithmic transformation was performed on original DWI images to generate log maps (logb0, logb100, and logb600). Then, a resampling method was performed to extract multiple 2D axial planes of HCCs from the log map to increase the dataset for training. Subsequently, 2D CNN was used to extract deep features of the log map for HCCs. Finally, fusion of deep features derived from three b-value log maps was conducted for HCC malignancy classification. Specifically, a deeply supervised loss function was devised to further improve the performance of lesion characterization. The data set was split into two parts: the training and validation set (60 HCCs) and the fixed test set (40 HCCs). Four-fold cross validation with 10 repetitions was performed to assess the performance of deep features extracted from single b-value images for HCC grading using the training and validation set. Receiver operating characteristic curve (ROC) and area under the curve (AUC) values were used to assess the characterization performance of the proposed deep feature fusion method to differentiate low-grade and high-grade in the fixed test set.

RESULTS

The proposed fusion of deep features derived from logb0, logb100, and logb600 with deeply supervised loss function generated the highest accuracy for HCC grading (80%), thus outperforming the method of deep feature derived from the ADC map directly (72.5%), the original b0 (65%), b100 (68%), and b600 (70%) images. Furthermore, AUC values of the deep features of the ADC map, the deep feature fusion with concatenation, and the proposed deep feature fusion with deeply supervised loss function were 0.73, 0.78, and 0.83, respectively.

CONCLUSION

The proposed fusion of deep features derived from the logarithm of the three b-value images yields high performance for HCC grading, thus providing a promising approach for the assessment of DWI in lesion characterization.

Liver Imaging Reporting and Data System Category 5: MRI Predictors of Microvascular Invasion and Recurrence After Hepatectomy for Hepatocellular Carcinoma

OBJECTIVE. We investigated in Liver Imaging Reporting and Data System category 5 (LR-5) observations whether imaging features, including LI-RADS imaging features, could predict microvascular invasion (MVI) and posthepatectomy recurrence in high-risk adult patients with hepatocellular carcinoma (HCC). MATERIALS AND METHODS. We retrospectively identified 149 high-risk patients who underwent 3-T MRI within 1 month before hepatectomy for HCC; 81 of 149 patients with no HCC recurrence were followed for more than 1 year. Tumors with clear surgical margins were confirmed in each hepatectomy specimen. MVI was evaluated histologically by a histopathologist. Tumor recurrence was determined by clinical and imaging follow-up. Two independent radiologists reviewed the prehepatectomy MR images and assessed LI-RADS v2018 imaging features as well as some non-LI-RADS features in all LR-5 observations in consensus. Alpha-fetoprotein level, tumor number, and imaging features were analyzed as potential predictors for MVI and posthepatectomy recurrence using multivariate logistic regression and Cox proportional hazards models. RESULTS. One hundred forty-nine patients with pathologically confirmed HCC were included; 64 of 149 (43.0%) patients had MVI, whereas 48 of 129 (37.2%) patients had tumor recurrence within 3 years after hepatectomy. Mosaic architecture (odds ratio, 3.420; p < 0.001) and nonsmooth tumor margin (odds ratio, 2.554; p = 0.011) were independent predictors of MVI. Multifocal tumors (hazard ratio, 2.101; p = 0.034), absence of fat in mass (hazard ratio, 2.109; p = 0.015), and nonsmooth tumor margin (hazard ratio, 2.415; p = 0.005) were independent predictors of posthepatectomy recurrence. CONCLUSION. In high-risk patients with LR-5 HCC, mosaic architecture and non-smooth tumor margin independently predicted MVI. Multifocal tumors, absence of fat in mass, and nonsmooth tumor margin independently predicted recurrence.

Nomogram and Artificial Neural Network for Prognostic Performance on the Albumin-Bilirubin Grade for Hepatocellular Carcinoma Undergoing Transarterial Chemoembolization

PURPOSE

To construct the albumin-bilirubin (ALBI) grade and the Child-Turcotte-Pugh (CTP) score based on nomograms, as well as to develop an artificial neural network (ANN) to compare the prognostic performance of the 2 scores for hepatocellular carcinoma (HCC) that has undergone transarterial chemoembolization.

MATERIALS AND METHODS

This multicentric retrospective study included patients with HCC who underwent transarterial chemoembolization monotherapy as an initial treatment at 4 institutions between January 2008 and December 2016. In the training cohort, significant risk factors associated with overall survival (OS) were identified by univariate and multivariate analyses. The prognostic nomograms and ANN were established and then validated in 2 validation cohorts.

RESULTS

A total of 838 patients (548, 115, and 175 in the training cohort and validation cohorts 1 and 2, respectively) were included. The median OS was 10.4, 15.7, and 9.2 months in the training cohort and validation cohorts 1 and 2, respectively. In the training cohort, both ALBI grade and CTP score were identified as significant risk factors. The ALBI grade and CTP score based on nomograms were established separately and showed similar prognostic performance when assessed externally in validation cohorts (C-index in validation cohort 1: 0.823 vs 0.802, P = .417; in validation cohort 2: 0.716 vs 0.729, P = .793). ANN showed that ALBI grade had higher importance on survival prediction than CTP score.

CONCLUSIONS

ALBI grade performs at least no worse than CTP score regarding survival prediction for HCC receiving transarterial chemoembolization. Considering the easy application, ALBI grade has the potential to be regarded as an alternative to CTP score.

Prognostic Value and Prediction of Extratumoral Microvascular Invasion for Hepatocellular Carcinoma

BACKGROUND

There are few reports on microvascular invasion (MVI) located intra- or extratumorally and prognosis of hepatocellular carcinoma (HCC).

OBJECTIVE

The aim of this study was to evaluate patient outcome according to the location of MVI, and to build a nomogram predicting extratumoral MVI.

METHODS

We included 681 consecutive patients who underwent hepatic resection (HR) or liver transplantation (LT) for HCC from January 1994 to June 2012, and evaluated patient outcome according to the degree of vascular invasion (VI). A nomogram for predicting extratumoral MVI was created using 637 patients, excluding 44 patients with macrovascular invasion, and was validated using an internal (n = 273) and external patient cohort (n = 256).

RESULTS

The 681 patients were classified into four groups based on pathological examination (148 no VI, 33 intratumoral MVI, 84 extratumoral MVI, and 29 macrovascular invasion in patients who underwent HR; 238 no VI, 50 intratumoral MVI, 84 extratumoral MVI, and 15 macrovascular invasion in patients who underwent LT). Multivariate analysis revealed that extratumoral MVI was an independent risk factor for overall survival in patients who underwent HR (hazard ratio 2.62, p < 0.0001) or LT (hazard ratio 1.99, p = 0.0005). Multivariate logistic regression analysis identified six independent risk factors for extratumoral MVI: α-fetoprotein, tumor size, non-boundary type, alkaline phosphatase, neutrophil-to-lymphocyte ratio, and aspartate aminotransferase. The nomogram for predicting extratumoral MVI using these factors showed good concordance indices of 0.774 and 0.744 in the internal and external validation cohorts, respectively.

CONCLUSIONS

The prognostic value of MVI differs according to its invasiveness. The nomogram allows reliable prediction of extratumoral MVI in patients undergoing HR or LT.

Prognostic and Therapeutic Implications of Microvascular Invasion in Hepatocellular Carcinoma

Hepatocellular carcinoma (HCC) is a morbid condition for which surgical and ablative therapy are the only options for cure. Nonetheless, over half of patients treated with an R0 resection will develop recurrence. Early recurrences within 2 years after resection are thought to be due to the presence of residual microscopic disease, while late recurrences > 2 years after resection are thought to be de novo metachronous HCCs arising in chronically injured liver tissue. Microvascular invasion (MVI) is defined as the presence of micrometastatic HCC emboli within the vessels of the liver, and is a critical determinant of early recurrence and survival. In this review, we summarize the pathogenesis and clinical relevance of MVI, which correlates with adverse biological features, including high grade, large tumor size, and epithelial-mesenchymal transition. Multiple classification schemas have been proposed to capture the heterogeneous features of MVI that are associated with prognosis. However, currently, MVI can only be determined based on surgical specimens, limiting its clinical applicability. Going forward, advances in axial imaging technologies, molecular characterization of biopsy tissue, and novel serum biomarkers hold promise as future methods for non-invasive MVI detection. Ultimately, MVI status may be used to help clinicians determine treatment plans, particularly with respect to surgical intervention, and to provide more accurate prognostication.

Predictive model for microvascular invasion of hepatocellular carcinoma among candidates for either hepatic resection or liver transplantation

BACKGROUND

Microvascular invasion is the strongest prognostic factor of survival in patients with hepatocellular carcinoma. We therefore developed a predictive model for microvascular invasion of hepatocellular carcinoma to help guide treatment strategies in patients scheduled for either hepatic resection or liver transplantation.

METHODS

Patients with hepatocellular carcinoma who underwent hepatic resection or liver transplantation from 1994 to 2016 were divided into training and validation cohorts. A predictive model for microvascular invasion was developed based on microvascular invasion risk factors in the training cohort and validated in the validation cohort.

RESULTS

A total of 910 patients (425 having received hepatic resection, 485 having received liver transplantation) were included in the training (n = 637) and validation (n = 273) cohorts. Multivariate analysis identified α-fetoprotein ≥100 ng/mL (relative risk 3.05, P < .0001), tumor size ≥40 mm (relative risk 1.98, P = .0002), nonboundary hepatocellular carcinoma type (relative risk 1.91, P = .001), neutrophil-to-lymphocyte ratio (relative risk 1.86, P = .002), and aspartate aminotransferase (relative risk 1.53, P = .02) as associated with microvascular invasion. The estimated probability of microvascular invasion ranged from 17.0% in patients with none of these factors to 86.9% in the presence of all factors. This model achieved a C-index of 0.732 in the validation cohort. The 5-year overall survival of patients with ≥50% probability of microvascular invasion was poorer than that of patients with <50% probability (hepatic resection; 39.1% vs 61.2%, P < .0001, liver transplantation; 5-year overall survival, 54.8% vs 79.0%, P = .05).

CONCLUSION

This model developed from preoperative data allows reliable prediction of microvascular invasion in candidates for either hepatic resection or liver transplantation.

A wide-margin liver resection improves long-term outcomes for patients with HBV-related hepatocellular carcinoma with microvascular invasion

BACKGROUND

The impact of the resection margin on survival outcomes in patients with hepatocellular carcinoma remains to be determined. This study aimed to examine the association between the width of resection margin and the presence of microvascular invasion in hepatitis B virus-related hepatocellular carcinoma.

METHODS

We reviewed data on 2,508 consecutive patients who underwent liver resection for a solitary, hepatitis B virus-related hepatocellular carcinoma for operative morbidity, tumor recurrence, and overall survival.

RESULTS

Microvascular invasion was identified histologically in 929 patients (37.0%). A wide margin of resection (≥1 cm, n = 384) resulted in better 5-year recurrence and overall survival versus a narrow margin of resection (<1 cm, n = 545) among patients with microvascular invasion (71.1% versus 85.9%; 44.9% versus 25.0%; both P < .001), but not in patients without microvascular invasion (P = .131, .182). Similar results were identified after propensity-score matching. A wide margin resection also had a lesser incidence of early recurrence developed within the first postoperative 24 months (58.1% versus 72.7%; P < .001). Compared with a wide resection margin, a narrow margin was associated with worse recurrence and overall survival in patients with microvascular invasion (hazard ratio: 1.50 and 1.75). In addition, a wide or a narrow resection margin had differences in the rate of grade I-III, but not grade IV complications (31.0% versus 21.7%; P = .017; 3.5% versus 1.6%; P = .147) among cirrhotic patients with microvascular invasion.

CONCLUSION

The presence of microvascular invasion was associated with a worse prognosis after resection. A wide resection margin resulted in better long-term prognoses versus a narrow resection margin among patients with hepatitis B virus-related hepatocellular carcinoma with microvascular invasion.

Nomograms based on inflammatory biomarkers for predicting tumor grade and micro-vascular invasion in stage I/II hepatocellular carcinoma

Background: Increasing evidences reveal that inflammation plays a critical role in tumorigenesis and progression. We aimed to develop the nomograms based on inflammatory biomarkers to predict micro-vascular invasion (MVI) and tumor grade in stage I/II hepatocellular carcinoma (HCC).Methods: A retrospective cohort of 627 patients with stage I/II HCC between January 2007 and December 2014 was included in the study. Logistic regression was performed to identify the independent risk factors of tumor grade and MVI. The significant predictors including neutrophil-to-lymphocyte ratio (NLR), derived neutrophil-to-lymphocyte ratio (dNLR), lymphocyte-to-monocyte ratio (LMR), tumor volume age, and tumor size were subsequently incorporated to build the nomograms. The prediction accuracies of the nomograms were evaluated using the area under the receiver operating characteristic (ROC) curve.Results: The independent risk factors for tumor grade were NLR, dNLR, and tumor volume (P<0.001, P=0.001, and P<0.001, respectively), which were assembled into tumor grade nomogram. MVI nomogram was developed by dNLR, LMR, age, and tumor size (P<0.001, P<0.001, P<0.001, and P=0.001, respectively) which were the independent predictors for MVI. The area under the ROC curve of nomograms for predicting tumor grade and MVI were 0.727 (95% confidence intervals [CI]: 0.690-0.761) and 0.839 (95% CI: 0.808-0.867), respectively. Patients who had a nomogram score of less than 100 and 79 were considered to have high possibility of moderate grade and have low risks of MVI presence, respectively.Conclusion: We successfully developed nomograms predicting tumor grade and MVI based on inflammatory biomarkers with high accuracy, leading to a rational therapeutic choice for stage I/II HCC.

The Predictors of Microscopic Vessel Invasion Differ Between Primary Hepatocellular Carcinoma and Hepatocellular Carcinoma with a Treatment History

BACKGROUND AND AIM

Previous studies have shown that microscopic vessel invasion (MVI) occurs in hepatocellular carcinoma (HCC) with a treatment history due to its poorer malignant behavior in comparison with primary HCC. The aim of the present study was to determine the predictors of MVI and overall survival in HCC patients with a treatment history.

METHODS

This retrospective study included 580 patients who underwent hepatectomy and whose preoperative imaging showed no evidence of macroscopic vessel invasion. The patients were classified into two groups: primary HCC (n = 425) and HCC with a treatment history (n = 155). MVI was defined as the presence of either microscopic portal vein invasion or venous invasion, which was invisible on preoperative imaging.

RESULTS

MVI was identified in 34 (21.9%) patients with a treatment history. A multivariate analysis showed that a high des-gamma-carboxy prothrombin (odds ratio [OR] 5.16, P = 0.002) and a large tumor diameter (OR 2.57, P = 0.030) were the significant predictor of MVI in HCC with a treatment history. Moreover, the presence of MVI (hazard ratio [HR] 2.27, P = 0.001) and tumor diameter >27 mm (HR 2.04, P = 0.006) remained significant predictors of the overall survival in HCC with a treatment history. The tumor diameter cutoff value for predicting MVI (27 mm) in HCC with a treatment history was smaller than in primary HCC (37 mm).

CONCLUSIONS

The presence of MVI was a significant predictor in the HCC patients with a treatment history. The tumor diameter is an important factor that can be used to predict the presence of MVI, especially in HCC with a treatment history.

Combined Volumetric and Density Analyses of Contrast-Enhanced CT Imaging to Assess Drug Therapy Response in Gastroenteropancreatic Neuroendocrine Diffuse Liver Metastasis

OBJECTIVE

We propose a computer-aided method to assess response to drug treatment, using CT imaging-based volumetric and density measures in patients with gastroenteropancreatic neuroendocrine tumors (GEP-NETs) and diffuse liver metastases.

METHODS

Twenty-five patients with GEP-NETs with diffuse liver metastases were enrolled. Pre- and posttreatment CT examinations were retrospectively analyzed. Total tumor volume (volume) and mean volumetric tumor density (density) were calculated based on tumor segmentation on CT images. The maximum axial diameter (tumor size) for each target tumor was measured on pre- and posttreatment CT images according to Response Evaluation Criteria In Solid Tumors (RECIST). Progression-free survival (PFS) for each patient was measured and recorded.

RESULTS

Correlation analysis showed inverse correlation between change of volume and density (Δ(V + D)), change of volume (ΔV), and change of tumor size (ΔS) with PFS (r = -0.653, P=0.001; r = -0.617, P=0.003; r = -0.548, P=0.01, respectively). There was no linear correlation between ΔD and PFS (r = -0.226, P=0.325).

CONCLUSION

The changes of volume and density derived from CT images of all lesions showed a good correlation with PFS and may help assess treatment response.