Current status of preoperative risk assessment for posthepatectomy liver failure in patients with hepatocellular carcinoma

Abstract Liver resection is an effective therapeutic option for patients with hepatocellular carcinoma. However, posthepatectomy liver failure (PHLF) remains a major cause of hepatectomy‐related mortality, and the accurate prediction of PHLF based on preoperative assessment of liver functional reserve is a critical issue. The definition of PHLF proposed by the International Study Group for Liver Surgery has gained acceptance as a standard grading criterion. Liver function can be estimated using a variety of parameters, including routine blood biochemical examinations, clinical scoring systems, dynamic liver function tests, liver stiffness and fibrosis markers, and imaging studies. The Child–Pugh score and model for end‐stage liver disease scores are conventionally used for estimating liver decompensation, although the alternatively developed albumin‐bilirubin score shows superior performance for predicting hepatic dysfunction. Indocyanine green clearance, a dynamic liver function test mostly used in Japan and other Asian countries, serves as a quantitative estimation of liver function reserve and helps determine indications for surgical procedures according to the estimated risk of PHLF. In an attempt to improve predictive accuracy, specific evaluation of liver fibrosis and portal hypertension has gained popularity, including liver stiffness measurements using ultrasonography or magnetic resonance elastography, as well as noninvasive fibrosis markers. Imaging modalities, including Tc‐99m‐labeled galactosyl serum albumin scintigraphy and gadolinium‐enhanced magnetic resonance imaging, are used for preoperative evaluation in combination with liver volume. This review aims to provide an overview of the usefulness of current options for the preoperative assessment of liver function in predicting PHLF.


| INTRODUC TI ON
Hepatocellular carcinoma (HCC) is the most common primary liver cancer and commonly arises in patients with chronic liver disease. 1 There are many therapeutic options for HCC, including transplantation, ablation, transarterial chemoembolization, and systemic therapy.Liver resection is widely accepted as an effective curative treatment for HCC, 2 and the capacity of the liver to regenerate and restore its function allows the removal of the part of the hepatic parenchyma with lesions.However, resection of an excessive volume of a diseased liver could result in insufficient functional reserve of the remnant liver, leading to posthepatectomy liver failure (PHLF). 35][6] Furthermore, PHLF significantly affects the long-term survival of HCC patients. 7,80][11][12] Preventing the occurrence of PHLF by accurately estimating the liver functional reserve and determining the permissive future remnant volume during preoperative planning is essential.
Much effort has been made to develop selection criteria to identify patients at high risk of PHLF; however, accurate predictors of PHLF based on preoperative evaluation of liver function remain controversial.The aim of this narrative review is to provide an overview of the utility of current options for preoperative assessment of liver function and the performance of each parameter in predicting PHLF.

| LITER ATURE S E ARCH
The predictors of PHLF are heterogeneous, and the criteria for surgical indications for patients with HCC differ greatly across institutions and countries.In addition, advances in surgical techniques and diagnostic tools, including imaging modalities and detection of novel biomarkers, have impacted perioperative management and postoperative outcomes in patients with HCC, further complicating the preoperative prediction of PHLF.This article aims to review the current status and trends of preoperative risk assessment for liver resection and comprehend this heterogeneity.Because of a broad variety of liver function tests and the substantial heterogeneity of studies, a formal systematic review was not conducted.Rather, a pragmatic electronic literature search in the PubMed and Medline databases was performed using the keywords "liver failure OR liver insufficiency OR liver dysfunction OR liver decompensation" AND "posthepatectomy OR postoperative OR hepatectomy OR liver resection."We particularly focused on literature published within the last 15 years.Only studies in humans published in English were considered.Studies were excluded if they fulfilled any of the following criteria: (1) a focus on liver resection for non-HCC tumors, such as cholangiocarcinoma and metastatic cancer; (2) inclusion of only long-term outcomes; (3) examination of correlation among the parameters or grading systems but not with outcomes; (4) lack of precise description of the definition of postoperative outcomes; and (5)   no report on the detailed methodology.

| DEFINITI ON OF PHLF
The incidence of PHLF has been reported to range from 1.2% to 32%, which potentially reflects the differences in patient demographics, pathology of underlying diseases, procedures performed, and the definition of PHLF. 13Various definitions of PHLF have been proposed, some of which have gained wide acceptance (Table 1).

Balzan et al. in 2005
showed that a combination of prothrombin time <50% and bilirubin >50 μmol/L on postoperative day 5 was an accurate predictor of the risk of hepatectomy-related mortality. 14These are termed the 50-50 criteria, and their usefulness for the early diagnosis of PHLF has been validated. 15Mullen et al. 16 proposed that a peak serum bilirubin concentration >7 mg/dL is a powerful predictor of 90-day mortality and complications after major hepatectomy.Although the peak bilirubin criterion showed TA B L E 1 Representative definitions for PHLF.better predictive performance (area under the curve, 0.982; sensitivity, 93.3%; and specificity, 94.3%) than the 50-50 criteria, the exclusion of patients with cirrhosis in this analysis has raised questions about its validity.
The International Study Group of Liver Surgery (ISGLS) proposed a standardized definition for PHLF in 2011. 13They defined PHLF as an increased prothrombin time-international normalized ratio (PT-INR) and concomitant hyperbilirubinemia on or after postoperative day 5.The severity is categorized into three grades, as follows: Grade A, resulting in abnormal laboratory parameters but requiring no deviation from standard care; Grade B, resulting in a deviation from regular clinical management and requiring noninvasive treatment; and Grade C, resulting in a deviation from the regular clinical management and requiring invasive treatment.The perioperative mortality rates of patients with grades A, B, and C PHLF are 0%, 12%, and 54%, respectively. 13e ISGLS definition has a higher sensitivity for predicting hepatectomy-related mortality than the 50-50 criteria and the peak bilirubin >7 mg/L criterion. 17Therefore, the ISGLS definition has gained acceptance as a standard grading criterion for PHLF and has been universally used in studies involving liver resection.

| PREOPER ATIVE A SS E SS MENT FOR LIVER FUN C TIONAL RE S ERVE
The risk assessment for PHLF is mainly based on the optimization of preoperative liver function reserve. 3Liver function can be estimated using various preoperative parameters, including blood biochemical examinations, clinical scoring systems, dynamic liver function tests, fibrosis markers, including liver stiffness (LS), imaging studies, and parameters of portal hypertension, including the hepatic venous pressure gradient (HVPG) 5,22 (Table 2, Figure 1).

| ROUTINE B LOOD B I O CHEMI C AL TE S TS AND CLINI C AL SCORE S
The Child-Pugh score is an essential tool for stratifying the prognosis of patients with HCC and as a general guide for indication for surgical resection 23,24 and is the gold standard grading system for liver function.This score is based on five simple parameters: encephalopathy, ascites, serum total bilirubin level, serum albumin level, and prothrombin time. 64The MELD score, which incorporates renal function and general liver functional indicators, was originally designed to predict survival in patients with cirrhosis after insertion of a transjugular intrahepatic portosystemic shunt 65 and has been used to prioritize candidates for liver transplantation.The MELD score has also been applied for the early prediction of postoperative morbidity and mortality in patients undergoing liver resection, 27 but its predictive performance in patients without advanced cirrhosis is controversial. 12,66though the Child-Pugh score has conventionally been used for risk assessment of surgical treatment for patients with HCC, it has limitations, such as subjective parameters and arbitrary cutoff points.
The ALBI score has emerged as an evidence-based scoring system to assess liver function in patients with HCC.This score includes only the albumin and bilirubin values and is therefore more objective. 31The ALBI score has been widely accepted as a prognostic tool and has a good correlation with survival, time to recurrence, and tolerability of surgical, locoregional, and systemic therapies for HCC. 67,6835]69 However, the utility of the preoperative ALBI grade for estimating the risk for PHLF has been demonstrated mainly in retrospective studies, and further investigations with highquality designs to evaluate its predictive accuracy are needed. 38though the ALBI score has gained popularity because of its simple and objective parameters for predicting prognosis in HCC patients, there have been attempts to determine more accurate predictive models for PHLF using routine blood tests, including prothrombin time, aminotransferase, and platelet count, as well as specific indicators, such as the ICG test, LSM, and imaging modalities. 6Importantly, the platelet count, which is a conventional item not included in the ALBI score, is known to have an impact on the postoperative outcome for HCC, as it reflects clinically significant portal hypertension (CSPH). 70In addition, the platelet count is an essential component of noninvasive diagnostic tools for liver fibrosis, including the APRI and FIB-4 index.Multiple studies have identified the platelet count as a significant predictor of PHLF, 6,71 and models that incorporate the platelet count combined with albumin, 72 the ALBI score, 73 ICG test, 74 and other predictive factors 5,75 have shown a better predictive performance than the ALBI score.Of note, these models include resection volume or future liver remnant (FLR) as a parameter, which allows surgeons to plan surgical procedures based on liver functional evaluation using simple and conventional factors.Recently, nomograms have been developed based on multiple independent preoperative predictors, which enable multidisciplinary risk assessment to determine the indications for hepatectomy in patients with HCC 20,76,77 (Table 3).

| ICG test
The clearance of intravenously administered exogenous substances that are metabolized or excreted via liver perfusion has been used to quantitatively examine liver function. 40The ICG retention test is as a well-accepted method for preoperatively assessing liver functional reserve and is routinely performed in Japan and other Asian countries. 41 is a dynamic method that measures the hepatic clearance of ICG 15 min after its intravenous injection (ICG-R15).ICG clearance is usually delayed in patients with liver damage, and an increase in the ICG-R15 reflects the degree of liver dysfunction.The Makuuchi criteria, a decisional algorithm for the extent of hepatectomy according to the ICG-R15, have reduced hepatectomy-related morbidity and mortality, especially during the developmental stage of liver surgery in Japan. 39e ICG test has gained popularity as a preoperative liver functional test, and it has superior predictive performance compared to the Child-Pugh and MELD scores. 28Although the ICG-R15 is not a linear parameter, the plasma disappearance rate of ICG (KICG) is useful for the quantification of liver function when combined with the estimated-FLR as the KICG of the remnant liver (remKICG), 78 which is correlated with the occurrence of PHLF not only in HCC patients 79,80 but also in those with biliary cancer 81 and individuals who have undergone portal vein embolization. 82Some studies have used predictive models incorporating the ICG-R15, 74,[83][84][85] including the Albumin-Indocyanine Green Evaluation model, 86,87 which has better performance than the Child-Pugh score but is comparable with the ALBI grade for predicting PHLF (Table 3).The potential limitation of the ICG clearance test is that its result is affected by biliary obstruction and hemodynamic alterations, such as intrahepatic shunt, portal hypertension, and thrombosis.

| Liver maximum capacity
The liver maximum capacity (LiMAx) test evaluates hepatic metabolism by measuring the 13  for the diagnosis and prognostication of nonalcoholic steatohepatitis (NASH).4][95][96] The APRI and FIB-4 index are also useful for estimating liver functional reserve, as they correlate with the risk of perioperative mortality 97 and have better predictive accuracy for PHLF than the MELD and Child-Pugh scores 34,98-102 (Table 4).

| Specific liver fibrosis markers
Other specific markers for liver fibrosis examined by blood tests include hyaluronic acid, type IV collagen 7S, and M2BPGi.
4][105][106][107] Type IV collagen 7S is a biomarker of liver fibrogenesis, and its serum concentration correlates with hepatic dysfunction following liver resection. 108,109More recently, M2BPGi, which is a unique fibrosis-related glyco-alteration detected by a glycan sugar chain-based immunoassay, has been proposed as a novel marker for liver fibrosis. 110Serum M2BPGi levels have a predictive accuracy for the diagnosis of liver fibrosis progression comparable to that of LSM and superior to that of APRI, hyaluronic acid, and type IV collagen 7S. 50,51Further, M2BPGi can predict PHLF better than other preoperative parameters, including KICG, especially in patients with HCVrelated HCC. 111 Taken together, the results of these studies indicate that serum markers for liver fibrosis are useful for the preoperative assessment of liver functional reserve.However, these results are mostly based on retrospective analysis in a limited number of centers, and further well-designed prospective studies are required to determine the markers with the best accuracy.

| L SM
Liver stiffness measurement has been widely accepted as a noninvasive assessment procedure for liver fibrosis and is an alternative to liver elastographic tissue map.There is good to excellent agreement across different ultrasonographic systems for LSM. 116,1179][120] The potential limitations of ultrasonographic elastography include the need for training, limited availability, high cost, and failure due to artifacts, operator inexperience, ascites, obesity, narrow intercostal space, and confounders, such as inflammation, venous congestion, cholestasis, non-fasting, and exercise. 121

| MRE
Magnetic resonance elastography was developed as a noninvasive imaging method for quantifying liver fibrosis with high accuracy. 122It can be easily incorporated into current abdominal MRI protocols and is capable of providing a stiffness map of the entire liver as well as a comprehensive evaluation in conjunction with MRI across the abdomen.[125][126] However, there are limitations, including cost, availability, and patientdependent factors such as the presence of magnetically susceptible implants, compliance with breath-hold, and claustrophobia. 121

| Usefulness of LSM for prediction of PHLF
Liver stiffness measurement techniques have the potential to be applied to risk assessment for PHLF based on the significant correlation between the progression of liver fibrosis and extent of liver dysfunction. 127The utility of VCTE for predicting postoperative complications has been described. 128,129A prospective study demonstrated that ARFI-based LSM is useful for predicting PHLF based on the ISGLS definition, with higher accuracy than conventional preoperative tests, including KICG, and other fibrosis markers, such as hyaluronic acid, type IV collagen, the APRI, and the FIB-4 index. 130The PHLF prediction model of the ARFI value incorporating FLR allows surgeons to make decisions regarding surgical procedures based on the estimation of permissive resection volume and has superior predictive performance to remKICG.2][143][144][145] LSM by MRE can also be used as a risk assessment modality for major complications after liver resection, including PHLF, and has good predictive performance compared to the ICG test, MELD score, APRI, FIB-4 index, and VCTE value [146][147][148][149][150][151] (Table 5).LSM is a promising technique for evaluating preoperative liver functional reserve, and further investigations are required to develop LSM-based criteria for determining surgical indications.

| Hepatobiliary scintigraphy
Tc-99m-GSA scintigraphy is a well-accepted imaging modality for assessing liver function. 152Tc-99m-GSA is a liver scintigraphy agent that binds to the asialoglycoprotein receptor on hepatocytes. 153Hepatic dysfunction as detected by the abnormal distribution of functioning hepatocytes with Tc-99m-GSA is correlated with hepatic disorders, including steatosis, fibrosis, and necrosis due to chronic liver injury.
The HH15 (blood clearance ratio), representing retention of the tracer in the blood, and LHL15 (hepatic uptake ratio), representing uptake of the tracer in the liver, are commonly used as parameters.One of the benefits of Tc-99m-GSA scintigraphy is its ability to quantify the function of specific parts of the liver as well as the function of the entire liver, 154 which is useful for estimating the function of the FLR during preoperative risk assessment and decision-making in surgical procedures.Tc-99m-GSA is also applicable for patients undergoing portal vein embolization or two-stage procedures, as liver function in such patients is not uniform across liver segments. 1553][164][165] Tc-99m-mebrofenin scintigraphy is also used to obtain functional liver imaging to estimate FLR function and the risk of PHLF after major hepatectomy and two-stage procedures. 83,166,167-99m-mebrofenin scintigraphy correlates with ICG retention because its absorption, excretion, and lack of hepatic biotransformation are similar to those of ICG. 168

| Gd-EOB-DTPA-enhanced MRI
Gd-EOB-DTPA-enhanced MRI (EOB-MRI) has been proposed for the evaluation of liver functional reserve, as the specific uptake of Gd-EOB-DTPA by hepatocytes reflects their function.2][173][174][175] The advantages of EOB-MRI include high spatial resolution and combined anatomical and functional assessment, which enable the evaluation of regional liver function and diagnosis of hepatic lesions prior to liver resection.Moreover, it is reasonable to incorporate the FLR volume for precise estimation of postoperative residual function.CSPH is not necessarily a contraindication for liver resection 10 because minor resection in patients with moderate CSPH yields competitive survival outcomes. 23The extent of hepatectomy should be determined based on preoperative risk assessment of the severity of portal hypertension as well as other liver functional indicators to prevent the occurrence of PHLF, especially in patients with cirrhosis.HPVG is the gold-standard direct assessment of portal hypertension and a significant predictor of hepatic decompensation and patient survival.Preoperative HVPG is associated with postoperative liver dysfunction and mortality after liver resection in patients with HCC and liver cirrhosis. 53CSPH is defined as an HVPG >10 mg, and its relevance to a higher risk of PHLF has been proposed. 177Nevertheless, HVPG is not routinely measured in clinical practice, as it is a potentially invasive technique with complex procedures and limited reproducibility due to inter-operator variability.

| Noninvasive assessment for CSPH
Alternative noninvasive parameters have been used to assess CSPH.The standard surrogate criteria include the presence of gastroesophageal varices or thrombocytopenia (platelet count <100 000/mL) and splenomegaly (diameter >12 cm). 70LS can also estimate portal hypertension, allowing highly accurate noninvasive identification of CSPH, 54 particularly when combined with spleen diameter and platelet count. 1784][185] More recently, CT-based LSN has been proposed as a diagnostic tool for detecting CSPH. 186The LSN score is associated with severe complications and PHLF after liver resection. 22,62antification of LSN can be performed using routine CT images, and it may be a promising method for assessing liver functional reserve in the preoperative setting; however, further large cohort studies are needed to confirm its accuracy.

| CON CLUS ION
Accurate prediction of PHLF baswed on preoperative assessment of liver functional reserve remains challenging, and much effort has

114 8. 1 |
biopsy.The high diagnostic accuracy of LSM is based on the pathogenesis of liver fibrosis, in which the deposition of excessive amounts of extracellular matrix due to chronic injury increases tissue elasticity, enabling quantification of the extent of liver damage.LS is also affected by inflammation, passive venous congestion, portal hypertension, and biliary obstruction, which are potential confounders.Recently, LSM has gained popularity as a noninvasive assessment of liver fibrosis in patients with NAFLD and NASH because of the rapid increase in NAFLD/ NASH-related end-stage liver diseases worldwide. 112-Ultrasonographic elastography Liver stiffness measurement is performed using ultrasonographyor MRI-based techniques.Ultrasonographic elastography includes vibration-controlled transient elastography (VCTE), point shear wave elastography (pSWE), acoustic radiation force impulse (ARFI), and twodimensional shear wave elastography (2D-SWE). 45,115VCTE has been validated in a large number of studies with good reproducibility, but it has the limitation of a lack of imaging.pSWE and 2D-SWE can be performed in combination with real-time standard B-mode imaging in which the region of interest can be adjusted by the operator; the former technique acquires point measurements, and the latter yields a 2D regional liver function reserve than other modalities, such as Tc-99m-GSA scintigraphy, because of its superior spatial resolution.176EOB-MRI-based parameters have good to excellent accuracy (AUC 0.75-0.96) in predicting PHLF, and they have better accuracy than the ICG test, MELD score, and ALBI score.[173][174][175]However, the heterogeneity in the variance of EOB-MRI-derived parameters and limited sample size in the existing studies highlight the necessity for well-designed, prospective, multicenter studies with large sample sizes.Further, liver functional imaging, including hepatobiliary scintigraphy and MRI, are more costly than non-imaging tests.Further investigations are necessary to determine whether the potential benefits of these imaging tests can overcome this disadvantage because of the overall cost reduction due to the improvement of patient outcomes.10| A SS E SS MENT FOR P ORTAL HYPERTENS ION 10.1 | HVPGClinically significant portal hypertension is strongly correlated with hepatic decompensation and mortality in HCC patients.
been made to develop criteria to ensure the safety of liver resection in patients with HCC.Multiple studies have evaluated the predictive performance of various preoperative parameters, which are broadly categorized as clinical scores based on routine blood tests, dynamic liver function tests, LS and noninvasive fibrosis markers, liver function imaging, and biomarkers for CSPH.These categories are not completely independent, and some parameters overlap across groups.For example, platelet count, generally included in routine blood tests, is a composition of fibrosis markers, including the APRI and FIB-4 index, and it is also useful for estimating the presence of CSPH, and LSM, which is mostly based on imaging analysis, accurately evaluates CSPH and liver fibrosis.Additionally, imaging modalities, including EOB-MRI and hepatobiliary scintigraphy, are also categorized as dynamic liver function tests.Of note, imaging techniques can simultaneously evaluate function and anatomy and preoperatively provide useful information for estimating safe and feasible FLR.Although the superiority of single predictors is controversial, a combination of parameters with consideration of their role in each category should enable comprehensive risk assessment for PHLF, leading to the proposal of predictive models based on the clinical background of individual patients.This review helps organize the current status of the preoperative prediction of PHLF, highlighting the necessity for further well-designed, large investigations to identify the best combination of parameters for the establishment of novel criteria for liver resection.E TH I C S S TATEM ENTS Approval of the research protocol: N/A.Informed Consent: N/A.Registry and the Registration No. of the study/trial: N/A.Animal Studies: N/A.O RCI D Takahiro Nishio https://orcid.org/0000-0002-9676-7235R E FE R E N C E S

TA B L E 2 Categories of liver function tests. Liver function tests Items/components
90,91:12CO 2 ratio in the exhaled breath, which is derived from the rate of metabolism of intravenously injected 13 Cmethacetin.The LiMAx test result strongly correlates with liver function reserve, and a preoperative volume/function analysis combining FLR and LiMAx enables an accurate estimation of remnant liver function prior to surgery.88,89TheLiMAxtest has gained acceptance mostly in Western countries, and the LiMAx decision tree algorithm has improved preoperative assessments for PHLF and postoperative outcomes.90,91 48| S ERUM MARKER S FOR LIVER FIB ROS IS7.1 | APRI and FIB-4 indexLiver fibrosis is a common consequence of chronic liver injury, and the extent of fibrosis is highly correlated with liver functional reserve and prognosis in patients with HCC.Liver biopsy is the standard option for evaluating liver fibrosis, but it is an invasive procedure and has several limitations, such as complications, sampling errors, intra-and interobserver variability, and expense.92Toaddresstheselimitations,noninvasiveliverfibrosismarkerssuitablefor routine use have been developed.The APRI (aspartate transaminase/[upper limit of normal] × 100/platelet count[10 9/L])46and the FIB-4 index ([age (in years) × aspartate aminotransferase (U/L)]/[platelet count(10 9/L) × alanine aminotransferase (U/L) 1/2 ])48can be commonly assessed using simple and conventional parameters.These liver fibrosis indices show excellent accuracy in predicting significant fibrosis and cirrhosis, and they have recently gained attention as noninvasive tools Predictors and prediction models for PHLF based on routine tests.
Predictors and prediction models for PHLF based on imaging and CSPH parameters.