Mihil Ppmihil , , Mihil Piznis , Nigel D Heton

a Department of Hepato-Pancreato-Biliary Surgery, Royal Blackburn Hospital, Blackburn BB2 3HH, UK

b Department of General Surgery, Whittington Hospital, London N19 5NF, UK

c Department of Liver Transplant and Hepato-Pancreato-Biliary Surgery, Institute of Liver Studies, Kings Health Partners at King’s College Hospital NHS Trust, London SE5 9RS, UK

Keywords: Small-for-size syndrome Liver resection Future liver remnant

ABSTRACT Background: Primary and secondary liver tumors are not always amenable to resection due to location and size. Inadequate future liver remnant (FLR) may prevent patients from having a curative resection or may result in increased postoperative morbidity and mortality from complications related to small-for- size syndrome (SFSS).Data sources: This comprehensive review analyzed the principles, mechanism and risk factors associated with SFSS and presented current available options in the evaluation of FLR when planning liver surgery. In addition, it provided a detailed description of specific modalities that can be used before, during or after surgery, in order to optimize the conditions for a safe resection and minimize the risk of SFSS.Results: Several methods which aim to reduce tumor burden, preserve healthy liver parenchyma, induce hypertrophy of FLR or prevent postoperative complications help minimize the risk of SFSS.Conclusions: With those techniques the indications of radical treatment for patients with liver tumors have significantly expanded. The successful outcome depends on appropriate patient selection, the indi- vidualization and modification of interventions and the right timing of surgery.

Introduction

Over the past 70 years, liver resection has been shown to be an effective treatment for primary and secondary liver tumors. The limits of resection are defined by patient safety, technical feasibil- ity and long-term outcome [1] . The aim of surgery is to achieve complete oncological clearance. The extent of resection is deter- mined by tumor size and location, the function and maintenance of the vascular integrity of future liver remnant (FLR) and the qual- ity of the liver (fat, chronic liver disease, chemotherapy induced liver injury), which in turn is related to the FLR/body weight ratio and the presence of underlying portal hypertension [ 2 , 3 ].

Small-for-size syndrome (SFSS) is a term used to denote a clin- ical condition, where the residual liver post-surgery (resection or transplant) is insufficient for the recovery of normal liver function. The term was first used in living donor liver transplantation to de- scribe a graft that was too small for a given recipient [ 4 , 5 ]. Cur- rently, the term may be used for all cases of recipient liver failure following orthotopic or auxiliary transplantation including whole cadaveric grafts or partial grafts from living donors or split cadav- eric donors. The likelihood of the syndrome is multifactorial and includes severity of decompensation of recipient chronic liver dis- ease, degree of portal hypertension, graft size in relation to body weight of the recipient, and conditions that predispose to marginal grafts (fatty infiltration, elevated liver enzymes, prolonged systemic instability, older age, longer ischemic time). SFSS syndrome also applies in living donor patients when they develop liver insuffi- ciency after donation [5–7] .

In a similar way liver dysfunction following extended liver re- section is due to insufficient residual liver mass in relation to the body surface and weight of the patient and the condition of the liver prior to resection [5–7] . In both clinical scenarios (re- section or transplant) the syndrome is characterized by prolonged cholestasis, persistent coagulopathy [international normalized ratio (INR)＞1.5] and ascites, portal hypertension and gastrointestinal bleeding which manifest 1-2 weeks postoperatively and in the ab- sence of liver recovery, and death from recurrent sepsis 4-6 weeks later [ 8 , 9 ]. An increase in the prothrombin time and hyperbiliru- binemia, on or after the 5th postoperative day (POD) is usually in- dicative of SFSS. Serum bilirubin＞50μmol/L and INR＞1.7 on POD5 are associated with 50% survival (50-50 criteria, Table 1 ). In the setting of the syndrome, various models have been developed to predict the outcome ( Table 1 ) [10–12] .

Table 1 Calculation and clinical models diagnosing and predicting severity and mortality of small-for-size syndrome.

Table 2 Risk factors associated with small for size syndrome in liver resection.

The histological appearance of SFSS includes cholestasis, hepa- tocyte ballooning, mitochondrial swelling, collapse of the space of Disse, with areas of regeneration adjacent to those with ischemic necrosis [ 9 , 13 ]. The mechanisms of injury include increased por- tal flow and hyperperfusion through the small liver, with recip- rocal reduction in hepatic arterial inflow, higher portal pressure and sinusoidal resistance with biliary injury/dysfunction. The me- chanical injury is exacerbated by the inflammatory and immune response with an element of ischemia-reperfusion injury, leading to ischemic necrosis [ 6 , 9 , 13 ].

The reported incidence of liver dysfunction following liver re- section varies between 0.7% and 34% reflecting the quality of the liver, the extent of resection and heterogeneity of the patient pop- ulation and tumor type and stage [14] . Overall mortality of post- hepatectomy SFSS is＜5% and the morbidity is 15%-30% [15] . Of note, resection of four or more segments results in a higher in- cidence of postoperative mortality (7.4% vs. 2.7%) and morbidity (25.6% vs. 6.4%) [15] . The functional liver remnant volume is as- sociated with severity of liver dysfunction and postoperative com- plications. The morbidity is associated with an FLR of less than 40% and severe dysfunction with an FLR of less than 25% [16] .

Risk factors associated with SFSS

The risk factors associated with SFSS can be divided into pa- tient and surgical procedure related as shown in Table 2 [ 7 , 17 ]. Pa- tient factors include the quality and function of the liver particular in the presence of pre-existing liver disease. Steatosis, steatohep- atitis, fibrosis, chronic viral hepatitis B and C, cholangitis, alcohol related liver disease, which are associated with portal hyperten- sion and established cirrhosis represent serious limitations to liver resection [ 18 , 19 ]. Patients with hepatocellular carcinoma (HCC) and liver cirrhosis are at particular risk of SFSS, which restricts the number of cases suitable for resection [20] .

Chemotherapy may induce steatohepatitis and/or sinusoidal in- jury. These patients require more residual liver mass to ensure postoperative recovery [21] . Patients with tumors obstructing the biliary tree with associated cholestasis show decreased metabolic and regenerative capacity. Other risk factors including age (＞69 years), malnutrition, diabetes mellitus and male sex, influence remnant liver regeneration, though the mechanisms are not fully understood [22–24] . During surgery the extent of resection is im- portant and is primarily dictated by tumor size, number and loca- tion. Significant blood loss and hemodynamic instability requiring significant transfusion and inotropic support result in additional liver injury and have been associated with increased liver dys- function and SFSS [25] . Prolonged application of inflow occlusion (Pringle manoeuvre) or total vascular occlusion may also result in a significant ischemia-reperfusion injury [ 26 , 27 ]. The complexity of surgery (e.g., vascular/biliary reconstruction,insitucold perfusion,exvivoresection) and their associated technical complications (e.g. portal vein/hepatic vein thrombosis/narrowing) are also risk factors of developing SFSS [ 28 , 29 ].

Evaluation of potential FLR

Preoperative helical imaging [computed tomography (CT) or magnetic resonance imaging (MRI)] is widely accepted for diagno- sis and staging of liver tumors. In addition, it allows for planning surgical resection, estimation of the residual liver volume, assess- ment of response to previous neoadjuvant treatment, estimation of the degree of fatty infiltration and comparison before and after potential volume manipulations [ 21 , 30 , 31 ]. Standardized volumet- ric calculation of FLR is related to the body surface area (BSA) and body weight (BW) and a number of formulae and equations have been employed aiming to predict the volume of liver that can be safely resected ( Table 3 ) [ 6 , 32 ].

Table 3 Formulae to calculate volume of the hepatic remnant.

Imaging can also provide details regarding the contour and con- sistency of the parenchyma and look for signs of portal hyper- tension that would indicate chronic liver disease and ascites as- sociated with decompensation [30] . Technical difficulties occur in the presence of undetected lesions, marked intrahepatic biliary di- latation, previous radiological intervention (e.g. ablation), vascu- lar obstruction and hypertrophic/atrophic segments due to tumor growth [6] . In such cases dynamic tests such as Gadox- etic acid-enhanced MR,99mTc-mebrofenin hepatobiliary scintig- raphy (HBSBSA), and technetium 99m-galactosyl human serum albumin single-photon emission computed tomography, have been proposed to increase the accuracy of functional FLR assessment [33–35] .

Quality and function of the residual liver will not always cor- relate with volume when there is evidence of liver disease or pa- tient factors which influence function (obesity, diabetes, metabolic syndrome, the effect of systemic chemotherapy). Traditional scor- ing systems [Child-Pugh, model for end-stage liver disease (MELD)] are routinely utilized to assess the grade and prognosis of cirrho- sis. Despite having similar prognostic values in most cases, par- ticularly when assessing patients undergoing surgery, Child-Pugh score has higher specificity than MELD [ 8 , 14 , 36 ]. The extent of injury after chemotherapy is often difficult to estimate and even with apparently normal liver function tests significant injury may be seen intraoperatively, leading to a change in surgical plans [21] . Platelet-based liver scores (PBLS) can be an additional tool to as- sess chemotherapy related liver injury and liver fibrosis. Those in- clude aspartate aminotransferase to platelets ratio index (APRI) and the fibrosis-4 score (FIB-4) [34] . Those formulae have been used in the past to evaluate degree of fibrosis and cirrhosis in chronic hepatitis C patients and appear to have potential value in assess- ing risk of post-hepatectomy liver dysfunction in HCC resections, and severity of sinusoidal injury after treatment with oxaliplatin for colorectal liver metastases (CRLM) [34] . Other tests such as the LiMAx test, are based on the hepatic metabolism of13C-methacetin into acetaminophen and13CO 2 . Measurement and analysis of ex- haled13CO 2 by a modified nondispersive isotope-selective infrared spectroscope can be used to assess baseline liver function. The esti- mated postoperative LiMAx result based on the predicted FLR per- centage may address potential risk of SFSS prior to resection. It has been suggested if estimated postoperative LiMAx value is＜80–100μg/kg/h (preoperative normal value＞300μg/kg/h) then sig- nificant risk for post-hepatectomy liver failure exists [37] . Transient elastography as a means of detection of liver cirrhosis/fibrosis has been evaluated in the past and potentially may identify patients with significant fibrosis who are at high risk for resection [38] . In one meta-analysis by Geng et al. of 10 504 patients, the sensitivity of transient elastography for detecting liver fibrosis is 81% [38] .

Indocyanine green (ICG) clearance is a dynamic test that quan- titatively evaluates liver function, and one study has shown that it may be a useful predictor of hospital mortality after major liver re- section [39] . When liver function is impaired; ICG retention rate at 15 min (ICGR15) increases (＞15%) indicating that major liver re- section will not be tolerated [ 30 , 40 ]. This test is routinely used in many Asian centers prior to resection for primary and secondary liver tumors, but less frequently in Europe [ 41 , 42 ]. Some scholars suggest it has value for patients with borderline functional FLR and particularly those with evidence of chronic liver disease [43] . How- ever, the ICG clearance test accuracy can be affected by alterations in parenchymal blood flow particularly with lesions obstructing the biliary tree or where there is intrahepatic shunting. Some scholars suggest performing the ICG test intraoperatively after occlusion of the inflow and outflow of the segments to be resected, to try to improve its accuracy [44] .

Many researchers advocate an acceptable predicted FLR of 25% in a healthy liver and 40% when function is compromised [45] . Optimizing the patient’s condition prior to the surgery helps im- prove outcomes. This includes appropriate delaying of surgery to allow for the liver to recover from previous systemic treatment, effective biliary drainage in the presence of obstruction and nu- trition support [15] . Poor nutritional status may negatively affect liver regeneration postoperatively and is a recognized risk factor for SFSS. Patients with large volume disease, previous gastrointesti- nal surgery, systemic chemotherapy, liver cirrhosis and those who would require complex reconstruction and have a borderline pre- dicted FLR are at higher risk [17] . Not all patients, however, are safe at these levels and additional actions are required to mini- mize the risk of SFSS and allow for more patients to be resected with a greater degree of safety. Those actions include i) conserve healthy parenchyma (e.g. neoadjuvant treatment that may down- stage the tumor or ablation of selected lesions when indicated in the setting of multiple tumors); ii) induce enhanced hypertrophy of FLR with selective contralateral portal vein flow occlusion [e.g. portal vein embolization (PVE)]; and iii) prevent congestion of FLR by modulating excessive portal perfusion in the setting of post- hepatectomy portal hypertension (e.g. splenic artery ligation, por- tosystemic shunt).

Neoadjuvant approaches that reduce tumor burden and provide disease control and downstaging

Neoadjuvant therapy aims to reduce tumor size and extent, fa- cilitating subsequent definitive treatment, usually with surgery, re- sulting in better outcome with decreased morbidity and mortality. It also helps to identify patients with undetected micrometastatic disease with aggressive biology who are at high risk of early re- currence after surgery [46] . For liver tumors, reducing the size and the number of lesions allows for parenchymal-preserving surgery by increasing the amount of uninvolved parenchyma contributing to the FLR, especially if there is a bilobar distribution. Tumors that are critically located may become amenable to resection, sparing vascular structures, avoiding associated multi-segment liver resec- tions and maintaining inflow and outflow vascular integrity for the FLR without reconstruction. Eventually surgery may become less complex with a reduction in intraoperative blood loss, inflow occlusion and postoperative stress allowing the liver to regener- ate and recover its function. Treatment modalities with the inten- tion of controlling or potentially downstaging the disease, include systemic and transarterial chemotherapy, transarterial chemoem- bolization (TACE), radioembolization [systemic internal radiation therapy (SIRT) variations] and external beam radiation therapy.

Systemic chemotherapy

Currently the most common indication for chemotherapy is CRLM in the West and HCC in the East. For CRLM, systemic ther- apy is primarily directed by mutation status and biomarker ex- pression [e.g.RAS,BRAF,HER2, tropomyosin receptor kinase (TPK) fusion], the presence of DNA mismatch deficiency repair/high microsatellite instability (dMMR/MSI-H), and the primary tumor location. Standard combination regimens are either oxaliplatin or irinotecan based [folinic acid-fluorouracil-oxaliplatin (FOL- FOX), folinic acid-fluorouracil-irinotecan (FOLFIRI), capecitabine- oxaliplatin (XELOX), capecitabine-irinitecan (XELIRI), fluorouracil- leucovorin-oxaliplatin-irinotecan (FOLFOXIRI), etc] with acceptable mortality and morbidity, given the related hepatotoxicity [47] . With the use of molecular-targeting agents such as cetuximab or panitumumab in left/sigmoid colon primary,RASwild-type tu- mors may increase resectability compared to FOLFOX or FOLFIRI alone. FOLFOXIRI and bevacizumab or adding bevacizumab to backbone chemotherapy doublets (preferred in irinotecan-based regimens) are also potential options forRAS/BRAFmutated (ex- ons 2-4 ofKRAS, exons 2-4 of theNRASand exon 15 of theBRAFgene) of any location orRASwild-type right colon tumors, but the associated hepatotoxicity limits their use [ 46 , 48–51 ]. Se- lected studies with patients who successfully converted follow- ing neoadjuvant chemotherapy are shown in Table 4 [52–78] . A small number of patients (3.5%–6.5%) that have proved defi- cient in dMMR/MSI-H are uniquely susceptible to immune check- point inhibitor immunotherapy therefore first-line pembrolizumab monotherapy [targeting programmed death receptor 1 (PD-1)], rather than cytotoxic chemotherapy, should be recommended [79] . Another option is the combination of nivolumab and ipilimumab, [monoclonal antibodies against cytotoxic T-lymphocyte antigen 4 (CTLA-4)] especially for refractory or relapsed dMMR/MSI-H cases. In addition, even in tumors with proficient dMMR/MSI-H but with high levels of tumor mutational burden (TMB), pembrolizumab is an option for second-line therapy in combination with other cyto- toxic agents [79] .

Table 4 Selected studies of patients with unresectable colorectal liver metastases undergoing neoadjuvant systemic chemotherapy with conversion to surgical resection.

Initial inadequate treatment response should not be conceived as conversion failure because second-line treatment may be of- fered in selected patients with good performance status. For ex- ample: (i) switching FOLFOX/FOLFIRI to each other and keeping be- vacizumab if well tolerated or replacing bevacizumab with afliber- cept or ramucirumab; (ii) starting encorafenib an inhibitor of ty- rosine kinases in combination with cetuximab inRASwild-BRAFV600Emutant tumors; (iii) adding trastuzumab and lapatinib in HER-2 positive tumors; (iv) adding a tropomyosin receptor kinase (TRK) inhibitor larotrectinib or entrectinib , when positive [80–83] .

The combination of hepatic artery infusion (HAI) [using fluorouracil (5-FU), floxuridine or oxaliplatin] with systemic chemotherapy is an alternative option for previously refractory tu- mors or for patients who present initially with very high disease volume, with even better response rates in chemotherapy naive pa- tients [84] . Such patients and those who have failed first-line sys- temic treatment can be potentially considered for early augmenta- tion with combined HAI before proceeding with first- or second- line systemic chemotherapy alone [84] . HAI has a high rate of procedure-related complications (22%) including arterial thrombo- sis, extrahepatic perfusion, incomplete hepatic perfusion, and hem- orrhage. Liver toxicity (elevated liver enzymes, cholestasis, biliary sepsis) also can be severe and defer patients from proceeding to surgery ( Table 5 ) [85–100] . In particular the use of floxuridine is associated with biliary sclerosis and concurrent dexamethasone in- fusion is recommended [84] .

In general, patients undergoing neoadjuvant conversion chemotherapy for CRLM should be assessed for response ev- ery two months (10 weeks for pembrolizumab). If unresectable after 6 months then further treatment is unlikely to offer any benefit with regard to resection. For those who become resectable, surgery should be planned at least 4 weeks post-treatment or 6-8 weeks if bevacizumab is included, to lessen the severity of chemotoxicity [ 47 , 50 , 51 ].

Neoadjuvant systemic therapy may be an option for selected patients with non-metastatic, large, locally advanced, intrahep- atic or hilar cholangiocarcinoma, if they are not jaundiced and have a good functional reserve at presentation [101–104] . First- line regimens consist of gemcitabine alone or in combination with cisplatin (preferred). Alternative options include gemcitabine in combination with S-1 (tegafur/gimeracil/oteracil) or other agents (oxaliplatin, nab-paclitaxel, bevacizumab) [ 105 , 106 ]. FOLFOX or capecitabine based regimens are alternative options for gemc- itabine refractory cases [105] . Second or later targeted treatment for highly selected patients who may harbor an actionable ge- netic alteration may be available as well. Specific examples include infigratinib and pemigatinib for fibroblast growth factor receptor (FGFR) translocations, ivosidenib for isocitrate dehydrogenase (IDH) mutations, larotrectinib forTRKfusion-positive cases, dabrafenib forBRAFV600Emutations and immunotherapy with a checkpoint inhibitor for deficient DNA mismatch repair, overexpression of pro- grammed cell death ligand 1 (PD-L1), or high levels of tumor mu- tational burden [ 106 , 107 ]. Infigratinib has shown strong anti-tumor activity in selectiveFGFR-2fusions and may be an additional op- tion in the front-line setting of chemonaive patients [108] . Sys- temic treatment in locally advanced cholangiocarcinoma with or without external beam radiation has shown potential in reversing unresectable status and facilitating R0 resection [ 101–104 , 109 , 110 ]. Selected studies are shown in Table 6 [111–120] . Another option for selected patients is the combination of HAI with gemcitabine plus oxaliplatin alone or HAI with floxuridine with concomitant sys- temic chemotherapy with either mitomycin C/gemcitabine or gem- citabine/oxaliplatin ( Table 6 ) [ 101–104 , 109–116 ].

Table 5 Selected studies of patients with unresectable colorectal liver metastases undergoing neoadjuvant hepatic artery infusion with or without systemic chemotherapy with conversion to surgical resection.

Table 6 Selected studies of patients with locally advanced non-metastatic hilar/peripheral cholangiocarcinoma undergoing neoadjuvant treatment to convert for surgical resec- tion.

Patients with gastroenteropancreatic neuroendocrine tumors (NETs) and non-resectable liver metastases present a discrete group with highly variable clinical course. The decision for surgi- cal intervention is influenced by many factors including size, loca- tion, grade, differentiation status and potential for resection of the primary tumor, patient’s symptoms from disease bulk or hormone hyper secretion and liver disease volume [121] . Liver surgery may be offered in selected patients in association with other treatment modalities, with either curative intention or for debulking and may provide prolonged disease control and amelioration of symptoms. Initial systemic treatment can identify good responders who will potentially benefit from more aggressive approach [122] .

For patients with low to intermediate-grade metastatic NETs, first-line options include somatostatin analogs, molecularly tar- geted agents such as everolimus and sunitinib (especially for pan- creatic NETs) and temozolomide-based or streptozocin-based com- binations. Other cytotoxic agents such as 5-FU, capecitabine and oxaliplatin can be also added [ 64 , 123 ]. For patients with metastatic high grade (G3), poorly differentiated neuroendocrine carcinoma (NEC), standard regimens consist of cisplatin or carboplatin com- bined with etoposide , or irinotecan plus cisplatin. Other options in- clude streptozocin or temozolomide based schemes combined with doxorubicin, 5-FU, bevacizumab, etc [124] . A subgroup of G3 tu- mors particularly of pancreatic origin which have a Ki-67 index 20%–55% but relatively well-differentiated histology (NET G3) are usually offered first-line temozolomide rather than platinum-based chemotherapy [125] . Appropriate selection of patients who show response or even stable disease during systemic treatment, with or without other liver-directed therapies may potentially become candidates for surgery for either complete resection or significant debulking [ 65 , 126 , 127 ].

Systemic and/or hormonal treatment for patients with non- colorectal, non-neuroendocrine liver metastases is a prerequi- site when considering resection even with initially resectable disease [128] . The decision with regards to surgery is primarily directed by tumor type, chemotherapy response, evidence and lo- cation of extrahepatic disease and patients’ age and fitness [128] . Patients who have good response or stable disease are potential candidates. Also, patients with low volume silent extrahepatic dis- ease would potentially benefit from resection and continuation of systemic treatment (e.g. breast primary with liver metastases and low volume bone disease). Higher 5-year survival rates have been reported after liver resection for genitourinary 38%-62% and breast liver metastases 27%-53% respectively [129] .

TACE

TACE, drug eluting beads (DEB-TACE) chemoembolization or bland particle embolization have been extensively used as a first- line treatment before microwave ablation for patients with HCC which is deemed to be unresectable due to size, location or pre- existing portal hypertension. Depending on tumor burden and stage, TACE is also an option for selected patients with HCC awaiting liver transplantation in order to prevent progression and dropout from the waiting list, to decrease post-liver transplanta- tion recurrence (especially for patients with complete post-TACE pathological response), or to achieve allocation eligibility after suc- cessful downstaging and reassessment within the Duvoux crite- ria (patients initially beyond Milan transplant criteria) [ 130 , 131 ]. Depending on patient selection, doxorubicin alone or in combina- tion with mitomycin C and cisplatin and irinotecan eluting beads are the most effective regimens used [132] . TACE is associated with procedure-related major complications and liver toxicity. In one study of 2863 transarterial embolization (TAE)/TACE proce- dures performed in 1120 patients, the incidence of major complica- tions (rupture, abscess, pseudoaneurysm, gallbladder perforation) was 2.1% per patient and 0.84% per procedure [133] . Liver toxic- ity is presented with evidence of decompensation, ascites, eleva- tion of serum bilirubin and liver enzymes and less often with sig- nificant liver dysfunction and death (severe jaundice, coagulopa- thy, encephalopathy, renal impairment, etc). Patients with chronic liver disease, previous TACE and those who have received systemic treatment are at higher risk [134]. Miksad et al. [135] reported on 572 patients with HCC who had deterioration of liver function tests and INR from baseline in the immediate post-TACE period [biliru- bin 30%, aspartate aminotransferase (AST) 44%, alanine aminotrans- ferase (ALT) 25%, INR 25%]. In a study by Garwood et al. [136] of 443 TACE procedures for HCC, the incidence of reversible hepato- toxicity occurred after 90 procedures (20%) and permanent hepa- totoxicity after 41 procedures (9%) respectively, with 4% mortality from decompensated liver disease.

Evidence of main or peripheral portal vein thrombosis (PVT) is a relative contraindication to TACE but a recent meta-analysis by Silva et al. [137] has shown that when selective arterial catheteri- zation is performed, the risk of adverse effects is similar to that in TACE without PVT. TACE can be repeated for completion of treat- ment response, and be used in adjunction with other local or sys- temic therapies for better outcome (e.g. TACE with radiation, TACE with sorafenib, TACE with ablation) [138] . The use of sequential TACE and local ablation (radiofrequency, microwave, cryoablation) provides a synergistic action improving the effectiveness of each procedure. The advantages are as follows: (i) reduced thermal loss in embolized areas results in increased ablation zone; (ii) satellite lesions identified by TACE and supplied by collaterals can be in- cluded in the ablation zone; (iii) completion of necrosis in the pe- riphery of large tumors by ablation, chemotherapy effect enhanced by local ablation heat. The interval time varies from 3 days to 7 weeks but usually TACE precedes followed by ablation after 4-6 weeks [139] .

Preoperative TACE is also a potential option for selected patients in whom a resection is planned and studies have shown onco- logical benefit for those with (i) post-TACE complete pathological response; (ii) well compensated underlying liver disease; (iii) sin- gle tumor; and iv) alpha fetoprotein (AFP)＜100 ng/mL before TACE [ 140 , 141 ]. Surgery can be performed 4-6 weeks post-TACE. Patients with large HCCs which are partially treated with TACE prior to resection seem to benefit less as they have worse disease- free and overall survival outcomes comparing to those undergo resection without preoperative TACE. However, a subgroup of pa- tients with large or multifocal tumors and insufficient FLR, who are not candidates for resection or transplantation, may benefit from devascularization and reduction of tumor size, often in combina- tion with volume manipulation techniques (e.g. TACE with PVE). In these patients, depending on the degree of portal hypertension, re- section may be offered if tumor control is established with signs of liver hypertrophy of the area to be kept. Studies have shown sat- isfactory FLR hypertrophy with increased conversion rate with se- quential TACE and PVE (usually after 2 weeks) ( Table 7 ) [142–153] .

Table 7 Selected studies of patients with HCC undergoing sequential TACE and PVE with conversion to surgical resection.

TACE has also been used for CRLM as an alternative neoadjuvant treatment for resectable disease with a view to improve post-hepatectomy oncological outcome (DEBIRI trial with 76% complete or major pathological tumor response rate post-resection) [154] . More commonly it has been used for patients with unresectable tumor(s) in a combined setting of multiple lines of systemic chemotherapy and/or HAI therapy and potential re- peated TACE sessions for refractory cases [ 155 , 156 ]. Despite het- erogeneity of cases, studies have shown post-TACE tumor response rates ranging between 47% and 56%, depending on overall disease burden and location [157–159] .

For patients with non-resectable neuroendocrine liver metas- tases, studies have shown that single or multiple sessions of TACE can achieve partial or complete radiological response rates of 35%–75% [160] and partial response or no progression in 70% of patients [161] . For selected patients, anatomical hepatectomies or multiple wedge resections, or a combination of redo liver resec- tions and repeated TACE can provide R0 resection or a significant debulking of high-volume metastases with symptomatic improve- ment, resulting in sufficient long-term disease control [162] .

Finally, the use of TACE or TAE in benign liver lesions such as cavernous hemangioma (CH), focal nodular hyperplasia (FNH) and hepatic adenoma (HA) has also been reported. In general, surgery is the definitive treatment for symptomatic lesions, those with size＞5 cm, bleeding tumors and adenomas with malignant poten- tial [ 163 , 164 ]. Surgical resection is not always feasible due to tu- mor size and proximity to major vascular structures. Selective TAE can be used as an initial approach in an actively bleeding tumor or as a “downstaging” method when future resection is planned, in order to avoid extended parenchymal loss and reduce intrao- peartive blood loss and length of inflow occlusion when the tu- mor is devascularized [165] . Size reduction varies and sometimes can reach 90% in smaller lesions. Embolization can be performed using mostly gelatine sponge, steel coils, polyvinyl alcohol (PVA) or ethanol [ 166 , 167 ]. Additional delayed tumor size reduction has been observed when pingyangmycin-lipiodol or bleomycin is used due to their long-term anti-angiogenic properties which cause an ongoing fibrotic reaction [168] .

Radioembolization

Radioembolization using intra-arterial injection of Yttrium-90 (90-Y)-labeled microspheres (resin or glass spheres) is an estab- lished method which delivers high focal doses of radiation to liver tumors, inducing extensive necrosis within an acceptable safety profile [ 169 , 170 ]. Its indications include multiple bilobar lesions or solitary large tumors. It has been used as an alternative treat- ment to medical therapy, as a palliative technique for symptomatic patients with hepatic-predominant unresectable disease or as a debulking method of high-volume liver metastasis. The term ra- diation segmentectomy is used when a specific area is targeted with higher dose radiation causing “ablation” of the underlying tumor [171] . Studies have shown favorable response rates (rang- ing from 26% to 91%, depending on the tumor type including HCC 50%, colorectal 18%-46%, neuroendocrine 63% and breast metas- tases 55%-57%), survival benefit with amelioration of symptoms, and downstaging to allow resection in selected patients [ 172 , 173 ]. Most common complications include mild post-embolization syn- drome, cholangitis and cholecystitis, gastric and duodenal ulcera- tion, pancreatitis and pneumonitis. The most serious complication is radiation-induced hepatitis and patients with serum bilirubin＞2-3 mg/dL should not be eligible [174] .

For patients with advanced HCC, radioembolization may be an option for bulky or multifocal tumors, macrovascular invasion and associated portal vein thrombosis, and for those awaiting trans- plantation with indications following the same principles as with TACE [ 130 , 169 ]. Patients with locally advanced intrahepatic cholan- giocarcinoma are also candidates for radioembolization and those with significant response may potentially become candidates for R0 resection [ 174 , 175 ]. Best results are achieved with concomi- tant systemic treatment with conversion rates ranging from 10.8% to 46% [ 117 , 176 , 177 ]. Similarly, patients with CRLM who receive first-line systemic treatment combined with at least one session of radioembolization, may show increased conversion rate versus chemotherapy alone (38.1% vs. 28.9%) [178] .

Interestingly compensatory hypertrophy of untreated liver parenchyma has been reported in patients undergoing transarte- rial radioembolization, although this seems to be to a lesser ex- tent (23% vs. 69%) and over a longer period of time (1–3 months vs. 4–6 weeks) than patients undergoing PVE/portal vein ligation (PVL). The hypertrophy continues for up to 4–12 months pos- sibly due to ongoing radiation induced fibrosis, causing selec- tive portal hypertension and diverting portal flow to the unaf- fected areas [173] . In one study hypertrophy of FLR was 45% at 9 months [179] and in another study was 30.8% at 6 months and 40.1% at 12 months respectively [180] . Patients also poten- tially benefit from disease control and prevention of tumor devel- opment in the FLR as opposed to PVE/PVL [181] . When any kind of intra-arterial treatment is planned in patients receiving systemic chemotherapy, timing should be adjusted, as a number of agents may interact to increase toxicity and potential complications (e.g. capecitabine sensitizes tissues to the effect of radiotherapy, VEGF- R inhibitors increase the risk of arterial injury during endovascular access) [182] .

Radiotherapy

External beam radiation delivered either by conventional approaches or with conformal treatment, brachytherapy, stereo- tactic body radiation therapy (SBRT) or nano-knife has been investigated for primary liver tumors (HCC and cholangiocarci- noma) as a neoadjuvant approach [ 183 , 184 ]. SBRT delivers high doses of radiation in a single treatment or in a small number of fractions with high precision minimizing the radiation effect to adjacent liver parenchyma and offers locoregional control and potential for conversion of initially unresectable tumors. Indications include HCC＜5 cm with associated tumor vein thrombosis where other treatment modalities (ablation/TACE) are not suitable [ 183 , 184 ]. Although the primary role of SBRT for perihilar and intrahepatic cholangiocarcinoma is to control tumor locoregionally, some patients may show favorable response and become candidates for resection. Initial systemic chemotherapy followed by radiation and concurrent chemotherapy may maximize results ( Table 6 ) [111–120] . Toxicity includes biliary complications and duodenal ulceration and obstruction. Proton beam therapy (PBT) has also been used for the local control of HCC, cholan- giocarcinoma and liver secondaries with potential to concentrate radiation over the targeted volume, sparing adjacent normal liver tissue [185] . Patients can be treated with various dose fraction- ations and best results are seen when the biologically effective dose (BED) exceeds 80 GyE [ 186 , 187 ]. Dose has to be de-escalated when tumors are less than 2 cm away from porta hepatis, stom- ach or duodenum. PBT may be used to control and downstage tumors often in combination with systemic treatment minimizing associated liver injury and potentially create a window for surgical exploration in selected patients [185–187] .

Liver volume manipulation techniques

PVE

Preoperative PVE is the occlusion of portal flow to a se- lected portion of the liver containing the tumor to be re- sected. It has been widely adopted as a method of increasing the FLR and induces liver regeneration by clonal expansion of hepatocytes [188–193] . FLR may increase up to 62%, 3–8 weeks post-embolization (as a percentage augmentation from baseline FLR) [194] converting up to 72%–85% of previously unresectable pa- tients. Of those who fail to become surgical candidates, tumor pro- gression, inadequate FLR and procedure-related complications (in- advertent embolization, recanalization, thrombosis) are the com- monest reasons [ 37 , 195–197 ]. Tumor progression in this scenario may be seen as a way evaluating oncological aggressiveness and aid patient selection [ 37 , 196 ]. Longer intervals post-PVE may be re- quired for patients with underlying liver disease, previous systemic treatment, diabetes and poor nutritional status. Post-embolization functional and volumetric assessment combination of FLR is help- ful in subsequent liver resection (usually 4–8 weeks) ( Fig. 1 ) [197] .

Fig. 1. Increase of FLR volume (segments 2-3) before ( A ) and after ( B ) right PVE including segment 4 branch. PVE: portal vein embolization; FLR: future liver remnant.

Fig. 2. Increase of FLR volume (segments 2–4) before ( A ) and after ( B ) simultaneous PVE and HVE. FLR: future liver remnant; PVE: portal vein embolization; HVE: hepatic vein embolization.

PVE is used primarily for right-sided tumors. The radiological approach varies amongst centers and depends on tumor location, the extension of liver resection and embolization (segment 4), type of embolic agent used, as newer agents becoming more efficient, and operator preference [195] . Transhepatic/transjugular ipsilateral access benefits from leaving intact FLR but is technically challeng- ing due to sometimes difficult angulation and the risk of tumor seeding and bleeding. In the contralateral approach the main limi- tation is the risk of compromising FLR and its portal supply [195] . The choice of embolic agent depends on the local expertise, avail- ability, and cost. Cyanoacrylate causes more effective and long- lasting occlusion with minimal associated adjacent tissue reaction, compared with thrombin or gelfoam. PVA in combination with coils is also an excellent alternative. In general, PVE is considered more effective (hypertrophy 69% vs. 46% with large particles) if ap- plied selectively occluding peripheral branches with small spheri- cal particles [ 194 , 195 , 198 ].

PVE can be applied alone or in adjunction with other surgical or interventional techniques with view to maximize hypertrophic effect in FLR and control accelerated tumor progression during the interval period. Continuation of neoadjuvant treatment during that time, if well tolerated, is not contraindicated and may decrease the risk of tumor progression and appears not to have a nega- tive impact on the liver regeneration [199–201] . When extended right liver resection is anticipated, embolization of segment 4 is necessary and enhances hypertrophy of segments 2-3 [199] . This approach may be useful in cases with bilobar tumors (e.g. CRLM), where the lesions in the FLR need to be resected first, followed by PVE and right trisectionectomy. Interestingly, it has been found that metastasectomy in the FLR prior to PVE appears to impair hy- pertrophy induced by PVE. However if PVE includes segment 4, this compensates for the negative impact of first-stage resection, inducing hypertrophy similar to that observed in patients under- going single-stage right hepatectomy [202] .

There have been several studies showing the benefit of se- quential TACE and PVE (followed by 1-2 weeks) in gaining sat- isfactory and fast hypertrophy of the FLR, especially in the setting of HCC and liver cirrhosis. Obliteration of arteriopor- tal shunts can maximize the effect of PVE, and tumor necro- sis may facilitate resection and allow for additional parenchymal preservation [ 143 , 148 , 152 , 193 , 203 , 204 ]. In a similar manner the as- sociating portal embolization and arterial ligation (APEAL) proce- dure is a two-step approach when tumors in the FLR are either re- sected or ablated and at the same time the tumor dominant right anterior or right posterior hepatic artery is ligated, followed by PVE [ 193 , 205 ].

Sequential (in case of inadequate FLR after initial PVE) or si- multaneous PVE and hepatic vein embolization (HVE), known as liver venous deprivation, has also been used to further increase FLR volume ( Fig. 2 ). Outflow obstruction, congestion and atrophy of the embolized segments, result in increased pressure, impaired arterial inflow and subsequently remodeling of parenchymal tissue, neo-collateral creation and interlobar sinusoidal diversion of por- tal flow to the unaffected segments [ 35 , 193 , 206 , 207 ]. HVE seems suitable for patients at risk of failure of conventional PVE such as those with a very small FLR volume and most commonly has been used for patients with extrahepatic biliary cancer [208] . One or two hepatic veins depending on the segments planned to be resected can be embolized at the same time (e.g. right and mid- dle hepatic vein for a right trisectionectomy) [209] . In one multi- institutional review study of 68 patients, the hypertrophy of FLR after HVE ranged from 33% to 63.3% allowing for surgery in 85.3% of previously deemed unresectable patients whom the majority (82.4%) had hepatobiliary cancers [208] . HPVE in cirrhotic patients is not contraindicated but because of subsequent liver congestion and potential worsening portal hypertension, it should be used in selected patients [208] . Patients with obstructive jaundice, in the view of planned HVE, would benefit from biliary drainage of FLR and this can be done simultaneously during HVE [210] .

Fig. 3. Increase of FLR volume (segments 2–4) before ( A ) and after ( B ) first stage of ALPPS procedure. FLR: future liver remnant; ALPPS: associated liver partition and portal vein ligation for staged hepatectomy.

Fig. 4. Intraoperative picture at completion of first stage ALPPS. ALPPS: associated liver partition and portal vein ligation for staged hepatectomy.

PVE with contralateral application of autologous hematopoietic stem cells which stimulate liver regeneration and induce enhanced hypertrophy of FLR has also been reported [211] . Stem cells can be obtained from peripheral blood after stimulation using gran- ulocyte colony-stimulating factor and leukapheresis or with bone marrow aspiration. The stem cells are delivered to the FLR via its portal branch 1-2 days after PVE [212] . Enhanced FLR hypertrophy in less time than PVE alone was reported in one study by Fürst et al. where gain volumes was 79% vs. 39% with PVE alone and the intervals to liver resection were 27 days vs. 45 days [213] . Similar results in terms of hypertrophy of FLR within a certain period of time were reported in two different studies by Ludvík et al. and by am Esch et al. in which gain volumes were 173.2 mL vs. 98.9 mL and 138.66 mL vs. 62.95 mL in the PVE alone group after 3 and 2 weeks respectively [ 214 , 215 ]. In another study of 55 patients with CRLM by Treska et al., patients with PVE and stem cell stimulation were more likely to undergo surgery than the patients with PVE alone although rapidity in volume gain did not differ between the two groups [212] .

PVL and two-stage hepatectomy

Surgical open or laparoscopic PVL is an alternative to PVE based on the same principle of inducing hypertrophy of FLR by increasing the portal flow to it. It is more invasive and requires general anesthesia and many studies have shown similar results in terms of morbidity, mortality, conversion rates and disease progression [ 216 , 217 ]. Hypertrophy of FLR achieved by PVL is suf- ficient to facilitate complete resection but not as effective as PVE (27% vs. 39% and 43.1% vs. 53.4% in two studies) [ 217 , 218 ]. This is probably due to the more proximal portal vein occlusion in PVL al- lowing the development of collaterals along the two sides of the future cut surface [219] . In addition, there is usually no interrup- tion to the venous inflow to segment 4 although this can be done intraoperatively.

The most common indication for PVL and two-stage hepatec- tomy is bilobar CRLM. During the first stage, resection of metas- tases in the FLR and ligation of right portal vein is performed, fol- lowed by completion right hepatectomy after an interval period. Intraoperative staging can be done during the first operation with the option of resecting the primary tumor in case of synchronous metastasis. Other indications include misjudgment of a marginal FLR discovered during planned liver resection and lack of access to interventional radiology facilities [216–220] .

Associated liver partition and portal vein ligation for staged hepatectomy (ALPPS)

ALPPS has emerged as a technique which potentially can cre- ate an adequate FLR in a short period of time [221] . The princi- ple is based on collateral communication disruption between the two liver parts after the initial parenchymal transection and the exclusion of portal inflow to the tumor-involved segments [ 3 , 222 ]. This approach maximizes the compensatory increase in portal flow to the FLR and significantly accelerates (10-14 days vs. 3-8 weeks post-PVE/PVL) hypertrophy. FLR increase varies among studies with different patients, interval time between stages and tumor types and ranges from 48% to 110% [ 223 , 224 ]. During the interval phase the “diseased occluded” liver, which still has intact arterial sup- ply and biliary drainage, maintains some metabolic and synthetic function ( Figs. 3 and 4 ) [224] .

Liver volume does not always correlate with the observed FLR and SFSS may occur despite apparent generous radiological hyper- trophy [ 3, 225, 226]. Various studies have looked at the histopathol- ogy of the hypertrophied liver and have shown the superiority of ALPPS (compared with PVE and PVL) in driving hepatocyte prolifer- ation as reflected by increased cell kinetic growth rate. This results in increased liver volume by hepatocyte number with less cellu- lar edema or lipid accumulation [ 225 , 227 , 228 ]. However, optimal function of those hepatocytes is questionable, as they appear at a least healthy status. They are smaller in size and show signs of immaturity due to higher glycogen concentration, less lipofuscin granules, different or ganelle distribution pattern and narrowing of sinusoidal space [229] .

Despite the increasing number of ALPPS performed worldwide, there is no clear consensus as to whether it is superior to PVE and PVL. ALPPS seems to be an attractive strategy in the setting of tu- mor vascular invasion or thrombosis, fast growing tumors and bilo- bar disease distribution (clearing the FLR in the first stage). In ad- dition there are occasions when PVE or other techniques are either contraindicated or fail to create sufficient FLR [226] .

ALPPS clearly has higher morbidity (14%-50%) and mortality (0-29%) from complications mainly related to SFSS, compared to that reported after conventional trisectionectomy with or without PVE [230] . Predisposing factors include type of primary tumor (cholangiocarcinoma and HCC vs. colorectal) underlying liver dis- ease, age (＞60 years) and surgical complexity with concomitant vascular or biliary reconstruction [ 231 , 232 ]. Liver regeneration in patients with HCC and liver cirrhosis is less predictable and ALPPS seems to have an inferior hypertrophy effect with impaired func- tionality of the FLR, reflecting higher mortality rates in those pa- tients (11%-31%) [233] . D’Haese et al. reported 35 HCC patients with FLR incease of 47% after first stage of the ALPPS against 76% in patients with CRLM. Hypertrophy in HCC patients showed a linear negative correlation with the degree of liver fibrosis [234] .

The ALPPS procedure has been modified with the intention of the followings. (i) Reduce complications related to parenchy- mal transection such as bleeding, infected collection, bile leak, etc. Partial split is up to 60%–80%, by use of a tourniquet along future transection line to “squeeze” intrahepatic portal branches instead of a physical split, creating an ablation zone [ radiofrequency ablation /microwave ablation (RFA/WMA)] instead of a splitting line. (ii) Avoid extensive necrosis and adhesions (re- section of segment 4 during first stage, anterior approach with minimal or no hilar dissection and mobilization followed by PVE, use of a plastic bag, fibrin glue or protective film, laparoscopic ap- proach at either stage). (iii) Enhance FLR hypertrophy (longer in- terval period, right bile duct ligation). (iv) Increase resectability of different locations of liver tumors often “mixing” right and left liver segments (left ALPPS, monosegmental ALPPS, mesohepatec- tomy with doubleinsitusplitting) [ 225 , 231 , 232 ].

Despite the increased morbidity and mortality related to ALPPS procedure, detailed assessment with careful selection of patients and employment of technical modifications as appropriate has led to improved outcomes [232] . In a study from the ALPPS reg- istry, the mortality in 320 patients was 8.8%, mostly (75%) due to SFSS [235] . Risk factors for death from SFSS included advanced age (＞60 years), increased intraoperative blood loss, longer duration of first stage, previous cholestasis and perihilar and gallbladder tu- mors vs. intrahepatic cholangiocarcinoma, HCC and metastatic tu- mors. Of note patients with MELD score＞10 after first stage were at increased risk of developing SFSS. However, FLR before stage two and interval time were not predictive, indicating that liver qual- ity and function are not proportional to the rapid hypertrophy in- duced by ALPPS [235] . Baili et al. reported 83 patients from various centers who underwent modified ALPPS with a mortality rate of 13.3% [236] . In a meta-analysis by Gavriilidis et al. comparing 315 patients with ALPPS and 702 patients with any other form of por- tal vein occlusion, mortality and major morbidity rates were 8% vs. 4% and 47% vs. 22%, respectively [237] . However, FLR hypertrophy volumes and the number of patients who proceeded to definitive liver resection were higher in ALPPS group and the interval time to operation was shorter [237] .

Ablative techniques

Open, laparoscopic or percutaneous ablation of liver tumors of- fers options for patients with HCC and colorectal/neuroendocrine metastases. RFA or MWA causes coagulative necrosis of both the tumor and the surrounding healthy liver parenchyma [238] . MWA is rapid and provides a consistent burn with less heat-sink effect than RFA and may be as effective for lesions＜5 mm in prox- imity with major vascular structures [239]. In general, ablation is associated with better outcomes in patients with three or fewer lesions and those＜3 cm in diameter [230]. Ablation margin is also an important factor for success. Shady et al. reported that lo- cal tumor progression (LTP) rates for CRLM ablated with a margin of ≤5 mm,＞5–10 mm, and＞10 mm were 60%, 10.5% and 0% respectively [240].

Patient fitness, multifocal distribution, underlying liver disease and inadequate hepatic reserve are common indications for abla- tive therapy [238] . Patients benefit from shorter hospital stay and reduced complication rate compared to those of resection (13% vs. 26%). Despite higher local recurrence rates, adequate long-term lo- coregional control can be achieved [241] . In two studies of 205 and 353 patients with CRLM treated with resection or RFA, intrahepatic recurrence rates were higher in the ablation group in both studies (47.5% vs. 12.1% and 59.9% vs. 23.9%, respectively) [ 242 , 243 ]. A re- cent study has shown superiority of MWA against RFA in terms of time required to complete ablation, LTP rates especially for lesions near vascular structures, and overall survival. In one study of 110 patients with CRLM the mean total ablation time was shorter for MWA (18 min vs. 23.5 min) compared with that for RFA [240] . In a meta-analysis by Di Martino et al. of 860 patients treated with three forms of ablation (MWA, RFA, cryoablation), LTP was lower in the MWA subgroup (4.5% vs. 14% for RFA and 21% for cryoab- lation). Similarly, the median 5-year disease-free survival rate was 18% with RFA vs. 38.5% with MWA [244] .

Another option for ablation is irreversible electroporation (IRE) with delivery of short-pulsed high-voltage electric current through fine electrodes applied to the tumor, causing irreversible injury to cell membrane and cell death by apoptosis [245] . Its main indi- cations include tumors with close proximity to major vascular or vital structures such as the bile ducts or bowel, and allowing in- growth of hepatocytes to the treated area [246] . Recurrence-free rates vary with tumor size and type, and range between 55% and 93% during the follow-up of 3–12 months [ 247 , 248 ]. Repeated IRE combined with MWA is an option for more bulky tumors [246] .

Liver ablation can be applied in adjunction with one- or two- stage hepatectomy (with or without PVE/PVL or in ALPPS), in pa- tients with multiple tumors, where selected lesions are resected and others are ablated based on their location and proximity to major pedicles. In this setting, ablation can be as effective as re- section when an ablation margin＞5 mm is achieved [244] . Aug- mentation of a compromised resection margin is also possible. This modality potentially allows for more parenchymal preservation of FLR with acceptable long-term locoregional control. Colorectal and neuroendocrine secondaries are common indications for this ap- proach ( Table 8 ) [ 243 , 249–259 ].

Intraoperative post-resection manipulation: portal vein flow modulation techniques

Post-hepatectomy elevated portal vein pressure increases the risk of SFSS regardless of FLR volume especially in patients with chemotoxicity or chronic liver disease and some degree of pre- existing portal hypertension [260] . The risk is low with portal vein pressure＜15 mmHg, intermediate with portal vein pressure 15–18 mmHg and significant when＞18–20 mmHg [261] . Techniques modifying portal inflow which reduce portal vein pressure and prevent parenchymal congestion have been used in liver trans- plantation resulting also in reciprocal increase in hepatic arterial inflow [261] . Reported surgical modifications have included splenic artery ligation, splenectomy, meso-caval shunt, spleno-renal shunt, portocaval shunt, and splenic artery embolization [262–265] .

Tanaka and Ogura reported reduced portal vein pressure from 16 mmHg to 11 mmHg in 7 adult living donor liver transplan- tation (LDLT) recipients after splenic artery ligation [261] . Troisi et al. reported 8 adult LDLT recipients who underwent hemiporto- caval shunt with significant decrease in portal vein flow compared to the group without flow modulation, 190 mL/min/100 g vs. 401 mL/min/100 g liver with associated higher graft survival at 1 year (75% vs. 20%) [266] . In 10 adult LDLT recipients, portal vein flow decreased from 360.2 mL/min/100g to 240.5 mL/min/100g graft weight following splenic artery ligation [267] . Botha et al. found significant decrease in portal vein pressure/central vein pressure gradient from 18 mmHg to 5 mmHg in 16 adult LDLT recipients with hemiportovaval shunt and reduced portal vein flow from 1490 mL/min to 1018 mL/min with graft survival of 81% at 1 year [263] . Soin et al. reported that the mean portal vein pressure decreased from 26 mmHg to 15 mmHg in 123 adult LDLT recipients after inflow modification with either hemiportocaval shunt or splenic artery ligation [268] . A systematic review by Troisi et al. [267] in- cluding 12 studies and 449 LDLT recipients, reported significant reduction in portal flow in those undergoing single or combined inflow modulation as demonstrated by portal vein flow or pres- sure measurements ( Table 9 ) [ 263 , 269–279 ]. In general, the splenic artery occlusion with or without splenectomy was sufficient, but if further reduction in portal vein pressure was required, the use of a portocaval shunt may help to avoid SFSS. The complication rate was between 16.6% and 70% and included high volume as- cites, SFSS, portal vein thrombosis, early graft dysfunction and por- tal steal phenomenon. The retransplantation rate was between 0% and 25% mainly due to SFSS with 1-year survival between 69% and 100% [267] .

Table 8 Selected studies of combined liver resection and ablation for metastatic colorectal cancer.

Table 9 Secondary outcomes of portal flow modulation for living donor recipients.

For hepatic resections, portal vein modulation has not been per- formed routinely but is helpful when the liver remnant appears congested. Obtaining intraoperative measurement of pre- and post- resection portal vein pressure and flow, and portal vein-hepatic veins/central venous pressure gradient (usually using direct fine gauge catheters introduced to the portal vein), is a key step. Por- tal vein pressure＞15–20 mmHg or PV/CVP gradient＞15 mmHg or portal flow＞250–360 mL/min/100g liver, are usually sugges- tive of a hyperperfused liver remnant, indicating that patient will benefit from inflow modulation [ 267 , 280 ]. The type of interven- tion is influenced by local expertise and surgeon preference and depends on initial pressure values and response to each treatment as there is no consensus favoring a specific technique. Occasionally portal vein insufficiency is present and related to variceal collat- erals diverting flow from the liver and will need to be surgically ligated [ 281 , 282 ]. Sato et al. observed a reduction in portal vein pressure after splenic artery ligation from 29 mmHg to 24 mmHg in 6 patients with cirrhotic liver who underwent major liver re- section for HCC. None of these patients developed SFSS postopera- tively [283] . Schwarz et al. reported that no SFSS in 4 patients who underwent liver resection with splenic artery ligation for CRLM who had developed nodular regenerative hyperplasia related to previous oxaliplatin chemotherapy [284] .

The use of pharmacological reduction of portal vein pressure with agents such as somatostatin or terlipressin has been reported and may be an additional option in reducing portal flow and re- ducing the risk of SFSS. Somatostatin causes splanchnic vasocon- striction and has been used for the treatment of variceal bleed- ing and portal hypertension in liver cirrhosis [285] . Ozden et al. reported successful treatment of SFSS in an adult LDLT recipient using somatostatin and propranolol from the second postopera- tive day [286] . Continuous somatostatin infusion intraoperatively and for the next five days with a median decrease in portal vein pressure of 3.5 mmHg has been reported in 10 patients undergo- ing major liver resection [267] . Three patients that had the lowest portal vein pressure reduction＜2.5 mmHg developed SFSS [287] . Troisi et al. [288] investigated on liver hemodynamics in 18 ca- daveric orthotopic liver transplant recipients, who were given 500μg somatostatin intraoperatively before liver resection followed by continuous infusion of 250μg/h in the anhepatic phase and there- after up to 5 days and compared with 11 placebo patients. Hep- atic venous presssure gradient (HVPG) and portal vein pressure de- creased most in somatostatin group (81.66% vs. 58.82% and 47.12% vs. 31.25%, respectively), and hepatic artery flow decreased less in somatostatin group (10.39% vs. 45%). Currently the SOMAPROTECT study is in progress, which evaluates the presence of postoperative ascites after somatostatin infusion following open liver resection for HCC [285] .

Terlipressin has been used to increase systemic vascular re- sistance, mean arterial pressure and ameliorate renal impairment during liver transplantation. It has also been shown to reduce por- tal vein pressure, lessen the severity of postoperative ascites, re- sulting in a shorter hospital stay [ 281 , 289 , 290 ]. Fayed et al. showed that portal vein flow was reduced in adult LDLT recipients when terlipressin was given continuously for 3 days starting at time of surgery with 1597 mL/min compared to 1898 mL/min in a con- trol group [291] . Li et al. [272] reported 65 patients who under- went major liver resection and had elevated intraoperative por- tal vein pressure＞12 mmHg who were given 1 mg terlipressin. Those who responded (46/65) with any degree of portal vein pres- sure reduction continued terlipressin infusion for another 4 days. The mean portal vein pressure decreased from 15.8 mmHg to 14.3 mmHg in the responders’ group who had less ascites postoper- atively than non-responders. The incidence of post-hepatectomy liver failure was also lower (26% vs. 53%) in the responders group [292] . Mahdy et al. [293] also observed a decrease in por- tal vein pressure from 17.88 mmHg to 15.96 mmHg after terli- pressin infusion intraoperatively and for the next 48 h in 25 pa- tients undergoing liver resections and pancreatoduodenectomies. In contrast Kohler et al. [294] found that perioperative admin- istration of terlipressin in 75 patients undergoing liver resection was not associated with a reduction in SFSS syndrome or other liver-related complications except for better preservation of kidney function as compared to controls. Somatostatin or terlipressin use is not invasive, is reversible and can be given during the acute re- generation phase in the first 5–7 days postoperatively [285] . Both agents reduce portal vein pressure, however, more studies focusing on clinical outcome and complications such as bowel ischemia are needed.

Preconditioning liver ischemia

Remote or regional liver preconditioning ischemia may decrease ischemic reperfusion injury, through previous exposure to a short period(s) of vascular occlusion [ 295 , 296 ]. Creating a brief period of ischemia, followed by reperfusion, to the liver or to a remote organ (e.g. lower limb with a tourniquet on the thigh), has shown that humoral factors (e.g. nitric oxide, adenosine, protein kinase C, etc) are released and able to modulate oxidative stress, enhance ATP recovery, and regulate endoplasmic reticulum injury, resulting in less hepatocyte apoptosis and cell death [ 297 , 298 ].

Studies have shown reduction in postoperative serum transam- inases and increased ICG clearance in patients undergoing liver resection following remote ischemic preconditioning of the lower limb and lower mortality and graft loss for liver transplant recip- ients who had received a graft from donors undergoing ischemic preconditioning [ 296 , 299 , 300 ]. However, the majority of clinical trials have not shown substantial benefit in terms of reducing SFSS and improving overall clinical outcome [ 299 , 301–303 ]. Ischemic preconditioning could potentially be beneficial for patients with chronic liver disease who are more susceptible to injury during warm ischemia or in cases where significant ischemia is antici- pated (prolonged parenchymal resection time, total vascular ex- clusion, etc) [ 301 , 302 ]. Hahn et al. [304] randomized 160 major liver resection patients to receive either intermittent portal clamp- ing (15 min ischemia, 5 min reperfusion), or ischemic precondi- tioning (10 min ischemia, 10 min reperfusion). In the cirrhotic-liver subgroup consisting of 60 patients the postoperative morbidity in terms of ascites production, signs of liver insufficiency, intensive care unit and overall hospital stay were significantly lower in the ischemic precondition group.

Discussion and recommendations

SFSS remains uncommon after major liver resection on a back- ground of parenchymal liver disease, but once established is asso- ciated with a mortality of 50%. Patients at risk of developing in- adequate hepatic reserve due to the extension of resection, disease burden, or tumors proximal to major vascular structures should be evaluated for options increasing FLR. The aim is to establish a safe FLR and achieve an R0 resection, and avoid postoperative hepatic failure. A combination of approaches may be necessary.

Initially all patients should be discussed in a multidisciplinary meeting and be evaluated for fitness which should include nutri- tional status. Laboratory markers to assess baseline liver function and identify patients with chronic liver disease are routinely used. CT volumetry will estimate the standardized FLR which can be cor- related with the total liver volume in relation with patient weight, and guide with the resection plan. Other tests such as Gadoxetic acid-enhanced MR or99mTc-mebrofenin HBSBSA may also provide useful information about the functional FLR. ICG clearance test is another option for liver function assessment especially in cirrhotic patients. Following assessment, a proposed strategy that how to approach the patients who have inadequate FLR is summarized in four basic steps. (i) Tumor downstaging (systemic chemother- apy, intra-arterial target treatment, TACE, SIRT, radiotherapy). (ii) Liver volume manipulation techniques PVE/PVL with or without HVE. (iiia) One- or two-stage liver resection + /- tumor(s) ablation, ALPPS. (iiib) Portal vein modulation methods (splenic artery liga- tion, splenectomy, portocaval shunt). (iv) Early use of postoperative terlipressin or somatostatin.

The exact pathway should be individualized and appropriate modifications or combination of different techniques should be uti- lized. Balzan et al. [305] described 5 patients with bilobar CRLM, with sparing of segment 4, with high disease burden after sys- temic chemotherapy who had two-stage liver resection. The first stage was right hepatectomy with middle hepatic vein preservation and simultaneous outflow modification of left lateral segment with intraoperative narrowing of left hepatic vein with a silicone sling. After 4 weeks enhanced hypertrophy of segments 4 and 1 against segments 2-3 (144% vs. 54%) was observed. The second stage was left lateral liver resection. Despite evidence of adequate FLR, one patient died due to SFSS [305] .

A challenging group of patients are those who have failed to gain a sufficient FLR after PVE. Subsequent ipsilateral HVE as sal- vage ALPPS may help. Reports on salvage ALPPS are limited and despite apparently adequate FLR, cases of SFSS and other compli- cations still occur. Maulat et al. [306] performed 7 rescue ALPPS with previous apparent satisfactory PVE and hypertrophy. Two pa- tients developed SFSS and the overall major (Clavien-Dindo＞3) morbidity and mortality was 42.9% and 14.3%, respectively [306] . Ulmer et al. [307] reported 9 patients with salvage ALPPS having 77% FLR hypertrophy after the first stage, with 33% major morbid- ity and 11% mortality but no SFSS. Enne et al. [308] described 20 patients with salvage ALPPS an average of 50 days following portal vein occlusion with 88% additional growth after the second stage with 5% major morbidity and no mortality. Sparrelid et al. [309] re- ported no mortality or SFSS for 6 patients with a 61% increase in FLR after first stage of ALPPS.

The option of performing staged compete inflow occlusion to the liver segment containing the tumor was mentioned previously (sequential TACE and PVE for large solitary HCC). The role of TACE is to reduce disease burden allowing for more parenchymal spar- ing. In addition, TAE has been used to enhance hypertrophy fol- lowing insufficient PVE or ALPPS. In two patients with CRLM, FLR after PVE was only increased by 25% and 12%, but after sequen- tial TAE this reached 69% and 42% respectively allowing subsequent surgery [310] . In another report of a large HCC sequential TAE 13 days after first stage of ALPPS resulted in an additional 42% of FLR, allowing for second stage completion [311] . It has been proposed that portal occlusion to a liver segment that contains large volume disease may cause arterial steal phenomenon and impeding hyper- trophy, and TAE can potentially correct this [311] .

Experience in liver transplantation for unresectable CRLM is limited to small series and has shown high risk of recurrence with 3-year disease-free survival less than 35%; 5-year overall sur- vival has reached 60%-80% in highly selected patients [312] . The combination of staged hepatectomy and auxiliary liver transplan- tation using a partial graft (usually the left lateral segment) from a deceased or living donor may be an alternative option [313] . In the first operation left hepatectomy, right portal vein ligation and implantation of a left lateral segment are performed. Fol- lowing an interval with sufficient hypertrophy of the implanted left lateral segment, completion right hepatectomy offers onco- logical clearance. There are currently two clinical trials plan- ning to recruit 60 patients, looking at the outcomes of LDLT with two-stage hepatectomy for patients with isolated, irresectable CRLM [ 314 , 315 ].

Postoperative management of patients that develop SFSS is equally important in order to improve overall outcome and shorten hospital stay and recovery time especially for those who require adjuvant treatment. Management is directed by the severity of liver insufficiency and its complications. Early recognition and treatment is a key factor. Supportive measure in the presence of ascites, coagulopathy, infection, encephalopathy and bleeding and appropriate escalation in case of deterioration and additional organ dysfunction (renal, cardiovascular and respiratory support) remains the standard approach. Maintaining a high level of nutritional sta- tus preferably via the enteric route is important and provides an opportunity for effective liver regeneration [ 14 , 21 ]. When there is evidence of portal hypertension, portal flow modulation tech- niques as described above (porto-systemic shunt, splenic artery ligation/embolization, telipressin or somatostatin infusion) may be helpful to reduce parenchymal congestion, improve cholestasis and reduce ascites [316] . The use of extracorporeal artificial or bioarti- ficial liver support to maintain some degree of synthetic function and plasma detoxification has also been reported as a potential op- tion for highly selected patients who fail to recover liver function after conservative management, but with little high level evidence to justify that. Finally salvage orthotopic liver transplantation in patients with favorable tumor characteristics may be considered in selected cases [ 7 , 8 , 15 , 17 ].

In conclusion, the indications for radical treatment for patients with liver tumors have significantly expanded due to employment of new techniques, which minimize the risk of SFSS. The successful outcome depends on appropriate patient selection, the individual- ization and modification of interventions and the right timing of surgery.

Acknowledgments

None.CRediTauthorshipcontributionstatement

MichailPapamichail:Conceptualization, Writing - review & editing.MichailPizanias:Writing - original draft.NigelDHeaton:Writing - review & editing.

Funding

None.

Ethicalapproval

Not needed.

Competinginterest

No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the sub- ject of this article.