Machine learning in liver surgery:Benefits and pitfalls

The application of machine learning(ML)algorithms in various fields of hepatology is an issue of interest.However,we must be cautious with the results.In this letter,based on a published ML prediction model for acute kidney injury after liver surgery,we discuss some limitations of ML models and how they may be addressed in the future.Although the future faces significant challenges,it also holds a great potential.

Medical management of metabolic and cardiovascular complications after liver transplantation

Liver transplantation represents the only curative option for patients with endstage liver disease,fulminant hepatitis and advanced hepatocellular carcinoma.Even though major advances in transplantation in the last decades have achieved excellent survival rates in the early post-transplantation period,long-term survival is hampered by the lack of improvement in survival in the late post transplantation period(over 5 years after transplantation).The main etiologies for late mortality are malignancies and cardiovascular complications.The latter are increasingly prevalent in liver transplant recipients due to the development or worsening of metabolic syndrome and all its components(arterial hypertension,dyslipidemia,obesity,renal injury,etc.).These comorbidities result from a combination of pre-liver transplant features,immunosuppressive agent side-effects,changes in metabolism and hemodynamics after liver transplantation and the adoption of a sedentary lifestyle.In this review we describe the most prevalent metabolic and cardiovascular complications present after liver transplantation,as well as proposing management strategies.

Perioperative thrombotic complications in liver transplantation

Although the perioperative bleeding complications and the major side effects of blood transfusion have always been the primary concern in liver transplantation(OLT),the possible cohesion of an underestimated intrinsic hypercoagulative state during and after the transplant procedure may pose a major threat to both patient and graft survival.Thromboembolism during OLT is characterized not only by a complex aetiology,but also by unpredictable onset and evolution of the disease.The initiation of a procoagulant process may be triggered by various factors,such as inflammation,venous stasis,ischemia-reperfusion injury,vascular clamping,anatomical and technical abnormalities,genetic factors,deficiency of profibrinolytic activity,and platelet activation.The involvement of the arterial system,intracardiac thrombosis,pulmonary emboli,portal vein thrombosis,and deep vein thrombosis,are among the most serious thrombotic events in the perioperative period.The rapid detection of occlusive vascular events is of paramount importance as it heavily influences the prognosis,particularly when these events occur intraoperatively or early after OLT.Regardless of the lack of studies and guidelines on anticoagulant prophylaxis in this setting,many institutions recommend such an approach especially in the subset of patients at high risk.However,the decision of when,how and in what doses to use the various chemical anticoagulants is still a difficult task,since there is no common consensus,even for highrisk cases.The risk of postoperative thromboembolism causing severe hemodynamic events,or even loss of graft function,must be weighed and compared with the risk of an important bleeding.In this article we briefly review the risk factors and the possible predictors of major thrombotic complications occurringin the perioperative period,as well as their incidence and clinical features.Moreover,the indications to pharmacological prophylaxis and the current treatment strategies are also summarized.

Prediction of hepatocellular carcinoma biological behaviorin patient selection for liver transplantation

Morphological criteria have always been considered the benchmark for selecting hepatocellular carcinoma(HCC)patients for liver transplantation(LT).These criteria,which are often inappropriate to express the tumor’s biological behavior and aggressiveness,offer only a static view of the disease burden and are frequently unable to correctly stratify the tumor recurrence risk after LT.Alpha-fetoprotein(AFP)and its progression as well as AFP-m RNA,AFP-L3%,des-γ-carboxyprothrombin,inflammatory markers and other serological tests appear to be correlated with post-transplant outcomes.Several other markers for patient selection including functional imaging studies such as18F-FDG-PET imaging,histological evaluation of tumor grade,tissue-specific biomarkers,and molecular signatures have been outlined in the literature.HCC growth rate and response to pre-transplant therapies can further contribute to the transplant evaluation process of HCC patients.While AFP,its progression,and HCC response to pretransplant therapy have already been used as a part of an integrated prognostic model for selecting patients,the utility of other markers in the transplant setting is still under investigation.This article intends to review the data in the literature concerning predictors that could be included in an integrated LT selection model and to evaluate the importance of biological aggressiveness in the evaluation process of these patients.

Strategies to optimize the use of marginal donors in liver transplantation

Liver transplantation is the treatment of choice for end stage liver disease, but availability of liver grafts is still the main limitation to its wider use. Extended criteria donors(ECD) are considered not ideal for several reasons but their use has dramatically grown in the last decades in order to augment the donor liver pool. Due to improvement in surgical and medical strategies, results using grafts from these donors have become acceptable in terms of survival and complications; nevertheless a big debate still exists regarding their selection, discharge criteria and allocation policies. Many studies analyzed the use of these grafts from many points of view producing different or contradictory results so that accepted guidelines do not exist and the use of these grafts is still related to non-standardized policies changing from center to center. The aim of this review is to analyze every step of the donationtransplantation process emphasizing all those strategies, both clinical and experimental, that can optimize results using ECD.

Hepatitis C virus related cirrhosis decreased as indication to liver transplantation since the introduction of direct-acting antivirals: A single-center study

AIM To evaluate waiting list(WL) registration and liver transplantation(LT) rates in patients with hepatitis C virus(HCV)-related cirrhosis since the introduction of direct-acting antivirals(DAAs).METHODS All adult patients with cirrhosis listed for LT at Padua University Hospital between 2006-2017 were retrospectively collected using a prospectivelyupdated database; patients with HCV-related cirrhosis were divided by indication for LT [dec-HCV vs HCV/hepatocellular carcinoma(HCC)] and into two interval times(2006-2013 and 2014-2017) according to the introduction of DAAs. For each patient, indications to LT, severity of liver dysfunction and the outcome in the WL were assessed and compared between the two different time periods. For patients receiving DAA-based regimens, the achievement of viral eradication and the outcome were also evaluated. RESULTS One thousand one hundred and ninty-four [male(M)/female(F): 925/269] patients were included. Considering the whole cohort, HCV-related cirrhosis was the main etiology at the time of WL registration(490/1194 patients, 41%). HCV-related cirrhosis significantly decreased as indication to WL registration after DAA introduction(from 43.3% in 2006-2013 to 37.2% in 2014-2017, P = 0.05), especially amongst decHCV(from 24.2% in 2006-2013 to 15.9% in 2014-2017, P = 0.007). Even HCV remained the most common indication to LT over time(289/666, 43.4%), there was a trend towards a decrease after DAAs introduction(from 46.3% in 2006-2013 to 39% in 2014-2017, P = 0.06). HCV patients(M/F: 43/11, mean age: 57.7 ± 8 years) who achieved viral eradication in the WL had better transplant-free survival(log-rank test P = 0.02) and delisting rate(P = 0.002) than untreated HCV patients. CONCLUSION Introduction of DAAs significantly reduced WL registrations for HCV related cirrhosis, especially in the setting of decompensated cirrhosis.

Endovascular treatment for early hepatic artery occlusion after liver transplantation: Angioplasty or stent

To the Editor:We read with great interest the recent article by Zhu et al.[1].In the study,the authors analyzed the outcomes of 26 patients diagnosed with hepatic artery occlusion(HAO)and treated with an endovascular approach(EVT)within the first 30 days after ortho-topic liver transplantation(LT).The median interval from LT to EVT was 7 days,most patients were treated with angioplasty and only two(7.7%)needed stent placement.The authors should be congrat-ulated as they achieved a 100%of success rate with an 80.8%of 1-year survival rate.

Potentially fatal haemobilia due to inappropriate use of an expanding biliary stent

AIM: To highlight the fatal complication caused by expanding biliary stents and the importance of avoiding use of expanding stent in potentially curable diseases.METHODS: Arteriobiliary fistula is an uncommon cause of haemobilia. We describe a case of right hepatic artery pseudoaneurysm causing arteriobiliary fistula and presenting as severe malena and cholangitis, in a patient with a mesh metal biliary stent. The patient had lymphoma causing bile duct obstruction.RESULTS: Gastroduodenoscopy failed to establish the exact source of bleeding and hepatic artery angiography and selective embolisation of the pseudo aneurysm successfully controlled the bleeding.CONCLUSION: Bleeding from the pseudo aneurysm of the hepatic artery can be fatal. Mesh metal stents in biliary tree can cause this complication as demonstrated in this case.So mesh metal stent insertion should be avoided in potentially benign or in curable conditions. Difficulty in diagnosis and management is discussed along with the review of the literature.

Late hepatic artery pseudoaneurysm:A rare complication after resection of hilar cholangiocarcinoma

We report an unusual pathological entity of a pseudoaneurysm of the right hepatic artery, which developed two years after the resection of a type 11 hilar cholangiocarcinoma and secondary to an excessive skeletonization for regional lymphadenectomy and neoadjuvant external-beam radiotherapy. After a sudden and massive hematemesis, a multidetector computed tomographic angiography （MDCTA） showed a hepatic artery pseudoaneurysm. Angiography with embolization of the pseudoaneurysm was attempted using microcoils with adequate patency of the hepatic artery and the occlusion of the pseudoaneurysm. A new episode of hematemesis 3 wk later revealed a partial revascularization of the pseudoaneurysm. A definitive interventional radiological treatment consisting of transarterial embolization （TAE） of the right hepatic artery with stainless steel coils and polyvinyl alcohol particles was effective and welltolerated with normal liver function tests and without signs of liver infarction.

Repair of a mal-repaired biliary injury:A case report

Iatrogenic bile-duct injury post-laparoscopic cholecystectomy remains a major serious complication with unpredictable long-term results. We present a patient who underwent laparoscopic cholecystectomy for gallstones, in which the biliary injury was recognized intraoperatively. The surgical procedure was converted to an open one. The first surgeon repaired the injury over a T-tube without recognizing the anatomy and type of the biliary lesion, which led to an unusual biliary mal-repair. Immediately postoperatively, the abdominal drain brought a large amount of bile. A T-tube cholangiogram was performed. Despite the contrast medium leaking through the abdominal drain, the mal-repair was recognized intraoperatively. The surgical procedure was converted to an open one. The first surgeon repaired the injury over a T-tube without recognizing the anatomy and type of the biliary lesion, which led to an unusual biliary mal-repair. Immediately postoperatively, the abdominal drain brought a large amount of bile. A T-tube cholangiogram was performed. Despite the contrast medium leaking through the abdominal drain, the mal-repair was unrecognized. The patient was referred to our hospital for biliary leak. Ultrasound and cholangiography was repeated, which showed an unanatomical repair （right to left hepatic duct anastomosis over the T-tube）,with evidence of contrast medium coming out through the abdominal drain. Eventually the patient was subjected to a definitive surgical treatment. The biliary continuity was re-established by a Roux-en-Y hepaticojejunostomy, over transanastomotic external biliary stents. The patient is now doing well 4 years after the second surgical procedure. In reviewing the literature, we found a similar type of injury but we did not find a similar surgical real-repair. We propose an algorithm for the treatment of early and late biliary injuries.

Robotic versus laparoscopic liver resection for huge (≥10 cm) liver tumors: an international multicenter propensity-score matched cohort study of 799 cases

Background:The use of laparoscopic(LLR)and robotic liver resections(RLR)has been safely performed in many institutions for liver tumours.A large scale international multicenter study would provide stronger evidence and insight into application of these techniques for huge liver tumours≥10 cm.Methods:This was a retrospective review of 971 patients who underwent LLR and RLR for huge(≥10 cm)tumors at 42 international centers between 2002-2020.Results:One hundred RLR and 699 LLR which met study criteria were included.The comparison between the 2 approaches for patients with huge tumors were performed using 1:3 propensity-score matching(PSM)(73 vs.219).Before PSM,LLR was associated with significantly increased frequency of previous abdominal surgery,malignant pathology,liver cirrhosis and increased median blood.After PSM,RLR and LLR was associated with no significant difference in key perioperative outcomes including media operation time(242 vs.290 min,P=0.286),transfusion rate rate(19.2%vs.16.9%,P=0.652),median blood loss(200 vs.300 mL,P=0.694),open conversion rate(8.2%vs.11.0%,P=0.519),morbidity(28.8%vs.21.9%,P=0.221),major morbidity(4.1%vs.9.6%,P=0.152),mortality and postoperative length of stay(6 vs.6 days,P=0.435).Conclusions:RLR and LLR can be performed safely for selected patients with huge liver tumours with excellent outcomes.There was no significant difference in perioperative outcomes after RLR or LLR.

Minimally invasive liver resection for huge(≥10cm)tumors:an international multicenter matched cohort study with regression discontinuity analyses

Background:The application and feasibility of minimally invasive liver resection(MILR)for huge liver tumours(≥10 cm)has not been well documented.Methods:Retrospective analysis of data on 6,617 patients who had MILR for liver tumours were gathered from 21 international centers between 2009-2019.Huge tumors and large tumors were defined as tumors with a size≥10.0 cm and 3.0-9.9 cm based on histology,respectively.1:1 coarsened exact-matching(CEM)and 1:2 Mahalanobis distance-matching(MDM)was performed according to clinically-selected variables.Regression discontinuity analyses were performed as an additional line of sensitivity analysis to estimate local treatment effects at the 10-cm tumor size cutoff.Results:Of 2,890 patients with tumours≥3 cm,there were 205 huge tumors.After 1:1 CEM,174 huge tumors were matched to 174 large tumors;and after 1:2 MDM,190 huge tumours were matched to 380 large tumours.There was significantly and consistently increased intraoperative blood loss,frequency in the application of Pringle maneuver,major morbidity and postoperative stay in the huge tumour group compared to the large tumour group after both 1:1 CEM and 1:2 MDM.These findings were reinforced in RD analyses.Intraoperative blood transfusion rate and open conversion rate were significantly higher in the huge tumor group after only 1:2 MDM but not 1:1 CEM.Conclusions:MILR for huge tumours can be safely performed in expert centers It is an operation with substantial complexity and high technical requirement,with worse perioperative outcomes compared to MILR for large tumors,therefore judicious patient selection is pivotal.