Rapid and automated design of two-component protein nanomaterials using ProteinMPNN.

The design of protein-protein interfaces using physics-based design methods such as Rosetta requires substantial computational resources and manual refinement by expert structural biologists. Deep learning methods promise to simplify protein-protein interface design and enable its application to a wide variety of problems by researchers from various scientific disciplines. Here, we test the ability of a deep learning method for protein sequence design, ProteinMPNN, to design two-component tetrahedral protein nanomaterials and benchmark its performance against Rosetta. ProteinMPNN had a similar success rate to Rosetta, yielding 13 new experimentally confirmed assemblies, but required orders of magnitude less computation and no manual refinement. The interfaces designed by ProteinMPNN were substantially more polar than those designed by Rosetta, which facilitated in vitro assembly of the designed nanomaterials from independently purified components. Crystal structures of several of the assemblies confirmed the accuracy of the design method at high resolution. Our results showcase the potential of deep learning-based methods to unlock the widespread application of designed protein-protein interfaces and self-assembling protein nanomaterials in biotechnology.

Validation of the Mayo Imaging Classification System for Predicting Kidney Outcomes in ADPKD.

BACKGROUND: The Mayo Imaging Classification was developed to predict the rate of disease progression in patients with autosomal dominant polycystic kidney disease. This study aimed to validate its ability to predict kidney outcomes in a large multicenter autosomal dominant polycystic kidney disease cohort. METHODS: Included were patients with ≥1 height-adjusted total kidney volume (HtTKV) measurement and ≥3 eGFR values during ≥1-year follow-up. Mayo HtTKV class stability, kidney growth rates, and eGFR decline rates were calculated. The observed eGFR decline was compared with predictions from the Mayo Clinic future eGFR equation. The future eGFR prediction equation was also tested for nonlinear eGFR decline. Kaplan-Meier survival analysis and Cox regression models were used to assess time to kidney failure using Mayo HtTKV class as a predictor variable. RESULTS: We analyzed 618 patients with a mean age of 47±11 years and mean eGFR of 64±25 ml/min per 1.73 m 2 at baseline. Most patients (82%) remained in their baseline Mayo HtTKV class. During a mean follow-up of 5.1±2.2 years, the mean total kidney volume growth rates and eGFR decline were 5.33%±3.90%/yr and -3.31±2.53 ml/min per 1.73 m 2 per year, respectively. Kidney growth and eGFR decline showed considerable overlap between the classes. The observed annual eGFR decline was not significantly different from the predicted values for classes 1A, 1B, 1C, and 1D but significantly slower for class 1E. This was also observed in patients aged younger than 40 years and older than 60 years and those with PKD2 mutations. A polynomial model allowing nonlinear eGFR decline provided more accurate slope predictions. Ninety-seven patients (16%) developed kidney failure during follow-up. The classification predicted the development of kidney failure, although the sensitivity and positive predictive values were limited. CONCLUSIONS: The Mayo Imaging Classification demonstrated acceptable stability and generally predicted kidney failure and eGFR decline rate. However, there was marked interindividual variability in the rate of disease progression within each class.

Machine learning-based radiomic analysis and growth visualization for ablation site recurrence diagnosis in follow-up CT.

OBJECTIVES: Detecting ablation site recurrence (ASR) after thermal ablation remains a challenge for radiologists due to the similarity between tumor recurrence and post-ablative changes. Radiomic analysis and machine learning methods may show additional value in addressing this challenge. The present study primarily sought to determine the efficacy of radiomic analysis in detecting ASR on follow-up computed tomography (CT) scans. The second aim was to develop a visualization tool capable of emphasizing regions of ASR between follow-up scans in individual patients. MATERIALS AND METHODS: Lasso regression and Extreme Gradient Boosting (XGBoost) classifiers were employed for modeling radiomic features extracted from regions of interest delineated by two radiologists. A leave-one-out test (LOOT) was utilized for performance evaluation. A visualization method, creating difference heatmaps (diff-maps) between two follow-up scans, was developed to emphasize regions of growth and thereby highlighting potential ASR. RESULTS: A total of 55 patients, including 20 with and 35 without ASR, were included in the radiomic analysis. The best performing model was achieved by Lasso regression tested with the LOOT approach, reaching an area under the curve (AUC) of 0.97 and an accuracy of 92.73%. The XGBoost classifier demonstrated better performance when trained with all extracted radiomic features than without feature selection, achieving an AUC of 0.93 and an accuracy of 89.09%. The diff-maps correctly highlighted post-ablative liver tumor recurrence in all patients. CONCLUSIONS: Machine learning-based radiomic analysis and growth visualization proved effective in detecting ablation site recurrence on follow-up CT scans.

No need for fasting prior to doppler ultrasound of pediatric liver transplants: A self-controlled study.

BACKGROUND: Children frequently undergo routine Doppler-ultrasound (DUS) after liver transplantation (LT) for which they are fasted, but this may cause hunger and discomfort. OBJECTIVE: To determine if DUS measurements, with focus on the portal vein (PV), are affected by prandial changes, and if this affects distress and feasibility of the DUS. MATERIALS AND METHODS: Children were prospectively included to undergo a pre- and postprandial DUS on the same day at 6 months after LT. Pre- and anastomotic PV peak systolic velocity (PSV), and hepatic artery and hepatic vein DUS measurements were obtained. Pre- and postprandial measurements, and relative postprandial change of PV velocity ratio (VR) compared to PV anastomotic PSV, were compared using paired-sample t-tests and intraclass correlation coefficients (ICC). Obscuration by bowel gas, difficulty of DUS, and impact of fasting were assessed using 5-point rating scales. RESULTS: Twenty-eight children (median age 3.5 years, IQR 1.6-10.8) were included; four were subsequently excluded because they were not fasted (N = 2) or withdrew consent for the second DUS (N = 2). Measurements between pre- and postprandial DUS, and relative postprandial change of VR compared to PV anastomotic PSV, were not significantly different (p > .05). Test consistency was good (ICC = 0.69, 95% CI = 0.29-0.67) for PV anastomotic PSV, and excellent (95% CI = 0.61-0.93) for PV VR. Obscuration by bowel gas or ease of DUS did not change after eating (p > .05). The majority (16/28, 57.2%) found fasting difficult, and several (13/28, 46.4%) got upset when fasted. CONCLUSION: Children with an LT do not need to be fasted for routine DUS, which may decrease the burden of the examination.

Sonic hedgehog hepatocellular adenoma: magnetic resonance imaging features and correlation with histology.

OBJECTIVES: Sonic hedgehog hepatocellular adenoma (shHCA) is a new hepatocellular adenoma (HCA) subgroup characterized by high risk of hemorrhage. ShHCA account for below 10% of all HCA cases and are often associated with female gender, obesity, and non-alcoholic steatohepatitis. No specific MRI characteristics have been described to date. The objective of this study was to assess the value of using MRI to identify shHCA, and correlate MRI findings with histology. METHODS: We retrospectively collected MRI scans of 29 patients with shHCA from our center and from different liver referral centers to include 35 lesions. Diagnosis of shHCA was assessed by immunohistochemical overexpression of argininosuccinate synthase 1 or prostaglandin D2 synthase, then confirmed by molecular analysis of sonic hedgehog pathway activation and/or by proteomic analysis. RESULTS: In 46% (n = 16/35) of shHCA cases, we detected intralesional fluid-filled cavities defined on MR images as fluid-like foci markedly hyperintense on T2-weighted sequences, and hypointense on T1-weighted sequences, with or without delayed enhancement. Pathologically, these cavities were observed in 54% of cases as vacuoles filled with blood at different stages of degradation. Hemorrhage and/or necrosis were detected among 71% of cases by MRI analysis (n = 25/35) versus 82% pathologically. Seventeen percent of shHCA cases (n = 6/35) were completely homogeneous via MRI and pathological analysis. No MRI criteria was found in favor of focal nodular hyperplasia, HNF1A-mutated HCA, or typical inflammatory HCA. CONCLUSION: We reveal the presence of intralesional fluid-filled cavities among 46% of our shHCA cases that represent a new MRI finding possibly helpful for shHCA diagnosis. CLINICAL RELEVANCE STATEMENT: This multicenter study is the first clinical study about the radiological aspect of this new hepatocellular adenoma subgroup. This highlights a strong correlation between MRI and histological analysis, with a specific pattern emerging for diagnosis. KEY POINTS: • Sonic hedgehog hepatocellular adenoma is a new hepatocellular adenoma subgroup associated with high risk of hemorrhage, but imaging features of this subgroup remain unknown. • Analysis of MR images and correlation with pathology revealed intralesional fluid-filled cavities and necrotic-hemorrhagic changes. • Intralesional fluid-filled cavities have not yet been described in other adenoma subtypes and represent a new MRI finding for sonic hedgehog hepatocellular adenoma.

pH-Responsive Elastin-Like Polypeptide Designer Condensates.

Biomolecular condensates are macromolecular complexes formed by liquid-liquid phase separation. They regulate key biological functions by reversibly compartmentalizing molecules in cells, in a stimulus-dependent manner. Designing stimuli-responsive synthetic condensates is crucial for engineering compartmentalized synthetic cells that are able to mimic spatiotemporal control over the biochemical reactions. Here, we design and test a family of condensate-forming, pH-responsive elastin-like polypeptides (ELPs) that form condensates above critical pH values ranging between 4 and 7, for temperatures between 20 and at 37 °C. We show that the condensation occurs rapidly, in sharp pH intervals (ΔpH < 0.3). For eventual applications in engineering synthetic cell compartments, we demonstrate that multiple types of pH-responsive ELPs can form mixed condensates inside micron-sized vesicles. When genetically fused with enzymes, receptors, and signaling molecules, these pH-responsive ELPs could be potentially used as pH-switchable functional condensates for spatially controlling biochemistry in engineered synthetic cells.

Rapid and automated design of two-component protein nanomaterials using ProteinMPNN.

The design of novel protein-protein interfaces using physics-based design methods such as Rosetta requires substantial computational resources and manual refinement by expert structural biologists. A new generation of deep learning methods promises to simplify protein-protein interface design and enable its application to a wide variety of problems by researchers from various scientific disciplines. Here we test the ability of a deep learning method for protein sequence design, ProteinMPNN, to design two-component tetrahedral protein nanomaterials and benchmark its performance against Rosetta. ProteinMPNN had a similar success rate to Rosetta, yielding 13 new experimentally confirmed assemblies, but required orders of magnitude less computation and no manual refinement. The interfaces designed by ProteinMPNN were substantially more polar than those designed by Rosetta, which facilitated in vitro assembly of the designed nanomaterials from independently purified components. Crystal structures of several of the assemblies confirmed the accuracy of the design method at high resolution. Our results showcase the potential of deep learning-based methods to unlock the widespread application of designed protein-protein interfaces and self-assembling protein nanomaterials in biotechnology.

A dual-tracer approach using [11C]CH and [18F]FDG in HCC clinical decision making.

BACKGROUND: Early detection of recurrent or progressive HCC remains the strongest prognostic factor for survival. Dual tracer PET/CT imaging with [11C]CH and [18F]FDG can further increase detection rates as both tracers entail different metabolic pathways involved in HCC development. We investigated dual-tracer PET/CT in clinical decision making in patients suspected of recurrent or progressive HCC. All HCC patients who underwent both [11C]CH and [18F]FDG PET/CT in our institute from February 2018 to December 2021 were included. Both tracer PET/CT were within 4 weeks of each other with at least 6-month follow-up. Patients underwent dual tracer PET/CT because of unexplained and suspicious CT/MRI or sudden rise of serum tumour markers. A detected lesion was considered critical when the finding had prognostic consequences leading to treatment changes. RESULTS: Nineteen patients who underwent [11C]CH and [18F]FDG PET/CT were included of which all but six patients were previously treated for HCC. Dual-tracer critical finding detection rate was 95%, with [18F]FDG 68%, and [11C]CH 84%. Intrahepatic HCC recurrence finding rate was 65% for both tracers. [18F]FDG found more ablation site recurrences (4/5) compared to [11C]CH (2/5). Only [11C]CH found two needle tract metastases. Both tracers found 75% of the positive lymph nodes. Two new primary tumours were found, one by [18F]FDG and both by [11C]CH. CONCLUSIONS: Our study favours a dual-tracer approach in HCC staging in high-risk patients or when conventional imaging is non-conclusive.

Prediction of metastatic pheochromocytoma and paraganglioma: a machine learning modelling study using data from a cross-sectional cohort.

BACKGROUND: Pheochromocytomas and paragangliomas have up to a 20% rate of metastatic disease that cannot be reliably predicted. This study prospectively assessed whether the dopamine metabolite, methoxytyramine, might predict metastatic disease, whether predictions might be improved using machine learning models that incorporate other features, and how machine learning-based predictions compare with predictions made by specialists in the field. METHODS: In this machine learning modelling study, we used cross-sectional cohort data from the PMT trial, based in Germany, Poland, and the Netherlands, to prospectively examine the utility of methoxytyramine to predict metastatic disease in 267 patients with pheochromocytoma or paraganglioma and positive biochemical test results at initial screening. Another retrospective dataset of 493 patients with these tumors enrolled under clinical protocols at National Institutes of Health (00-CH-0093) and the Netherlands (PRESCRIPT trial) was used to train and validate machine learning models according to selections of additional features. The best performing machine learning models were then externally validated using data for all patients in the PMT trial. For comparison, 12 specialists provided predictions of metastatic disease using data from the training and external validation datasets. FINDINGS: Prospective predictions indicated that plasma methoxytyramine could identify metastatic disease at sensitivities of 52% and specificities of 85%. The best performing machine learning model was based on an ensemble tree classifier algorithm that used nine features: plasma methoxytyramine, metanephrine, normetanephrine, age, sex, previous history of pheochromocytoma or paraganglioma, location and size of primary tumours, and presence of multifocal disease. This model had an area under the receiver operating characteristic curve of 0·942 (95% CI 0·894-0·969) that was larger (p<0·0001) than that of the best performing specialist before (0·815, 0·778-0·853) and after (0·812, 0·781-0·854) provision of SDHB variant data. Sensitivity for prediction of metastatic disease in the external validation cohort reached 83% at a specificity of 92%. INTERPRETATION: Although methoxytyramine has some utility for prediction of metastatic pheochromocytomas and paragangliomas, sensitivity is limited. Predictive value is considerably enhanced with machine learning models that incorporate our nine recommended features. Our final model provides a preoperative approach to predict metastases in patients with pheochromocytomas and paragangliomas, and thereby guide individualised patient management and follow-up. FUNDING: Deutsche Forschungsgemeinschaft.

De novo designed ice-binding proteins from twist-constrained helices.

Attaining molecular-level control over solidification processes is a crucial aspect of materials science. To control ice formation, organisms have evolved bewildering arrays of ice-binding proteins (IBPs), but these have poorly understood structure-activity relationships. We propose that reverse engineering using de novo computational protein design can shed light on structure-activity relationships of IBPs. We hypothesized that the model alpha-helical winter flounder antifreeze protein uses an unusual undertwisting of its alpha-helix to align its putative ice-binding threonine residues in exactly the same direction. We test this hypothesis by designing a series of straight three-helix bundles with an ice-binding helix projecting threonines and two supporting helices constraining the twist of the ice-binding helix. Our findings show that ice-recrystallization inhibition by the designed proteins increases with the degree of designed undertwisting, thus validating our hypothesis, and opening up avenues for the computational design of IBPs.

Evaluation of National Surgical Practice for Lateral Lymph Nodes in Rectal Cancer in an Untrained Setting.

BACKGROUND: Involved lateral lymph nodes (LLNs) have been associated with increased local recurrence (LR) and ipsi-lateral LR (LLR) rates. However, consensus regarding the indication and type of surgical treatment for suspicious LLNs is lacking. This study evaluated the surgical treatment of LLNs in an untrained setting at a national level. METHODS: Patients who underwent additional LLN surgery were selected from a national cross-sectional cohort study regarding patients undergoing rectal cancer surgery in 69 Dutch hospitals in 2016. LLN surgery consisted of either 'node-picking' (the removal of an individual LLN) or 'partial regional node dissection' (PRND; an incomplete resection of the LLN area). For all patients with primarily enlarged (≥7 mm) LLNs, those undergoing rectal surgery with an additional LLN procedure were compared to those  undergoing only rectal resection. RESULTS: Out of 3057 patients, 64 underwent additional LLN surgery, with 4-year LR and LLR rates of 26% and 15%, respectively. Forty-eight patients (75%) had enlarged LLNs, with corresponding recurrence rates of 26% and 19%, respectively. Node-picking (n = 40) resulted in a 20% 4-year LLR, and a 14% LLR after PRND (n = 8; p = 0.677). Multivariable analysis of 158 patients with enlarged LLNs undergoing additional LLN surgery (n = 48) or rectal resection alone (n = 110) showed no significant association of LLN surgery with 4-year LR or LLR, but suggested higher recurrence risks after LLN surgery (LR: hazard ratio [HR] 1.5, 95% confidence interval [CI] 0.7-3.2, p = 0.264; LLR: HR 1.9, 95% CI 0.2-2.5, p = 0.874). CONCLUSION: Evaluation of Dutch practice in 2016 revealed that approximately one-third of patients with primarily enlarged LLNs underwent surgical treatment, mostly consisting of node-picking. Recurrence rates were not significantly affected by LLN surgery, but did suggest worse outcomes. Outcomes of LLN surgery after adequate training requires further research.

Hepatic artery and portal vein Doppler ultrasound reference values in children aged 0-17 years old.

Introduction: Doppler ultrasound of the portal vein peak systolic velocity and hepatic artery peak systolic velocity and resistive index in children is often performed during abdominal ultrasound for the assessment of liver and other abdominal pathology. However, evidence-based reference values are not available. We aimed to determine these reference values and to investigate whether they are age-dependent. Methods: Children who underwent abdominal ultrasound between 2020 and 2021 were retrospectively identified. Patients without hepatic or cardiac abnormalities at the time of ultrasound or during at least 3 months follow-up were eligible for the study. Ultrasound without hepatic hilum portal vein peak systolic velocity and/or hepatic artery peak systolic velocity and resistive index measurements were excluded. Age-dependent changes were analyzed using linear regression. Normal range reference values were described using percentiles for all ages, and for age subgroups. Results: One-hundred ultrasound examinations performed in 100 healthy children aged 0-17.9 years (median 7.8 years, interquartile range 1.1-14.1 years) were included. Ninety-nine portal vein peak systolic velocity and 80 hepatic artery peak systolic velocity and resistive index measurements were obtained. There was no significant association between portal vein peak systolic velocity and age (ß = -0.056, p = 0.68). There were significant associations between age and hepatic artery peak systolic velocity, and between age and hepatic artery resistive index (ß = -0.873, p = 0.04 and ß = -0.004, p < 0.001, respectively). Detailed reference values were provided for all ages, and for age subgroups. Conclusion: Reference values for the hepatic hilum portal vein peak systolic velocity, hepatic artery peak systolic velocity, and hepatic artery resistive index in children were established. Portal vein peak systolic velocity is not age-dependent, whereas hepatic artery peak systolic velocity and hepatic artery resistive index decrease when children get older.

A prospective multicentre trial on survival after Microwave Ablation VErsus Resection for Resectable Colorectal liver metastases (MAVERRIC).

AIM: This multi-centre prospective cohort study aimed to investigate non-inferiority in patients' overall survival when treating potentially resectable colorectal cancer liver metastasis (CRLM) with stereotactic microwave ablation (SMWA) as opposed to hepatic resection (HR). METHODS: Patients with no more than 5 CRLM no larger than 30 mm, deemed eligible for both SMWA and hepatic resection at the local multidisciplinary team meetings, were deliberately treated with SMWA (study group). The contemporary control group consisted of patients with no more than 5 CRLM, none larger than 30 mm, treated with HR, extracted from a prospectively maintained nationwide Swedish database. After propensity-score matching, 3-year overall survival (OS) was compared as the primary outcome using Kaplan-Meier and Cox regression analyses. RESULTS: All patients in the study group (n = 98) were matched to 158 patients from the control group (mean standardised difference in baseline covariates = 0.077). OS rates at 3 years were 78% (Confidence interval [CI] 68-85%) after SMWA versus 76% (CI 69-82%) after HR (stratified Log-rank test p = 0.861). Estimated 5-year OS rates were 56% (CI 45-66%) versus 58% (CI 50-66%). The adjusted hazard ratio for treatment type was 1.020 (CI 0.689-1.510). Overall and major complications were lower after SMWA (percentage decrease 67% and 80%, p < 0.01). Hepatic retreatments were more frequent after SMWA (percentage increase 78%, p < 0.01). CONCLUSION: SMWA is a valid curative-intent treatment alternative to surgical resection for small resectable CRLM. It represents an attractive option in terms of treatment-related morbidity with potentially wider options regarding hepatic retreatments over the future course of disease.

Doppler-ultrasound reference values after pediatric liver transplantation: a consecutive cohort study.

OBJECTIVES: Doppler ultrasound (DUS) is the main imaging modality to evaluate vascular complications of pediatric liver transplants (LT). The current study aimed to determine reference values and their change over time. METHODS: A consecutive cohort of pediatric patients undergoing an LT were retrospectively included between 2015 and 2020. Timepoints for standardized DUS were intra-operative and postoperative (day 0), days 1-7, months 1 and 3, and years 1 and 2. DUS measurements of the hepatic artery (HA), portal vein (PV), and hepatic vein(s) (HV) were included if there were no complications during 2 years follow-up. Measurements consisted of: peak systolic velocity (PSV) and resistive index (RI) for the HA, PSV for the PV, and venous pulsatility index (VPI) for the HV. Generalized estimating equations were used to analyze change over time. RESULTS: One hundred twelve pediatric patients with 123 LTs were included (median age 3.3 years, interquartile range 0.7-10.1). Ninety-five HAs, 100 PVs, and 115 HVs without complications were included. Reference values for HA PSV and RI, PV PSV, and HV VPI were obtained for all timepoints (4043 included data points in total) and presented using 5th-95th percentiles and threshold values. All reference values changed significantly over time (p = 0.032 to p < 0.001). CONCLUSIONS: DUS reference values of hepatic vessels in children after LT are presented, reference values change over time with specific vessel-dependent patterns. Timepoint-specific reference values improve the interpretation of DUS values and may help to better weigh their clinical significance. KEY POINTS: • Doppler ultrasound reference values of pediatric liver transplantations are not static but change over time. Applying the correct reference values for the specific timepoint may further improve the interpretation of the measurements. • The pattern of change over time of Doppler ultrasound measurements differs between the hepatic vessel and measurement; knowledge of these patterns may help radiologists to better understand normal postoperative hemodynamic changes.

A possible physiological mechanism of rectocele formation in women.

BACKGROUND: We aimed to determine the anorectal physiological factors associated with rectocele formation. METHODS: Female patients (N = 32) with severe constipation, fecal incontinence, or suspicion of rectocele, who had undergone magnetic resonance defecography and anorectal function tests between 2015 and 2021, were retrospectively included for analysis. The anorectal function tests were used to measure pressure in the anorectum during defecation. Rectocele characteristics and pelvic floor anatomy were determined with magnetic resonance defecography. Constipation severity was determined with the Agachan score. Information regarding constipation-related symptoms was collected. RESULTS: Mean rectocele size during defecation was 2.14 ± 0.88 cm. During defecation, the mean anal sphincter pressure just before defecation was 123.70 ± 67.37 mm Hg and was associated with rectocele size (P = 0.041). The Agachan constipation score was moderately correlated with anal sphincter pressure just before defecation (r = 0.465, P = 0.022), but not with rectocele size (r = 0.276, P = 0.191). During defecation, increased anal sphincter pressure just before defecation correlated moderately and positively with straining maneuvers (r = 0.539, P = 0.007) and defecation blockage (r = 0.532, P = 0.007). Rectocele size correlated moderately and positively with the distance between the pubococcygeal line and perineum (r = 0.446, P = 0.011). CONCLUSION: Increased anal sphincter pressure just before defecation is correlated with the rectocele size. Based on these results, it seems important to first treat the increased anal canal pressure before considering surgical rectocele repair to enhance patient outcomes.

Hyperammonemic Encephalopathy in a Patient with Pancreatic Neuroendocrine Tumor and Portosystemic Shunt.

Hyperammonemia can lead to encephalopathy and may be accompanied by a diagnostic dilemma. Imaging as well as biochemical analyses are the cornerstone for identifying possible underlying causes such as severe liver disease or urea cycle defect. We report a case of a patient that presented with neurological deficits based on hyperammonemia in the presence of a large pancreatic neuroendocrine tumor (PNET) and portosystemic shunts in the liver. Prior cases are rather scarce, and the exact mechanism is not fully understood. The case illustrates the added value of a multimodality imaging approach in patients presenting with hyperammonemia-induced encephalopathy.

Variation in the management of benign liver tumors: A European survey and case vignette study.

BACKGROUND: Management of focal nodular hyperplasia (FNH) and hepatocellular adenoma (HCA), is multidisciplinary and subject to practice variation. We aimed to evaluate variation in clinical management of FNH and HCA in Europe. METHODS: We distributed an online survey (November 2021-March 2022) among 294 European experts. The survey included questions on local practice and included eight clinical vignettes. The clinical vignettes focused on FNH or HCA management in the setting of sex, lifestyle modification, and pregnancy. RESULTS: The response rate was 32% and respondents included surgeons (38%), gastroenterologists/hepatologists (25%), radiologists (32%), and pathologists (1.6%) from ten European countries. We observed practice variation with regard to lifestyle modification and imaging follow-up in patients with FNH, and with regard to the management of HCA >5 cm before and during pregnancy. Finally, the management of HCA >5 cm after lifestyle modification deviated from EASL guideline recommendations. CONCLUSION: Our survey illustrates variability in FNH and HCA management in Europe. Several areas were identified for future research and guideline recommendations, including FNH follow-up and the management of HCA >5 cm. We propose the organization of Delphi consensus meetings to prioritize areas of research and update current guidelines to optimize management for all patients with benign liver tumors.

The Value of Deep Learning in Gallbladder Lesion Characterization.

BACKGROUND: The similarity of gallbladder cancer and benign gallbladder lesions brings challenges to diagnosing gallbladder cancer (GBC). This study investigated whether a convolutional neural network (CNN) could adequately differentiate GBC from benign gallbladder diseases, and whether information from adjacent liver parenchyma could improve its performance. METHODS: Consecutive patients referred to our hospital with suspicious gallbladder lesions with histopathological diagnosis confirmation and available contrast-enhanced portal venous phase CT scans were retrospectively selected. A CT-based CNN was trained once on gallbladder only and once on gallbladder including a 2 cm adjacent liver parenchyma. The best-performing classifier was combined with the diagnostic results based on radiological visual analysis. RESULTS: A total of 127 patients were included in the study: 83 patients with benign gallbladder lesions and 44 with gallbladder cancer. The CNN trained on the gallbladder including adjacent liver parenchyma achieved the best performance with an AUC of 0.81 (95% CI 0.71-0.92), being >10% better than the CNN trained on only the gallbladder (p = 0.09). Combining the CNN with radiological visual interpretation did not improve the differentiation between GBC and benign gallbladder diseases. CONCLUSIONS: The CT-based CNN shows promising ability to differentiate gallbladder cancer from benign gallbladder lesions. In addition, the liver parenchyma adjacent to the gallbladder seems to provide additional information, thereby improving the CNN's performance for gallbladder lesion characterization. However, these findings should be confirmed in larger multicenter studies.

Optimal radiological gallbladder lesion characterization by combining visual assessment with CT-based radiomics.

OBJECTIVES: Differentiating benign gallbladder diseases from gallbladder cancer (GBC) remains a radiological challenge because they can appear very similar on imaging. This study aimed at investigating whether CT-based radiomic features of suspicious gallbladder lesions analyzed by machine learning algorithms could adequately discriminate benign gallbladder disease from GBC. In addition, the added value of machine learning models to radiological visual CT-scan interpretation was assessed. METHODS: Patients were retrospectively selected based on confirmed histopathological diagnosis and available contrast-enhanced portal venous phase CT-scan. The radiomic features were extracted from the entire gallbladder, then further analyzed by machine learning classifiers based on Lasso regression, Ridge regression, and XG Boosting. The results of the best-performing classifier were combined with radiological visual CT diagnosis and then compared with radiological visual CT assessment alone. RESULTS: In total, 127 patients were included: 83 patients with benign gallbladder lesions and 44 patients with GBC. Among all machine learning classifiers, XG boosting achieved the best AUC of 0.81 (95% CI 0.72-0.91) and the highest accuracy rate of 73% (95% CI 65-80%). When combining radiological visual interpretation and predictions of the XG boosting classifier, the highest diagnostic performance was achieved with an AUC of 0.98 (95% CI 0.96-1.00), a sensitivity of 91% (95% CI 86-100%), a specificity of 93% (95% CI 90-100%), and an accuracy of 92% (95% CI 90-100%). CONCLUSIONS: Machine learning analysis of CT-based radiomic features shows promising results in discriminating benign from malignant gallbladder disease. Combining CT-based radiomic analysis and radiological visual interpretation provided the most optimal strategy for GBC and benign gallbladder disease differentiation. KEY POINTS: Radiomic-based machine learning algorithms are able to differentiate benign gallbladder disease from gallbladder cancer. Combining machine learning algorithms with a radiological visual interpretation of gallbladder lesions at CT increases the specificity, compared to visual interpretation alone, from 73 to 93% and the accuracy from 85 to 92%. Combined use of machine learning algorithms and radiological visual assessment seems the most optimal strategy for GBC and benign gallbladder disease differentiation.