Predicting 5-year recurrence risk in colorectal cancer: development and validation of a histology-based deep learning approach.

BACKGROUND: Accurate estimation of the long-term risk of recurrence in patients with non-metastatic colorectal cancer (CRC) is crucial for clinical management. Histology-based deep learning is expected to provide more abundant information for risk stratification. METHODS: We developed and validated a weakly supervised deep-learning model for predicting 5-year relapse-free survival (RFS) to stratify patients with different risks based on histological images from three hospitals of 614 cases with non-metastatic CRC. A deep prognostic factor (DL-RRS) was established to stratify patients into high and low-risk group. The areas under the curve (AUCs) were calculated to evaluate the performances of models. RESULTS: Our proposed model achieves the AUCs of 0.833 (95% CI: 0.736-0.905) and 0.715 (95% CI: 0.647-0.776) on validation cohort and external test cohort, respectively. The 5-year RFS rate was 45.7% for high DL-RRS patients, and 82.5% for low DL-RRS patients respectively in the external test cohort (HR: 3.89, 95% CI: 2.51-6.03, P < 0.001). Adjuvant chemotherapy was associated with improved RFS in Stage II patients with high DL-RRS (HR: 0.15, 95% CI: 0.06-0.38, P < 0.001). CONCLUSIONS: DL-RRS has a good predictive performance of 5-year recurrence risk in CRC, and will better serve the clinical decision-making.

FASN-mediated fatty acid biosynthesis remodels immune environment in Clonorchis sinensis infection-related intrahepatic cholangiocarcinoma.

BACKGROUND & AIMS: Intrahepatic cholangiocarcinoma (iCCA) is the second most common primary liver cancer and is highly lethal. Clonorchis sinensis (C. sinensis) infection is an important risk factor for iCCA. Here we investigated the clinical impact and underlying molecular characteristics of C. sinensis infection-related iCCA. METHODS: We performed single-cell RNA sequencing, whole-exome sequencing, RNA sequencing, metabolomics and spatial transcriptomics in 251 patients with iCCA from three medical centers. Alterations in metabolism and the immune microenvironment of C. sinensis-related iCCAs were validated through an in vitro co-culture system and in a mouse model of iCCA. RESULTS: We revealed that C. sinensis infection was significantly associated with iCCA patients' overall survival and response to immunotherapy. Fatty acid biosynthesis and the expression of fatty acid synthase (FASN), a key enzyme catalyzing long-chain fatty acid synthesis, were significantly enriched in C. sinensis-related iCCAs. iCCA cell lines treated with excretory/secretory products of C. sinensis displayed elevated FASN and free fatty acids. The metabolic alteration of tumor cells was closely correlated with the enrichment of tumor-associated macrophage (TAM)-like macrophages and the impaired function of T cells, which led to formation of an immunosuppressive microenvironment and tumor progression. Spatial transcriptomics analysis revealed that malignant cells were in closer juxtaposition with TAM-like macrophages in C. sinensis-related iCCAs than non-C. sinensis-related iCCAs. Importantly, treatment with a FASN inhibitor significantly reversed the immunosuppressive microenvironment and enhanced anti-PD-1 efficacy in iCCA mouse models treated with excretory/secretory products from C. sinensis. CONCLUSIONS: We provide novel insights into metabolic alterations and the immune microenvironment in C. sinensis infection-related iCCAs. We also demonstrate that the combination of a FASN inhibitor with immunotherapy could be a promising strategy for the treatment of C. sinensis-related iCCAs. IMPACT AND IMPLICATIONS: Clonorchis sinensis (C. sinensis)-infected patients with intrahepatic cholangiocarcinoma (iCCA) have a worse prognosis and response to immunotherapy than non-C. sinensis-infected patients with iCCA. The underlying molecular characteristics of C. sinensis infection-related iCCAs remain unclear. Herein, we demonstrate that upregulation of FASN (fatty acid synthase) and free fatty acids in C. sinensis-related iCCAs leads to formation of an immunosuppressive microenvironment and tumor progression. Thus, administration of FASN inhibitors could significantly reverse the immunosuppressive microenvironment and further enhance the efficacy of anti-PD-1 against C. sinensis-related iCCAs.

Cognitive reserve disorder in age-related hearing loss: cognitive cortical compensatory to auditory perceptual processing.

The aim of this study is to ascertain the mechanisms of cognitive reserve disorder in age-related hearing loss (ARHL), to study the correlation between ARHL and cognitive decline via EEG, and to reverse the adverse remodeling of auditory-cognitive connectivity with hearing aids (HAs). In this study, 32 participants were enrolled, including 12 with ARHLs, 9 with HAs, and 11 healthy controls (HCs), to undergo EEG, Pure Tone Average (PTA), Montreal Cognitive Assessment (MoCA), and other general cognitive tests. There were the lowest MoCA in the ARHL group (P = 0.001), especially in language and abstraction. In the ARHL group, power spectral density of the gamma in right middle temporal gyrus was significantly higher than HC and HA groups, while functional connectivity between superior frontal gyrus and cingulate gyrus was weaker than HC group (P = 0.036) and HA group (P = 0.021). In the HA group, superior temporal gyrus and cuneus had higher connectivity than in the HC group (P = 0.036). In the ARHL group, DeltaTM_DTA (P = 0.042) and CTB (P = 0.011) were more frequent than in the HC group, while there was less DeltaTM_CTA (P = 0.029). PTA was found to be associated with MoCA (r = -0.580) and language (r = -0.572), DeltaTM_CTB had a likewise correlation with MoCA (r = 0.483) and language (r = 0.493), while DeltaTM_DTA was related to abstraction (r = -0.458). Cognitive cortexes compensate for worse auditory perceptual processing in ARHL, which relates to cognitive decline. The impaired functional connectivity between auditory and cognitive cortexes can be remodeled by HAs. DeltaTM may serve as a biomarker for early cognitive decline and decreased auditory speech perception in ARHL.

A quantitative analysis of the improvement provided by comprehensive annotation on CT lesion detection using deep learning.

BACKGROUND: Data collected from hospitals are usually partially annotated by radiologists due to time constraints. Developing and evaluating deep learning models on these data may result in over or under estimation PURPOSE: We aimed to quantitatively investigate how the percentage of annotated lesions in CT images will influence the performance of universal lesion detection (ULD) algorithms. METHODS: We trained a multi-view feature pyramid network with position-aware attention (MVP-Net) to perform ULD. Three versions of the DeepLesion dataset were created for training MVP-Net. Original DeepLesion Dataset (OriginalDL) is the publicly available, widely studied DeepLesion dataset that includes 32 735 lesions in 4427 patients which were partially labeled during routine clinical practice. Enriched DeepLesion Dataset (EnrichedDL) is an enhanced dataset that features fully labeled at one or more time points for 4145 patients with 34 317 lesions. UnionDL is the union of the OriginalDL and EnrichedDL with 54 510 labeled lesions in 4427 patients. Each dataset was used separately to train MVP-Net, resulting in the following models: OriginalCNN (replicating the original result), EnrichedCNN (testing the effect of increased annotation), and UnionCNN (featuring the greatest number of annotations). RESULTS: Although the reported mean sensitivity of OriginalCNN was 84.3% using the OriginalDL testing set, the performance fell sharply when tested on the EnrichedDL testing set, yielding mean sensitivities of 56.1%, 66.0%, and 67.8% for OriginalCNN, EnrichedCNN, and UnionCNN, respectively. We also found that increasing the percentage of annotated lesions in the training set increased sensitivity, but the margin of increase in performance gradually diminished according to the power law. CONCLUSIONS: We expanded and improved the existing DeepLesion dataset by annotating additional 21 775 lesions, and we demonstrated that using fully labeled CT images avoided overestimation of MVP-Net's performance while increasing the algorithm's sensitivity, which may have a huge impact to the future CT lesion detection research. The annotated lesions are at https://github.com/ComputationalImageAnalysisLab/DeepLesionData.

Intraindividual Comparison of Contrast-Enhanced Ultrasound Using Perfluorobutane With Modified Criteria Versus CT/MRI LI-RADS Version 2018 for Diagnosing HCC in High-Risk Patients.

BACKGROUND. Previously proposed modifications to LI-RADS criteria for contrast-enhanced ultrasound (CEUS) performed using perfluorobutane contrast agent yielded increased sensitivity for hepatocellular carcinoma (HCC) without a significant decrease in specificity. OBJECTIVE. The purpose of our study was to compare the diagnostic performance of CEUS with perfluorobutane using modified LI-RADS criteria versus contrast-enhanced CT or MRI using LI-RADS version 2018 (v2018) for characterizing lesions as HCC in high-risk patients. METHODS. This retrospective study included 171 patients (140 men, 31 women; mean age, 54 ± 12 [SD] years) at high-risk for HCC with a pathologically confirmed liver observation evaluated by both CEUS using perfluorobutane and contrast-enhanced CT or MRI between March 2020 and May 2021. A matching algorithm was used to select two patients with HCC for each patient with a non-HCC lesion. Two readers evaluated observations using previously proposed modifications to CEUS LI-RADS version 2017 that classify certain observations as LR-5 rather than as LR-4 or LR-M on the basis of the presence of Kupffer phase defect after perfluorobutane administration; two different readers evaluated observations using CT/MRI LI-RADS v2018. Each reader pair reached consensus. Diagnostic performance was evaluated. RESULTS. A total of 114 patients had HCC, 43 had a non-HCC malignancy, and 14 had a benign lesion. Modified CEUS criteria using perfluorobutane and CT/MRI LI-RADS v2018 showed no significant difference (p > .05) in sensitivity (92.1% vs 89.5%), specificity (87.7% vs 84.2%), or accuracy (90.6% vs 87.7%) of LR-5 for diagnosis of HCC. Of six observations assessed as LR-4 only by CT/MRI LI-RADS v2018, modified CEUS criteria using perfluorobutane assessed one as LR-3 (benign lesion) and five as LR-5 (all HCC). Of seven observations assessed as LR-M only by CT/MRI LI-RADS v2018, modified CEUS criteria using perfluorobutane assessed one as LR-3 (non-HCC malignancy) and six as LR-5 (all HCC). Eight of 12 observations assessed as LR-5 only by CT/MRI LI-RADS v2018 and 11 of 13 observations assessed as LR-5 only by modified CEUS criteria using perfluorobutane were HCC. CONCLUSION. The diagnostic performance of LR-5 for HCC diagnosis was not significantly different between modified CEUS criteria using perfluorobutane and CT/MRI LI-RADS v2018. CLINICAL IMPACT. The findings support the application of modified CEUS criteria using perfluorobutane for diagnosing HCC in high-risk patients.

Non-invasive diagnosis strategy of hepatocellular carcinoma in low-risk population.

AIMS: With prevalence of hepatocellular carcinoma (HCC) in low-risk population (LRP), establishing a non-invasive diagnostic strategy becomes increasingly urgent to spare unnecessary biopsies in this population. The purposes of this study were to find characterisics of HCC and to establish a proper non-invasive method to diagnose HCC in LRP. METHODS: A total of 681 patients in LRP (defined as the population without cirrhosis, chronic HBV infection or HCC history) were collected from 2 institutions. The images of computed tomography (CT) and magnetic resonance imaging (MRI) were manually analysed. We divided the patients into the training cohort (n = 324) and the internal validating cohort (n = 139) by admission time in the first institution. The cohort in the second institution was viewed as the external validation (n = 218). A multivariate logistic regression model incorporating both imaging and clinical independent risk predictors was developed. C-statistics was used to evaluate the diagnostic performance. RESULTS: Besides the major imaging features of HCC (non-rim enhancement, washout and enhancing capsule), tumor necrosis or severe ischemia (TNSI) on imaging and two clinical characteristics (gender and alpha fetoprotein) were also independently associated with HCC diagnosis (all P < 0.01). A clinical model (including 3 major features, TNSI, gender and AFP) was built to diagnose HCC and achieved good diagnostic performance (area under curve values were 0.954 in the training cohort, 0.931 in the internal validation cohort and 0.902 in the external cohort). CONCLUSIONS: The clinical model in this study developed a satisfied non-invasive diagnostic performance for HCC in LRP.

Prediction of microvascular invasion in hepatocellular carcinoma with expert-inspiration and skeleton sharing deep learning.

BACKGROUND AND AIMS: Radiological prediction of microvascular invasion (MVI) of hepatocellular carcinoma (HCC) is essential but few models were clinically implemented because of limited interpretability and generalizability. METHODS: Based on 2096 patients in three independent HCC cohorts, we established and validated an MVI predicting model. First, we used data from the primary cohort to train a 3D-ResNet network for MVI prediction and then optimised the model with "expert-inspired training" for model construction. Second, we implemented the model to the other two cohorts using three implementation strategies, the original model implementation, data sharing model implementation and skeleton sharing model implementation, the latter two of which used part of the cohorts' data for fine-tuning. The areas under the receiver operating characteristic curve (AUCs) were calculated to compare the performances of different models. RESULTS: For the MVI predicting model, the AUC of the expert-inspired model was 0.83 (95% CI: 0.77-0.88) compared to 0.54 (95% CI: 0.46-0.62) of model before expert-inspiring. Taking this model as an original model, AUC on the second cohort was 0.76 (95% CI: 0.67-0.84). The AUC was improved to 0.83 (95% CI: 0.77-0.90) with the data-sharing model, and further improved to 0.85 (95% CI: 0.79-0.92) with the skeleton sharing model. The trend that the skeleton sharing model had an advantage in performance was similar in the third cohort. CONCLUSIONS: We established an expert-inspired model with better predictive performance and interpretability than the traditional constructed model. Skeleton sharing process is superior to data sharing and direct model implementation in model implementation.

Hydraulic prediction of drought-induced plant dieback and top-kill depends on leaf habit and growth form.

Hydraulic failure caused by severe drought contributes to aboveground dieback and whole-plant death. The extent to which dieback or whole-plant death can be predicted by plant hydraulic traits has rarely been tested among species with different leaf habits and/or growth forms. We investigated 19 hydraulic traits in 40 woody species in a tropical savanna and their potential correlations with drought response during an extreme drought event during the El Niño-Southern Oscillation in 2015. Plant hydraulic trait variation was partitioned substantially by leaf habit but not growth form along a trade-off axis between traits that support drought tolerance versus avoidance. Semi-deciduous species and shrubs had the highest branch dieback and top-kill (complete aboveground death) among the leaf habits or growth forms. Dieback and top-kill were well explained by combining hydraulic traits with leaf habit and growth form, suggesting integrating life history traits with hydraulic traits will yield better predictions.

Follicular Dendritic Cell Sarcomas: CT and MRI Findings in 20 Patients.

OBJECTIVE. The objective of this study was to assess the imaging features of follicular dendritic cell sarcoma (FDCS) on CT and MRI. MATERIALS AND METHODS. The clinical data and pretreatment findings of 20 patients with pathologically proven FDCS on CT (n = 15), MRI (n = 7), or both (n = 2) were analyzed retrospectively. Tumor location, number, size, morphology, attenuation or signal intensity, margin, presence of metastases, and contrast enhancement were evaluated. RESULTS. FDCS originated from lymph nodes (n = 6) or a variety of extranodal sites (n = 14). The tumors were typically solitary and well-circumscribed. Extranodal lesions (mostly in the abdomen or mediastinum with mean diameter, 11.8 cm) were larger than nodal lesions (mean diameter, 6.5 cm). Nodal-type cases presented with homogeneous masses on CT and MRI. However, on CT, all extranodal tumors (n = 12) showed heterogeneous attenuation, of which 91.7% (11/12) contained areas of lower attenuation because of internal necrosis and 50.0% (6/12) showed calcifications. On MRI, primary hepatic or splenic tumors (n = 3) also appeared as large heterogeneous masses. Seven patients (35.0%) had advanced-stage disease, and intraabdominal extranodal cases were more likely to have regional lymphadenopathy (n = 4) and distant metastases (n = 5). Hypervascularity was seen in 90.0% (18/20) of patients and progressive enhancement was seen in 11 (78.6%) of 14 tumors with multiphase imaging. CONCLUSION. FDCS is a rare, mostly solitary, well-delineated malignancy. A nodal-type FDCS typically presents as a small homogeneous mass, whereas an extranodal FDCS in the mediastinum or abdomen manifests as a large heterogeneous mass with internal necrosis and calcifications accompanied by regional lymphadenopathy. Hypervascularity and progressive enhancement can be seen in the majority of tumors.

Three-day postoperative antibiotics reduces post-hepatectomy infection rate in hepatitis B virus-related hepatocellular carcinoma.

BACKGROUND AND AIM: The evidences for use of postoperative antibiotics (POA) in hepatocellular carcinoma (HCC) patients who underwent hepatectomy are controversial. We aimed to explore the relationship between POA and hepatectomy-related infection in a hepatitis B virus (HBV)-related HCC population. METHODS: We retrospectively collected 934 HCC patients who underwent hepatectomy for curative intent from three tertiary hospitals in China. The incidences of postoperative infection including surgical site infection and remote site infection were recorded and calculated. Univariable and multivariable logistic regression analyses were performed to explore related factors of postoperative infection and POA. And the relationship between infection rates with different durations of POA was investigated. RESULTS: The overall infection rate was 8.2% (77/934), including 6.5% (61/934) of surgical site infection and 2.0% (19/934) of remote site infection. Multivariable analysis revealed that the administration of POA was negatively related with the incidence of postoperative infection significantly (odds ratio = 0.50, 95% confidence interval = 0.30 to 0.83; P = 0.008). Albumin-bilirubin score, Barcelona Clinic Liver Cancer (BCLC) stage and extent of hepatectomy were independently related to the POA. And 3-day regimen seemed to be the shortest duration of POA to gain the lowest incidence of postoperative infection. CONCLUSIONS: Postoperative antibiotic is necessary for HBV-related HCC patients to prevent postoperative infection, especially for those with higher albumin-bilirubin score, at BCLC stage B-C, or who underwent major hepatectomy. For HBV-related HCC patients, postoperative second-generation cephalosporins, or ceftriaxone for 3 days after surgery might be proper.

Adipose tissue area as a predictor for the efficacy of apatinib in platinum-resistant ovarian cancer: an exploratory imaging biomarker analysis of the AEROC trial.

BACKGROUND: Vascular endothelial growth factor (VEGF)-targeted therapy is effective in patients with ovarian cancer. Whether adipose tissue (AT) could predict the efficacy of VEGF receptor (VEGFR) inhibitors in ovarian cancer is unknown. We aimed to evaluate the ability of distinct AT depots to predict the efficacy of apatinib, a VEGFR inhibitor, in recurrent ovarian cancers included in the AEROC trial. METHODS: The AEROC was a single-arm phase 2 trial of apatinib and oral etoposide in patients with platinum-resistant or platinum-refractory ovarian cancer. Apatinib was administered continuously, and oral etoposide was administered every 21 days for a maximum of six cycles. This was a post hoc study based on the AEROC trial. Areas of visceral AT (VAT), subcutaneous AT (SAT), and intermuscular AT (IMAT) were measured using computed tomography scan at baseline to assess their association with the objective response rate, progression-free survival, and overall survival. RESULTS: Of the 35 treated patients, 31 patients with at least one post-baseline efficacy assessment by computed tomography scan were included in this study. After adjusting for apatinib exposure, high VAT (odds ratio [OR], 0.16; 95% confidence interval [CI], 0.03-0.90, P = 0.037) and SAT (OR, 0.16; 95% CI, 0.03-0.87, P = 0.034) were significantly associated with a higher objective response rate. Further, decreased risks of disease progression and death were associated with high VAT (hazard ratio [HR], 0.39; 95% CI, 0.17-0.92, P = 0.031, and HR, 0.12; 95% CI, 0.04-0.40, P < 0.001, respectively), SAT (HR, 0.35; 95% CI, 0.15-0.83, P = 0.027, and HR, 0.24; 95% CI, 0.08-0.67, P = 0.007, respectively), and IMAT (HR, 0.20; 95% CI, 0.06-0.74, P = 0.016, and HR, 0.13; 95% CI, 0.03-0.62, P = 0.011, respectively). CONCLUSIONS: High areas of VAT, SAT, and IMAT were significantly associated with better outcomes in patients with platinum-resistant or platinum-refractory ovarian cancer who received VEGFR inhibitors. AT assessments may be valuable as patient-specific imaging biomarkers for predicting response to VEGFR inhibitors. TRIAL REGISTRATION: ClinicalTrials.gov identifier: NCT02867956 .

Prognostic Role of Time to Surgery in Hepatocellular Carcinoma at Barcelona Clinic Liver Cancer Stage 0-A.

BACKGROUND: Postsurgical recurrence is common in early-stage hepatocellular carcinoma (HCC). Prolonged time to surgery (TTS) may lead to tumor progression. However, the impact of TTS on HCC prognosis is controversial in Western studies and unknown in China. We aim to investigate the impact of TTS on the prognosis of Chinese HCC patients at Barcelona Clinic Liver Cancer (BCLC) stage 0-A who underwent surgery. PATIENTS AND METHODS: We retrospectively enrolled 967 BCLC 0-A HCC patients who underwent surgery at three tertiary centers in China. Primary outcomes were recurrence-free survival (RFS) and overall survival (OS). Restricted cubic spline (RCS) was used to select the cutoff value of TTS. Propensity score matching (PSM) was performed to reduce confounding bias, and a time-dependent Cox model was utilized to investigate factors influencing TTS. RESULTS: The median TTS of BCLC 0-A HCC patients was 13 days (interquartile range: 10-21 days). For patients with TTS ≤ 70 days, the cutoff value of TTS was 13 days according to RCS. After PSM, corresponding 1-, 3-, and 5-year RFS of the TTS > 13 days and TTS ≤ 13 days groups were 75.6%, 55.3%, 46.4% and 71.2%, 52.3%, 38.8%, respectively (P = 0.103). Corresponding 1-, 3-, and 5-year OS of TTS > 13 days and TTS ≤ 13 days groups were 93.7%, 82.8%, 69.6% and 92.4%, 78.5%, 68.4%, respectively (P = 0.580). Time-dependent Cox analysis revealed that age and tumor size were factors influencing TTS. CONCLUSIONS: Our study suggests that, for patients with TTS ≤ 70 days, prolonged TTS had no impact on BCLC 0-A Chinese HCC patients receiving surgery.

Preoperative Prediction of Pancreatic Neuroendocrine Neoplasms Grading Based on Enhanced Computed Tomography Imaging: Validation of Deep Learning with a Convolutional Neural Network.

INTRODUCTION: The pathological grading of pancreatic neuroendocrine neoplasms (pNENs) is an independent predictor of survival and indicator for treatment. Deep learning (DL) with a convolutional neural network (CNN) may improve the preoperative prediction of pNEN grading. METHODS: Ninety-three pNEN patients with preoperative contrast-enhanced computed tomography (CECT) from Hospital I were retrospectively enrolled. A CNN-based DL algorithm was applied to the CECT images to obtain 3 models (arterial, venous, and arterial/venous models), the performances of which were evaluated via an eightfold cross-validation technique. The CECT images of the optimal phase were used for comparing the DL and traditional machine learning (TML) models in predicting the pathological grading of pNENs. The performance of radiologists by using qualitative and quantitative computed tomography findings was also evaluated. The best DL model from the eightfold cross-validation was evaluated on an independent testing set of 19 patients from Hospital II who were scanned on a different scanner. The Kaplan-Meier (KM) analysis was employed for survival analysis. RESULTS: The area under the curve (AUC; 0.81) of arterial phase in validation set was significantly higher than those of venous (AUC 0.57, p = 0.03) and arterial/venous phase (AUC 0.70, p = 0.03) in predicting the pathological grading of pNENs. Compared with the TML models, the DL model gave a higher (although insignificantly) AUC. The highest OR was achieved for the p ratio <0.9, the AUC and accuracy for diagnosing G3 pNENs were 0.80 and 79.1% respectively. The DL algorithm achieved an AUC of 0.82 and an accuracy of 88.1% for the independent testing set. The KM analysis showed a statistical significant difference between the predicted G1/2 and G3 groups in the progression-free survival (p = 0.001) and overall survival (p < 0.001). CONCLUSION: The CNN-based DL method showed a relatively robust performance in predicting pathological grading of pNENs from CECT images.

Residual convolutional neural network for predicting response of transarterial chemoembolization in hepatocellular carcinoma from CT imaging.

BACKGROUND: We attempted to train and validate a model of deep learning for the preoperative prediction of the response of patients with intermediate-stage hepatocellular carcinoma (HCC) undergoing transarterial chemoembolization (TACE). METHOD: All computed tomography (CT) images were acquired for 562 patients from the Nan Fang Hospital (NFH), 89 patients from Zhu Hai Hospital Affiliated with Jinan University (ZHHAJU), and 138 patients from the Sun Yat-sen University Cancer Center (SYUCC). We built a predictive model from the outputs using the transfer learning techniques of a residual convolutional neural network (ResNet50). The prediction accuracy for each patch was revaluated in two independent validation cohorts. RESULTS: In the training set (NFH), the deep learning model had an accuracy of 84.3% and areas under curves (AUCs) of 0.97, 0.96, 0.95, and 0.96 for complete response (CR), partial response (PR), stable disease (SD), and progressive disease (PD), respectively. In the other two validation sets (ZHHAJU and SYUCC), the deep learning model had accuracies of 85.1% and 82.8% for CR, PR, SD, and PD. The ResNet50 model also had high AUCs for predicting the objective response of TACE therapy in patches and patients of three cohorts. Decision curve analysis (DCA) showed that the ResNet50 model had a high net benefit in the two validation cohorts. CONCLUSION: The deep learning model presented a good performance for predicting the response of TACE therapy and could help clinicians in better screening patients with HCC who can benefit from the interventional treatment. KEY POINTS: • Therapy response of TACE can be predicted by a deep learning model based on CT images. • The probability value from a trained or validation deep learning model showed significant correlation with different therapy responses. • Further improvement is necessary before clinical utilization.

The presence of microvascular invasion guides treatment strategy in recurrent HBV-related HCC.

OBJECTIVES: We used the status of microvascular invasion (MVI) at primary resection to help treatment selection for hepatitis B virus-positive (HBV+) recurrent hepatocellular carcinoma (rHCC) patients in Barcelona Clinic Liver Cancer (BCLC) stage B-C. METHODS: From 2009 to 2017, we enrolled 221 consecutive HBV+ rHCC patients at BCLC stage B-C who underwent re-resection (RR), radiofrequency ablation (RFA), or transarterial chemoembolization (TACE). Post recurrence survival (PRS) and overall survival (OS) were compared between RR/RFA and TACE according to MVI status. A one-to-one propensity score matching analysis was performed. RESULTS: For MVI(-) patients, the median PRS was 62.3 months for the RR/RFA group and 21.1 months for the TACE group (p = 0.039). The corresponding OS was 71.4 months and 26.6 months, respectively (p = 0.010). For MVI(+) patients, the median PRS in the RR/RFA group and TACE group was 14.7 months and 10.1 months (p = 0.115). The corresponding OS was 23.4 months and 16.4 months, respectively (p = 0.067). After matching, the dominance of RR/RFA over TACE remained in MVI(-) patients for both PRS (62.3 months vs 15.3 months, p = 0.019) and OS (98.1 months vs 33.4 months, p = 0.046). No significant difference was found in MVI(+) patients for either PRS (14.7 months vs 11.8 months, p = 0.593) or OS (23.4 months vs 28.1 months, p = 0.662). CONCLUSIONS: MVI status definitely helps select treatment options in HBV+ rHCC patients. For MVI(-) patients, RR/RFA provided better survival than TACE while for MVI(+) patients, TACE shared similar survival outcomes. KEY POINTS: • This study aimed at the determination of the optimal treatment options (ablation /resection vs TACE) in case of recurrent HBV-related HCC. • It showed that MVI status, established at primary resection of HCC, was a powerful marker for selecting the best treatment option in these patients. • In MVI(-) patients, RR/RFA achieved a better survival than TACE. In MVI(+) patients, TACE shared similar survival.

Hepatic resection versus transarterial chemoembolization in infiltrative hepatocellular carcinoma: A multicenter study.

BACKGROUND AND AIM: Prognosis of infiltrative hepatocellular carcinoma (iHCC) is poor, and the treatments selection based on efficacy is unclear. We performed this multicenter study to compare the efficacy of hepatic resection and transarterial chemoembolization (TACE) in treating patients with iHCC. METHODS: We retrospectively analyzed the overall survivals (OS) in 319 patients with iHCC who were initially treated by hepatic resection (n = 133) or TACE (n = 186) at four tertiary centers. Fifty-eight patients in the TACE group were assessed as resectable and compared with the hepatic resection group in subgroup analysis. A propensity score matched (PSM) analysis was performed to reduce selection bias. Cox regression was performed to identify significant factors associated with OS. RESULTS: The median OS time was significantly longer in the hepatic resection group than that in the TACE group, before and after PSM (before PSM, 17.5 vs 7.3 months, P < 0.0001; after PSM, 14.0 vs 7.3 months, P < 0.0001). The multivariable analysis indicated TACE as a risk factor of OS (hazard ratio = 2.233, 95% confidence interval = 1.492 to 3.341, P < 0.0001), as well as portal venous tumor thrombosis grades 3-4 and alpha fetal protein (AFP) > 400 ng/mL. In the subgroup analysis, the better efficacy of hepatic resection over TACE persisted regardless of the grade of portal venous tumor thrombosis and the level of AFP. As for resectable patients, hepatic resection still showed significant survival benefit (before PSM, 17.5 vs 11.2 months, P = 0.0013; after PSM, 14.0 vs 10.9 months, P = 0.0304). CONCLUSION: Hepatic resection might be the better choice for patients with iHCC due to its better survival benefit than TACE.

Lymph-node Epstein-Barr virus concentration in diagnosing cervical lymph-node metastasis in nasopharyngeal carcinoma.

PURPOSE: Cervical lymph-node (CLN) metastasis commonly occurs in patients with nasopharyngeal carcinoma (NPC) metastasis. The presence of Epstein-Barr virus (EBV) genomes in neck lymph nodes may diagnose CLN. This research was designed to appraise the diagnostic value of EBV concentration for cervical lymph nodes in NPC. METHODS: Two hundred and fifty-three NPC patients with 276 CLNs were enrolled. MRI was performed to detect CLN metastasis, and plasma EBV concentration was measured by quantitative PCR before treatment. Ultrasonography (US) and US-FNA were subsequently performed in the suspicious lymph nodes. Fifteen patients (22 lymph nodes) underwent fine-needle aspiration cytology (FNAC), and the remaining 242 patients (254 lymph nodes) underwent core needle biopsy (CNB) for CLNs at the clinician's demand. The aspiration needle was rinsed with 1 ml of normal saline for EBV detection. The method of lymph-node EBV measurement was consistent with that for plasma. The MRI results and EBV concentrations in plasma and lymph nodes were recorded and analyzed. Plasma EBV concentrations ≥ 4000 copies/ml were regarded as positive. RESULTS: CLN-EBV concentrations ≥ 787.5 copies/ml were regarded as positive according to receiver-operating characteristic curve analysis. The AUC of the EBV (0.925) concentration in CLN metastasis was significantly larger than the AUC of MRI (0.714) (P < 0.001). The sensitivity and specificity were 94.09% and 48.72% for MRI in lymph-node metastasis and 95.36% (P > 0.05) and 84.62% (P < 0.01) for EBV DNA in CLN metastasis, respectively. The sensitivity and specificity of EBV in plasma were 77.2% and 71.8%, respectively. The diagnostic specificity and AUC of EBV in CLNs were higher than those of MRI and plasma EBV (P < 0.005). CONCLUSIONS: Ultrasound-guided CLN FNA to obtain EBV concentrations may provide a new method to diagnose CLN metastasis with high sensitivity and specificity.

CT-based radiomics for preoperative prediction of early recurrent hepatocellular carcinoma: technical reproducibility of acquisition and scanners.

PURPOSE: To test the technical reproducibility of acquisition and scanners of CT image-based radiomics model for early recurrent hepatocellular carcinoma (HCC). METHODS: We included primary HCC patient undergone curative therapies, using early recurrence as endpoint. Four datasets were constructed: 109 images from hospital #1 for training (set 1: 1-mm image slice thickness), 47 images from hospital #1 for internal validation (sets 2 and 3: 1-mm and 10-mm image slice thicknesses, respectively), and 47 images from hospital #2 for external validation (set 4: vastly different from training dataset). A radiomics model was constructed. Radiomics technical reproducibility was measured by overfitting and calibration deviation in external validation dataset. The influence of slice thickness on reproducibility was evaluated in two internal validation datasets. RESULTS: Compared with set 1, the model in set 2 indicated favorable prediction efficiency (the area under the curve 0.79 vs. 0.80, P = 0.47) and good calibration (unreliability statistic U: P = 0.33). However, in set 4, significant overfitting (0.63 vs. 0.80, P < 0.01) and calibration deviation (U: P < 0.01) were observed. Similar poor performance was also observed in set 3 (0.56 vs. 0.80, P = 0.02; U: P < 0.01). CONCLUSIONS: CT-based radiomics has poor reproducibility between centers. Image heterogeneity, such as slice thickness, can be a significant influencing factor.

Pancreatic tumor in type 1 autoimmune pancreatitis: a diagnostic challenge.

BACKGROUND: The co-occurrence of type 1 autoimmune pancreatitis (AIP) and pancreatic tumor (PaT) has been previously reported. Pure AIP cases have favorable prognosis and are primarily treated with steroids, while AIP cases with PaT are associated with poor prognosis where the primary management is pancreatic resection. However, it's a challenge to timely identify the concurrent PaT in AIP because of their similar clinical and radiological manifestations. METHODS: We retrospectively reviewed the data in two medical centers from January 2010 to April 2019. The inclusion criteria were as follows: 1) completion of abdominal CT imaging before invasive procedures to the pancreas, 2) a final diagnosis of type 1 AIP using the 2011 international consensus diagnostic criteria, 3) follow-up duration of at least one month unless AIP and PaT were identified simultaneously. The presence of PaT in AIP was made based on histopathological confirmation, and the absence of PaT in AIP was defined as no pathological or radiological evidence of concurrent PaT. Clinical and radiological characteristics including gender, age, surveillance period, serum IgG4 and Ca-199 levels, biopsy, extrapancreatic involvement, CT and MR (if performed) imaging characteristics were compared between AIP with and without PaT. The Fisher's exact test was used for qualitative variables, and nonparametric Mann-Whitney test for quantitative variables. A p value ≤0.05 was considered statistically significant. RESULTS: A total of 74 patients with type 1 AIP were included, of which 5 (6.7%) had the concurrent PaT. The subtypes were pancreatic ductal adenocarcinoma (3/5), solitary extramedullary plasmacytoma in the pancreas (1/5) and cholangiocarcinoma in the pancreatic segment (1/5), respectively. Gender (p = 0.044), the pattern of pancreatic enlargement (p = 0.003), heterogeneity (p = 0.015), low-density (p = 0.004) on CT and rim enhancement on MRI (p = 0.050) differed significantly between AIP with and without PaT. None of the low-density characteristics on CT or other assessed MRI characteristics could significantly differentiate the two groups (p>0.05). CONCLUSIONS: Female, focal pancreatic enlargement, pancreatic heterogeneity, low-density on CT and rim enhancement on MRI are suggestive of the concurrent PaT in type 1 AIP. The characteristics of low-density on CT or other MRI characteristics did not provide further diagnostic values.

Lipiodol deposition in portal vein tumour thrombus predicts treatment outcome in HCC patients after transarterial chemoembolisation.

OBJECTIVE: To study lipiodol deposition in portal vein tumour thrombus (PVTT) in predicting the treatment outcome of hepatocellular carcinoma (HCC) patients after transarterial chemoembolisation (TACE). METHODS: We retrospectively reviewed data from 379 HCC patients with PVTT who underwent TACE as the initial treatment at Sun Yat-Sen University Cancer Center from January 2008 to December 2015. Patients were grouped by positive and negative lipiodol deposition based on the extent of lipiodol deposition in PVTT. The overall survival (OS) and progression-free survival (PFS) were compared between negative and positive lipiodol deposition groups; furthermore, the value of the combinatorial evaluation of tumour responses and lipiodol deposition in PVTT in predicting prognosis was analysed in subgroup patients with stable disease (SD) after TACE. RESULTS: Of the 379 patients, 264 (69.7%) had negative and 115 (30.3%) had positive lipiodol deposition in PVTT after TACE. Multivariate analysis identified positive lipiodol deposition in PVTT as an independent prognostic factor for favourable OS (p = 0.001). The median OS and PFS of negative and positive lipiodol deposition groups were 4.70 vs. 8.97 months (p = 0.001) and 3.1 months vs. 5.8 months (p < 0.001). In subgroup patients, the median OS and PFS of negative and positive lipiodol deposition groups were 4.7 months vs. 10.5 months (p < 0.001) and 3.5 months vs. 7.0 months (p < 0.001), respectively. CONCLUSIONS: The patients with positive lipiodol deposition in PVTT had a longer OS than those with negative lipiodol deposition. Furthermore, the positive lipiodol deposition in PVTT can further differentiate HCC patients with favourable prognosis from SD patients. KEY POINTS: • Lipiodol deposition in PVTT is a prognostic indicator for HCC patients after TACE treatment. • Positive lipiodol deposition in PVTT is associated with a better prognosis.