Hepatic venous pressure gradient in the preoperative assessment of patients with resectable hepatocellular carcinoma

Accepted Manuscript Hepatic venous pressure gradient in the preoperative assessment of patients with resectable hepatocellular carcinoma Alessandro Cucchetti, Matteo Cescon, Rita Golfieri, Fabio Piscaglia, Matteo Renzulli, Flavia Neri, Alberta Cappelli, Federico Mazzotti, Cristina Mosconi, Antonio Colecchia, Giorgio Ercolani, Antonio Daniele Pinna PII: DOI: Reference:

S0168-8278(15)00597-8 http://dx.doi.org/10.1016/j.jhep.2015.08.025 JHEPAT 5811

To appear in:

Journal of Hepatology

Received Date: Revised Date: Accepted Date:

13 March 2015 17 August 2015 17 August 2015

Please cite this article as: Cucchetti, A., Cescon, M., Golfieri, R., Piscaglia, F., Renzulli, M., Neri, F., Cappelli, A., Mazzotti, F., Mosconi, C., Colecchia, A., Ercolani, G., Pinna, A.D., Hepatic venous pressure gradient in the preoperative assessment of patients with resectable hepatocellular carcinoma, Journal of Hepatology (2015), doi: http://dx.doi.org/10.1016/j.jhep.2015.08.025

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Hepatic venous pressure gradient in the preoperative assessment of patients with resectable hepatocellular carcinoma Alessandro Cucchetti; Matteo Cescon; Rita Golfieri; Fabio Piscaglia; Matteo Renzulli; Flavia Neri; Alberta Cappelli; Federico Mazzotti; Cristina Mosconi; Antonio Colecchia; Giorgio Ercolani, Antonio Daniele Pinna Department of Medical and Surgical Sciences, S.Orsola-Malpighi Hospital, Alma Mater Studiorum University of Bologna, Bologna, Italy Correspondence to: Alessandro Cucchetti M.D., Department of Medical and Surgical Sciences – DIMEC; Alma Mater Studiorum - University of Bologna, Via Massarenti, 9; 40138 Bologna; ITALY. Email: [email protected] Running title: portal hypertension in hepatic surgery Keywords: portal hypertension; hepatic venous pressure gradient; hepatocellular carcinoma; surgery; survival. Total manuscript word count: 5300 Number of tables: 4 Number of figures: 2 Abbreviations: HVPG: hepatic venous pressure gradient HCC: hepatocellular carcinoma PHLF: post-hepatectomy liver failure MELD: Model for End-stage Liver Disease PH: portal hypertension EASL: European Association for the Study of the Liver AASLD: American Association for the Study of Liver Diseases (AASLD) CSPH: clinically significant portal hypertension BCLC: Barcelona Clinic Liver Cancer kPa: kilo-Pascal No conflict of interest to declare; no financial supports to declare Authors’ contributions: A. Cucchetti, M. Cescon, R. Golfieri: contributed to the conception and the design of the study; M. Renzulli, C. Mosconi, A. Cappelli: performed HVPG measurements and radiological data collection; F. Neri, F. Mazzotti: organized the study, the procedures and collected clinical data; A. Cucchetti, M. Cescon, F. Piscaglia: analyzed data A. Cucchetti, F. Piscaglia, A. Colecchia: wrote the manuscript; G. Ercolani, A.D. Pinna: added critical revision for important intellectual content . All the authors approved the final version of the manuscript

2

ABSTRACT Backgrounds & Aims: To asses the relationship existing between hepatic venous pressure gradient (HVPG) and the occurrence of post-hepatectomy liver failure (PHLF) grade B/C after resection of hepatocellular carcinoma (HCC) and persistent worsening of liver function.

Methods: Data from 70 consecutive prospectively enrolled HCC patients undergoing resection were collected and analysed. PHLF grade B/C was defined by the International Study Group of Liver Surgery recommendations. The appearance of unresolved decompensation was also analyzed.

Results: Postoperative and 90-day mortality were null. The median HVPG value was 9mmHg (range: 4-18) and the median Model for End-stage Liver Disease (MELD) score was 8 (range: 6-14); 34 patients had an HVPG≥10mmHg (48.6%). Forty-nine patients had an uneventful (Grade A) postoperative course, including 17 with an HVPG≥10mmHg (24.2% of 70 patients). Grade B complications occurred in 20 patients (3 with an HVPG<10mmHg and 17 with an HVPG≥10mmHg; P<0.001); only one grade C complication occurred in a patient with an HVPG<10mmHg, subsequently successfully undergoing liver transplantation. Median MELD score returned to preoperative values after a transient postoperative increase, regardless of the HVPG values; after three months, it returned to the preoperative of 8 in patients with an HVPG<10mmHg and of 9 in patients with an HVPG ≥10mmHg (P=0.077 and 0.076 at paired test, respectively).

Conclusions: The hepatic venous pressure gradient can be used before surgery to stratify the risk of PHLF but the proposed cut-off of 10mmHg excludes approximately one-quarter of the patients who would benefit from surgery without short to mid-term postoperative sequelae.

3

INTRODUCTION Hepatocellular carcinoma (HCC) is the most frequent primary hepatic tumour in patients with liver cirrhosis [1]. Patients having well-compensated liver disease in the absence of clinically significant portal hypertension (PH) are considered optimal candidates for hepatic resection (HR), thanks to the very low risk of a worsening of the postoperative liver function and excellent longterm survival [2]. Accordingly, the European Association for the Study of the Liver (EASL) and the American Association for the Study of Liver Diseases (AASLD) guidelines consider elevated PH to be a contraindication for resection due to the relatively higher risk of postoperative liver decompensation [1,3]. In particular, resection should be reserved for patients with very wellpreserved liver function, defined as normal bilirubin with either hepatic venous pressure gradient (HVPG) <10mmHg or platelet count ≥100,000 [1,3]. Measurement of the HVPG is considered the gold standard for assessing the presence of clinically significant portal hypertension (CSPH) whereas much debate still exists as to whether the indirect signs of PH, such as the presence of esophageal varices detected by endoscopy or splenomegaly (major diameter >12cm) with a platelet count <100.000/mm3, can be considered adequate predictors of PH severity in surgical series [4-8]. At present, the literature regarding the role of the HVPG in risk stratification for postoperative liver failure (PHLF) in HCC patients is surprisingly poor when compared with the weight of the guideline recommendations. In Eastern countries, evaluation for surgery is based on the indocyanine green retention rate (ICG-R) and, in Western countries, many, if not most, centers do not routinely perform venous pressure measurements [2,9]. The first reports regarding the HVPG in HR candidates were published in 1996 and 1999 by the Barcelona Clinic Liver Cancer (BCLC) group and included 29 and 43 surgical patients, respectively [10,11]. On the basis of these two seminal studies, the EASL and the AASLD wrote the current guidelines regarding the role of the HVPG in selecting candidates for resection. For more than a decade, no additional evidence was provided to support these findings [4]. In 2012, two prospective studies were published [8,12]; one of these was again from the BCLC group [12] and included a maximum of 46 resected patients having a maximum HVPG value of 12.5mmHg [12]. Thus, at present there seems to be a great need for additional studies to strengthen the evidence of the current recommendations. The main aim of the present study was to investigate the usefulness of the HVPG in the risk stratification for PHLF, as defined in 2012 by the International Study Group of Liver Surgery (ISGLS) in the largest prospective cohort of patients enrolled to date. Residual mid-term liver function, previously defined as the appearance of unresolved decompensation [10-12], and the competing

4 role of the Model for End-stage Liver Disease (MELD) score as well as patient survival were analysed.

METHODS From October 2009 to November 2014, 217 patients having HCC were resected at our centre. Of them, 70 patients with resectable HCC, with or without clinical signs of portal hypertension [6], prospectively enrolled, were offered and accepted to participate in this longitudinal observational study after approval by the local institutional review board. The observational nature of the study was explained to the patients, and informed consent was obtained in each case, according to the Declaration of Helsinki. The policy of our centre regarding HCC resectability and the role of liver function tests has already been published. Briefly, we consider patients as potential candidates for resection on the basis of the MELD score and, more recently, with the help of liver stiffness measurement to confirm the degree of liver fibrosis [13,14]. Briefly, we considered candidates, even for major hepatectomies, those patients having a liver stiffness value <15 KPa and a MELD score of 6-7; in patients having a MELD score ≥10 and/or liver stiffness value above 20 KPa, resection was proposed, and performed, only if a limited curative resection (with adequate future remnant liver volume) could be planned, and/or as a bridge therapy in the setting of a liver transplantation strategy [13,15]. In those patients with borderline MELD scores and/or liver stiffness values, we allowed the removal of not more than one/two segments but limited curative resection was always the preferred approach. The HVPG value alone was not considered as a determinant for refusing resection in our centre. Patient demographics, laboratory and radiological data, tumour pathology and operative data were collected for all subjects undergoing curative (R0) hepatectomy. None of the patients in the present study population underwent pre-operative portal vein embolization. Intra-operative ultrasound was performed systematically in order to detect additional nodules not revealed preoperatively and to ascertain a tumour-free margin of at least 1cm. During parenchymal transection, clamping maneuvers was always adopted to control for bleeding; central venous pressure was maintained under 5-6 mmHg to prevent from bleeding from hepatic veins. Candidates for liver resection for recurrent tumours were excluded from the study. All 70 patients had chronic hepatitis with or without cirrhosis and belonged to Child–Pugh class A.

5 HVPG measurement Patients underwent hepatic vein catheterisation within a maximum of three weeks before surgery. The procedures were performed by three experienced radiologists with at least ten years of expertise in this setting. Under local anaesthesia, a 7F venous catheter introducer was placed in the right basilic vein of the right forearm using the Seldinger technique. Thereafter, a, 8.5-11mm balloon-tipped catheter was advanced into the right hepatic vein to measure the wedged and the free hepatic venous pressures using an external electro-mechanical transducer and polygraph. The HVPG was calculated as the difference between the wedged and the free hepatic pressures, as previously described [16,17]. All measurements were carried out in triplicate as previously described [16,17]. Endpoints of the study The main endpoint of the study was to assess the relationship existing between the HVPG and the development of post-hepatectomy liver failure grade B/C, immediately after resection as well as in the mid-term, as already previously reported and more recently suggested by expert opinions [4,10-12,18]. To this end, three months after hepatic resection, the patients were reevaluated with a computed tomography scan and biochemical exams in order to detect tumour recurrence and to assess residual liver function [10-12]. The PHLF was defined here on the basis of the ISGLS recommendations, showing a good relationship with postoperative survival in recent series [19,20]. In particular, PHLF grade A corresponds to a substantial regular postoperative course whereas grade B results in a deviation from the regular clinical management but manageable without invasive treatment, and grade C results in a deviation from the regular clinical management requiring invasive treatment. A detailed report of postoperative complications was recorded during the postoperative course and has been presented here. The 50-50 criteria were also recorded [20]. Finally, overall survival was computed for completeness of results. Statistical analysis Continuous variables were expressed as medians and ranges, and the values in the different subgroups were compared using the unpaired Mann-Whitney test (for comparison between groups) or the paired Wilcoxon Signed Rank test (for comparison of post-operative MELD score course within groups). Categorical variables were expressed as prevalence, and the subgroups were compared using the Fisher’s exact test. Overall survival was computed from the day of surgery until the death of the patient; liver transplantation and the last follow-up visit were treated as censoring events (end of observations: 1 May 2015). Differences in survival were

6 investigated using the Kaplan–Meier estimate and the log-rank test. The accuracy of the HVPG in predicting PHLF grade B/C was assessed by measuring the area under the receiver operating characteristic curve (AUROC). Differences between AUROCs were compared using the HanleyMcNeil method. Positive and negative predictive values (PPV and NPV) for the HVPG cut-off point of ≥10mmHg were calculated and reported. Relationships between MELD score, HVPG and liver stiffness values were investigated using both linear regression models and non-linear regression models. A significance level of 0.05 was used in all analyses. The statistical analysis was carried out using SPSS Version 13.0 software (SPSS, Chicago, IL) and MedCalc for Windows, version 12.5 (MedCalc Software, Ostend, Belgium).

RESULTS The entire study population had a median age of 62 years (range: 40 – 84) and a median MELD score of 8 (range: 6 – 14). All patients belonged to Child–Pugh class A. The HVPG measurement was successful in all 70 patients but 2 (2.9%) developed basilic vein thrombosis after the procedure which was treated with heparin with no delay in the surgical procedure. The median HVPG value was 9mmHg (range: 4-18) and 34 patients had an HVPG≥10mmHg (48.6%). Baseline clinical characteristics of the entire study population, and of the subgroups according to the HVPG cut-off, are reported in Table 1. As expected, patients with an HVPG≥10mmHg had higher international normalized ratios (INRs) (P=0.025), lower platelet counts (P=0.001), higher prevalences of oesophageal varices at endoscopy (P=0.003) and slightly higher MELD scores (P=0.010); these patients also had smaller tumours (P=0.023), had more frequently F4 fibrosis stage (P=0.003) and more frequently underwent limited resections (P=0.013) in comparison to patients with an HVPG<10mmHg. Postoperative complications The postoperative (during in-hospital course) and 90-day mortalities were null. The median in-hospital stay was 8 days (range: 5 -55). Forty-nine patients had an uneventful postoperative course (70.0%); of them, 17 had an HVPG≥10mmHg (24.2% of the 70 patients). Grade B complications occurred in 20 patients: 3 with an HVPG<10mmHg (8.3%) and 17 with an HVPG≥10mmHg (50.0%; P<0.001). Only one grade C complication occurred in a patient having an HVPG<10mmHg and this was the only case in which the 50-50 criterion on postoperative day 5 was fulfilled. This patient had a tumor burden within our transplantability criteria and underwent a right hepatectomy, developing postoperative liver failure early on, which required salvage

7 transplantation on postoperative day 10. To date, this patient is currently alive with normal graft function. Details of the postoperative complications are reported in Table 2. Ascites requiring albumin infusion, daily diuretics and/or delay in drainage removal, and pleural effusion requiring oxygen supply, albumin, diuretics and/or thoracentesis were the most common causes of deviation from a normal postoperative. No differences were observed in the median in-hospital stay between patients with and without HVPG≥10mmHg (Table 1; P=0.247). When focusing on the 34 patients with HVPG≥10mmHg, the prevalence of F4 stage of the 17 patients with uneventful postoperative was 64.7% (11 patients), whereas this prevalence was higher in patients with grade B/C complications (94.1%; 16 out of 17 patients; P=0.085). Prevalence of oesophageal varices was similar in these two groups (47.1% versus 41.2%; P>0.999); a low platelet count (<100x103/ml) was observed in 29.4% (5 patients) versus 52.9% (9 patients; P=0.296). From Table 3, it can be noted that patients who experienced an uneventful postoperative course were submitted to a wedge resection in 76.5% of cases (13 patients out of 17), whereas this prevalence was lower in patients with grade B/C complications (35.3%, 6 out of 17 patients; P=0.037). The median liver stiffness values were 21.6 kPa (data available for 12 patients; range: 6.5 – 30.3) in uncomplicated patients and 30.8 kPa (17 patients; range: 14.1 – 42.7) in complicated ones (P=0.007). It can be also noted that in patients with HVPG≥10mmHg but with a MELD score <10, undergoing wedge resections, PHLF B/C occurred in 2 out of 14 cases (14.3%); in similar patients submitted to more extended hepatectomies, PHLF B/C occurred in 8 out of 12 cases (66.7%). In the remaining 8 patients with MELD ≥10, PHLF B/C occurred in 7 cases (87.5%), even if in presence of limited resections.

Postoperative laboratory course The postoperative course of the MELD score is depicted in Figure 1. As predictable, in the early postoperative course, a significant increase in MELD values was observed in patients with either low or high HVPGs which subsequently grossly returned to preoperative levels at 90 days after surgery. In patients with an HVPG<10mmHg, after the exclusion of the patient undergoing salvage transplantation 10 days after surgical resection, the median MELD score at 90 days was 8 (range: 6 – 13), similar to the preoperative value of 8 (range: 6 – 14; P=0.077). Of these, only one patient showed mild ascites at ultrasound examination at 90 days after surgery and was treated with an increase in diuretics.

8 In patients with an HVPG≥10mmHg, the median MELD score at 90 days was 9 (range: 7 – 27), very close to the preoperative value of 8.5 (range: 6 – 11; P=0.076). At the three months evaluation, one patient was suffering from persistent liver decompensation (1 out of 34; 2.9%) but was not a candidate for liver transplantation because of concomitant breast cancer, one patient showed jaundice resulting from concomitant cholangitis, due to a previous papillary sphincterotomy performed to remove choledocal gallstones, and one patient had mild ascites at ultrasound examination but with normal laboratory function tests. The remaining 31 patients (91.2% of patients with PH) returned to the same preoperative liver function status. During a mean follow-up of 17 months (range: 10 days – 5 years), 13 patients were subsequently put on the waiting-list for liver transplantation. Of the 13 listed patients, 6 had a preoperative HVPG <10mmHg and 7 had a preoperative value ≥10mmHg. Of these, as previously reported, one underwent salvage transplantation at postoperative day 10 due to PHLF grade C; all the remaining patients were listed for tumour recurrence either within the Milan criteria or within a downstaging protocol. Of the 13 patients listed, only one with tumor recurrence died while on the waiting list as a result of hepatic decompensation, while the remaining listed patients are currently alive in waiting list as of May 2015. Overall, three patients died during follow-up, all with HVPG≥10mmHg (Table 3). The 1- and 3-year survival rates of patients with an HVPG <10mmHg was both 100%, whereas those of patients with an HVPG≥10mmHg were 97.1% and 79.4%, respectively (Figure 2; P=0.112). Discrimination of PHLF grade B/C The ROC curve analysis for PHLF B/C showed an area of 0.727 (95%C.I.: 0.598 – 0.857) for the MELD score and of 0.792 (95%C.I.: 0.677 – 0.907) for the HVPG (P=0.475). Narrowing this analysis to the 58 patients having also liver stiffness measurement, the AUROC for liver stiffness was 0.813 (95%C.I.: 0.688 – 0.940), not significantly different to that of HVPG (AUROC: 0.799; 95%C.I.: 0.677 – 0.920; P=0.785) or that of the MELD score (AUROC: 0.739; 95%C.I.: 0.604 – 0.874; P=0.321). The PPV and the NPV of an HVPG≥10mmHg were 50% and 88.9%, respectively. The investigation of relationship existing between MELD, HVPG and liver stiffness returned the higher r-squares for cubic interpolations as follows: MELD and HVPG = 0.207 (70 patients); MELD and liver stiffness = 0.283 (58 patients) and HVPG and liver stiffness = 0.539 (58 patients).

9

DISCUSSION The guidelines of the European Association for the Study of the Liver as well as those of the American Association for the Study of Liver Diseases currently consider clinically significant PH as a contraindication for liver resection in HCC, recommending other treatments [1,3]. However, there is no consensus among the various centres regarding the role of PH in the decision-making process regarding surgery, and few studies have specifically addressed this issue including various definitions of PH [4]. In particular, up to now, direct measurement of the HVPG has been carried out in only four studies over two decades, including a maximum of 46 surgical patients studied in this manner (Table 4). On the other hand, whether the preoperative indirect measurement of PH, based on the presence of oesophageal varices, splenomegaly and platelet count, has or does not have an impact on the outcome of surgery is under debate, due to an unclear relationship between clinical signs of PH and that of HVPG [2,4,8]. In 2015, a meta-analysis was conducted to explore the impact of clinically significant PH on postoperative outcomes (survival and clinical decompensation) in detail [4]. However, even this evidence was not convincing enough to definitively prove PH as a criterion for avoiding surgery in HCC patients, mainly due to the heterogeneity of end-points and definitions used [18,22,23]. Even if investigators acknowledged for potential biases, the investigators' conclusion on the absolute necessity to use HVPG in the diagnosis of CSPH, and on excluding patients with CSPH from surgery, were questionable [22,23] Some correspondents suggested an incorrect analytic approach [22] whereas other suggested that short- and long-term overall survival was similar in the presence or absence of PH if the patients had early-stage HCC or had undergone minor hepatectomy, or if propensity-score matching was used to eliminate baseline differences [6, 23]. On one critical point, however, various authors are all in agreement; additional prospective data are needed to obtain a clearer evidence for considering an HVPG≥10mmHg as an absolute contraindication for surgery [4,17,22,23]. The aim of the present prospective study was to further investigate the role of the HVPG in the decision-making process for hepatic resection of HCC. The results were obtained from a tertiary referral hospital experience with a patient selection policy which allows considering patients with an HVPG≥10mmHg as surgical candidates [13,14,24]. The most interesting finding was that approximately one-quarter of the patients enrolled experienced a normal and uneventful postoperative course in spite of having a pressure gradient ≥10mmHg, but, on the basis of the guidelines, they would not be candidates for resection. After only three months from surgery, liver

10 function tests returned to the same preoperative levels in all but one patient with PH, also including those patients who experienced transient PHLF. This is an important aspect which must be weighed against available alternative treatments. In fact, clinicians may be more interested in understanding the prognostic difference between surgery and other treatment modalities, providing a direct comparison for patients with significant PH [18]. What would have been the alternative treatment to be adopted for patients with PH? As recently outlined by Roayaie et al. for HCC patients who were “not ideal candidates” for resection, surgery was associated with better survival as compared to embolization and other treatments but was inferior to transplantation and ablation [2]. Liver transplantation applicability is strongly influenced by the chronic shortage of donors and, in our program, it is not a-priori excluded from the treatmentalgorithm of HCC, configuring hepatic resection as a bridge therapy to transplantation. [24]. The present results suggest that this policy does not harm patients undergoing hepatic resection, even in the presence of PH. The most frequent postoperative complication was, in fact, the occurrence of ascites which required the delayed removal of abdominal drains, the infusion of albumin and the use of diuretics, but clinical decompensation was transient and liver function returned to baseline levels just three months after surgery. As regards ablation, its superiority in “non-ideal candidates” for surgery, as suggested by the Roayaie results, can be the consequence of the consistent clinical difference reported between resected candidates versus ablated candidates and the Roayaie finding has to be interpreted carefully. The Author’s intuition surely opened new pathways for additional comparative studies between resection and ablation in patients with PH, in the light of the retrospective data and randomised controlled trials available [2,25-28]. We believe that HVPG can be useful in the pre-operative assessment of PH but also that the measurement should not lead to an absolute contraindication for surgery; instead, it should be used to better balance the risks and the benefits when considering surgery against other locoregional therapies, and to plan the extent of the resection to be performed. In our experience, patients with an HVPG≥10mmHg underwent resection since only limited parenchymal curtailment needed to curatively remove the tumour (Table 1 and 3). A reasonable use of the information provided by the HVPG must be considered from this point of view: similarly to what is in use in Eastern countries regarding the indocyanine green retention rate [9], a relative high HVPG should not contraindicate hepatic resection a priori if the tumour can be removed with a relatively limited resection. Of note, uneventful postoperative course was observed in patients with less advanced liver fibrosis as mainly measured by liver stiffness and MELD score, so that the combination of

11 these informations can be the way to select optimal candidates for surgery, even in presence of high HVPG values. Sub-stratification suggested that in patients with HVPG≥10mmHg but with a MELD score still below 10, surgery can be safety attempted if a limited resection can be pursued. This finding supports Roayaie results showing that patients with PH can safely undergo resection only if they have excellent liver function [2]. On the contrary, the presence of MELD≥10 and/or the need for a more extended hepatic resection will be burdened of a PHLF probability that would be unacceptable by some surgical centres even if a recovering from transient liver dysfunction was observed in most of cases. Further studies providing a decision algorithm combining all these features are warranted, with a special attention to the extent of hepatectomy. From the present analysis, there is another important aspect which merits discussion. The AUROC of HVPG was found to be higher (0.792) than that of MELD score (0.727) although not statistically different. A poor relationship was observed between these two measurements (best rsquare = 0.207). When focusing the analysis on the 58 patients having also liver stiffness measurement, the AUROC of HVPG and liver stiffness were quite similar, with a improved relationship between these two measurements (best r-square = 0.539). Taking togheter these results, it can be suggested that MELD score provides valuable information regarding the probability of complicated postoperative course that is independent of HVPG and liver stiffness, and that liver stiffness can be used as a non-invasive predictor of the severity of PH with acceptable accuracy [12]. It was obviously to be pointed out that It was impossible to determine here if these findings were true-positive results or represented a type II error. However, it should be noted that, to date and to the best of our knowledge, the present study is the largest carried out on this topic comparing HVPG, MELD score and liver stiffness predictive accuracies. Therefore, larger prospective studies are needed to confirm the eventual superiority of HVPG over MELD as well as an eventual comparison with ICG-R in order to assess their utility in the preoperative assessment of surgical candidates. Another potential limitation of our study could be the relatively short follow-up of the entire study population, with only two patients having at least 5-year survival after surgery. In particular, the statistical absence of a survival difference between the two groups (Figure 2; P=0.112) does not necessarily mean the “evidence of absence”. This is probably the consequence of the small power due to numbers and relatively short follow-up, which makes difficult to reach robust conclusions in this regards. However, it should be pointed out that previous published experiences did not provide strong results in this regard. In fact, Bruix and Llop did not report data

12 on survival, and a median follow-up of only 8 months was reported by Llop; Boleslawski reported a maximum of 3-year survival. In 1999, Llovet et al. reported a longer survival analysis which, unfortunately, was a mixture of patients with and without HVPG assessment. Considering these aspects, the present analysis provided adequate data; however, studies regarding the preoperative HVPG, with a longer longitudinal assessment, are necessary. In fact, we need to acknowledge that even if the present study is the largest conducted to date, the sample size remains limited, and to confirm (or to refuse) our findings larger studies are strongly awaited. In conclusion, the present results confirmed that an HVPG≥10mmHg was associated with a higher risk of ascitic decompensation in the short term after surgery; however, the too strict application of this cut-off excludes a large proportion of potential surgical candidates since liver function tests were not deteriorated significantly 3 months after surgery. Conversely, the measurement of the HVPG should be utilised to modulate the planning for a hepatectomy, avoiding large parenchymal curtailment in patients with significant portal hypertension, in order to provide a good postoperative outcome. A definitive answer regarding the use of the HVPG in HCC patients should come from the comparison of different treatment modalities in patients with an HVPG≥10mmHg. Of note, the accuracy of the HVPG in predicting PHLF does not seem to be superior to a simple non-invasive index, such as the MELD score. Present results can also be used to update meta-analyses in this regards.

13

REFERENCES 1. European Association for the Study of the Liver; European Organisation for Research and Treatment of Cancer. EASL-EORTC clinical practice guidelines: management of hepatocellular carcinoma. J Hepatol 2012;56:908-943. 2. Roayaie S, Jibara G, Tabrizian P, Park JW, Yang J, Yan L, et al. The Role of hepatic resection in the treatment of hepatocellular cancer. Hepatology. 2015; 62:440-51. 3. Bruix J, Sherman M, Practice Guidelines Committee, American Association for the Study of Liver Diseases. Management of hepatocellular carcinoma. Hepatology 2005; 42: 1208-1236. 4. Berzigotti A, Reig M, Abraldes JG, Bosch J, Bruix J. Portal hypertension and the outcome of surgery for hepatocellular carcinoma in compensated cirrhosis: A systematic review and metaanalysis. Hepatology. 2015;61:526-36. 5. Capussotti L, Ferrero A, Vigan `o L, Muratore A, Polastri R, Bouzari H. Portal hypertension: contraindication to liver surgery? World J Surg 2006; 30: 992–999. 6. Cucchetti A, Ercolani G, Vivarelli M, Cescon M, Ravaioli M, Ramacciato G et al. Is portal hypertension a contraindication to hepatic resection? Ann Surg 2009; 250: 922–928. 7. Ishizawa T, Hasegawa K, Aoki T, Takahashi M, Inoue Y, Sano K et al. Neither multiple tumors nor portal hypertension are surgical contraindications for hepatocellular carcinoma. Gastroenterology 2008; 134: 1908–1916. 8. Boleslawski E, Petrovai G, Truant S, Dharancy S, Duhamel A, Salleron J, et al. Hepatic venous pressure gradient in the assessment of portal hypertension before liver resection in patients with cirrhosis. Br J Surg. 2012;99:855-63. 9. Kudo M, Izumi N, Kokudo N, Matsui O, Sakamoto M, Nakashima O, et al. Management of hepatocellular carcinoma in Japan: Consensus-Based Clinical Practice Guidelines proposed by the Japan Society of Hepatology (JSH) 2010 updated version. Dig Dis. 2011;29:339-64. 10. Bruix J, Castells A, Bosch J, Feu F, Fuster J, Garcia-Pagan JC, et al. Surgical resection of hepatocellular carcinoma in cirrhotic patients: prognostic value of preoperative portal pressure. Gastroenterology. 1996;111:1018-22 11. Llovet JM, Fuster J, Bruix J. Intention-to-treat analysis of surgical treatment for early hepatocellular carcinoma: resection versus transplantation. Hepatology. 1999;30:1434-40. 12. Llop E, Berzigotti A, Reig M, Erice E, Reverter E, Seijo S, et al. Assessment of portal hypertension by transient elastography in patients with compensated cirrhosis and potentially resectable liver tumors. J Hepatol. 2012;56:103-8.

14 13. Cescon M, Colecchia A, Cucchetti A, Peri E, Montrone L, Ercolani G, et al. Value of transient elastography measured with FibroScan in predicting the outcome of hepatic resection for hepatocellular carcinoma. Ann Surg. 2012;256:706-12 14. Cescon M, Cucchetti A, Grazi GL, Ferrero A, Viganò L, Ercolani G, et al. Indication of the extent of hepatectomy for hepatocellular carcinoma on cirrhosis by a simple algorithm based on preoperative variables. Arch Surg. 2009;144:57-63 15. Kim SU, Ahn SH, Park JY, Kim do Y, Chon CY, Choi JS, Kim KS, Han KH. Prediction of postoperative hepatic insufficiency by liver stiffness measurement (FibroScan((R))) before curative resection of hepatocellular carcinoma: a pilot study. Hepatol Int. 2008;2:471-7. 16. Groszmann RJ, Wongcharatrawee S. The hepatic venous pressure gradient: anything worth doing should be done right. Hepatology. 2004;39:280-2.ù 17. Berzigotti A, Casadei A, Magalotti D, Castaldini N, Losinno F, Rossi C, et al. Renovascular impedance correlates with portal pressure in patients with liver cirrhosis. Radiology. 2006;240:581-6. 18. Yang T, Lau WY, Zhang H, Wu M, Shen F. Should surgery be carried out in patients with hepatocellular carcinoma with portal hypertension? Hepatology. 2015 Jan 16. doi: 10.1002/hep.27703. [Epub ahead of print]. 19. Rahbari NN, Garden OJ, Padbury R, Brooke-Smith M, Crawford M, Adam R, et al. Posthepatectomy liver failure: a definition and grading by the International Study Group of Liver Surgery (ISGLS). Surgery. 2011;149:713-24. 20. Fukushima K, Fukumoto T, Kuramitsu K, Kido M, Takebe A, Tanaka M, et al. Assessment of ISGLS definition of posthepatectomy liver failure and its effect on outcome in patients with hepatocellular carcinoma. J Gastrointest Surg. 2014;18:729-36. 21. Balzan S, Belghiti J, Farges O, Ogata S, Sauvanet A, Delefosse D, et al. The "50-50 criteria" on postoperative day 5: an accurate predictor of liver failure and death after hepatectomy. Ann Surg. 2005;242:824-8 22. Cucchetti A, Cescon M, Pinna AD. Portal hypertension on the outcome of surgery for hepatocellular carcinoma in compensated cirrhosis: A systematic review and meta-analysis. More doubts than clarities. Hepatology. 2015 Jan 16. doi: 10.1002/hep.27702. [Epub ahead of print].

15 23. Zhong JH, Li LQ. Portal hypertension should not be a contraindication of hepatic resection to treat hepatocellular carcinoma with compensated cirrhosis. Hepatology. 2015 Jan 16. doi: 10.1002/hep.27700. [Epub ahead of print] 24. Cucchetti A, Vitale A, Del Gaudio M, Ravaioli M, Ercolani G, Cescon M, et al. Harm and benefits of primary liver resection and salvage transplantation for hepatocellular carcinoma. Am J Transplant. 2010;10:619-27 25. Livraghi T, Meloni F, Di Stasi M, Rolle E, Solbiati L, Tinelli C, et al. Sustained complete response and complications rates after radiofrequency ablation of very early hepatocellular carcinoma in cirrhosis: Is resection still the treatment of choice? Hepatology. 2008;47:82-9. 26. Chen MS, Li JQ, Zheng Y, Guo RP, Liang HH, Zhang YQ, et al. A prospective randomized trial comparing percutaneous local ablative therapy and partial hepatectomy for small hepatocellular carcinoma. Ann Surg 2006;243:321-8. 27. Huang J, Yan L, Cheng Z, Wu H, Du L, Wang J, et al. A randomized trial comparing radiofrequency ablation and surgical resection for HCC conforming to the Milan criteria. Ann Surg 2010;252:903-12 28. Feng K, Yan J, Li X, Xia F, Ma K, Wang S, et al. A randomized controlled trial of radiofrequency ablation and surgical resection in the treatment of small hepatocellular carcinoma. J Hepatol 2012;57:794-802

16

LEGENDS TO FIGURES Figure 1. Postoperative course of model for end-stage liver disease (MELD) score at different time points of patients with and without hepatic venous pressure gradient (HVPG) ≥10mmHg. POD = post-operative day. For both patients with HVPG value <10mmHg and ≥10mmHg the median delta-MELD between 90-days and pre-operative values was zero (IQR:0 – 1). The mean change of MELD was of 0.2 in patients with HVPG<10mmHg and of 0.8 in patients with HVPG≥10mmHg.

Figure 2. Patient survival after surgery in relationship with hepatic venous pressure gradient (HVPG) value <10mmHg (continuous line) or ≥10mmHg (dashed line; P=0.112). Three patients died during follow-up, all having HVPG≥10mmHg: one patient listed for liver transplantation because of tumor recurrence died for subsequent hepatic decompensation; one patient died of sepsis as a consequence of severe cholangitis due to a previous papillary sphincterotomy performed to remove choledocal gallstones and one patient died of liver decompensation but was not transplantable because of concomitant breast cancer. One patient with HVPG<10mmHg was transplanted on postoperative day 10: survival was censored the day prior to transplantation.

17

18

19

Table 1. Clinical characteristics of the entire study population and in relationship to the HVPG subgroups. All patients (n=70)

HVPG <10mmHg (n=36)

HVPG≥10mmHg (n=34)

p

Age (years)

62 (40 – 84)

62.5 (44 – 84)

62 (40 – 82)

0.529

Gender (male)

62 (88.6%)

35 (97.2%)

27 (79.4%)

0.026

HCV

46 (65.7%)

23 (63.9%)

23 (37.6%)

0.804

HBV

13 (18.6%)

9 (25.0%)

4 (11.8%)

0.221

Alcohol/Other

18 (25.7%)

8 (22.2%)

10 (29.4%)

0.588

Bilirubin (mg/dL)

0.73 (0.35 – 2.07)

0.67 (0.35 – 2.07)

0.77 (0.33 – 2.05)

0.145

4.3 (3.0 – 5.1)

4.3 (3.4 – 5.1)

4.2 (3.0 – 5.2)

0.121

INR

1.14 (1.00 – 1.56)

1.10 (1.00 – 1.56)

1.16 (1.00 – 1.48)

0.025

Creatinine (mg/dL)

0.83 (0.50 – 1.89)

0.88 (0.68 – 1.89)

0.77 (0.50 – 1.36)

0.073

AST (IU/L)

44 (15 – 148)

42 (17 – 137)

50 (15 – 148)

0.210

ALT (IU/L)

46 (11 – 182)

45 (12 – 182)

52 (11 – 141)

0.434

Platelet count (103 /ml)

140 (34 – 445)

158 (68 – 445)

113 (34 – 188)

0.001

Oesophageal varices

19 (27.1%)

4 (11.1%)

15 (44.1%)

0.003

Live Stiffness (kPa) *

17.9 (3.8 – 42.7)

9.9 (3.8 – 17.7)

26.3 (6.5 – 42.7)

0.001

Hepatitis infection

Albumin (g/dL)

Fibrosis stage **

0.003

F1-2

9 (15.6%)

9 (25.0%)

0 (0.0%)

F3

15 (21.4%)

8 (22.2%)

7 (20.6%)

F4

46 (65.7%)

19 (52.8%)

27 (79.4%)

8 (6 – 14)

8 (6 – 14)

8.5 (6 – 11)

0.010

3.1 (1.0 – 10.0)

3.5 (1.0 – 10.0)

2.5 (1.0 – 7.0)

0.023

Single nodule

45 (64.3%)

23 (63.9%)

22 (64.7%)

0.999

HVPG (mmHg)

9 (4 – 18)

7 (4 – 9)

12.5 (10 – 18)

0.001

Laparoscopic resection

2 (3.1%)

2 (5.6%)

0 (0.0%)

0.500

MELD score Tumour size (cm)

Extension of hepatectomy

0.013

Wedge/segmentectomy

45 (64.3%)

18 (50.0%)

27 (79.4%)

Two or more segments

25 (35.7%)

18 (50.0%)

7 (20.6%)

20

In-hospital stay (days) ***

8 (5 – 55)

8 (5 – 55)

9 (5 – 23)

0.247

All patients have either viral or non-viral chronic hepatitis and belonged to Child–Pugh class A. One patient can have more than one cause of chronic hepatitis, so the sum is not necessarily 100%. Tumour characteristics reported are those based on radiological imaging. Continuous variables are reported as medians and ranges. * Available for 58 patients (29 patients in each group) ** METAVIR fibrosis stage. The 7 patients without cirrhosis and HVPG≥10mmHg showed a maximum value of 13mmHg. *** The in-hospital stay of 55 days was due to one patient submitted to salvage liver transplantation; the removal of this patient has not impact on the results.

21 Table 2. Postoperative complications grade B/C according to the International Study Group of Liver Surgery (ISGLS) proposal.

At least one Grade B/C complication

All patients (n=70)

HVPG <10mmHg (n=36)

HVPG≥10 mmHg (n=34)

21

4 (11.1%)

17

(30.0%)

Ascites and/or pleural effusion requiring medical therapy and/or drainage

Ascites requiring albumin, diuretics and/or drainage

17

(50.0%)

3 (8.3%)

(24.3%)

15

(41.2%)

2 (5.6%)

(21.4%)

Pleural effusion requiring albumin, diuretics and/or

8

drainage

(11.4%)

Need for FFP transfusion or an INR value between 1.5

6 (8.6%)

14

13 (38.2%)

1 (2.8%)

7 (20.6%)

P 0.00 1

0.00 2

0.00 1

0.02 6

0 (0.0%)

6 (17.6%)

and 2.0

0.01 0

Bilirubin above 3mg requiring additional radiological

3 (4.3%)

1 (2.8%)

2 (5.9%)

and laboratory tests

Inadequate urine output (≤0.5 ml/kg/h)

0.60 9

1 (1.4%)

0 (0.0%)

1 (2.9%)

0.48 9

Fever with modification replacement

positive blood cultures requiring of antibiotics and central line

Biliary leak requiring delay in drainage removal

8 (11.4%)

1 (2.8%)

7 (20.6%)

0.02 6

1 (1.4%)

0 (0.0%)

1 (2.9%)

0.48 9

Other

5 (7.1%)

1 (2.8%)

4 (11.8%)

0.19 2

Need for salvage transplantation

1 (1.4%)

1 (2.9%)

0 (0.0%)

0.48 9

FFP: fresh frozen plasma. One patient can experience more than one concomitant complication.

22 Table 3. Clincal characteristics and outcome of the 34 patients with HVPG≥10 mmHg ordered by occurrence of post-hepatectomy liver failure (PHLF grades B/C) and HVPG values. Patient (sex/yrs)

Bilirubin (mg/dL)

MELD

LS (kPa)

HVPG (mmHg)

Resection type

PHLF B/C

3-m Decomp

Waiting list for transplant

Current status

M/62

0.85

8

30.3

10

Wedge

No

No

No

Alive

M/82

0.69

8

8.8

10

Left hep.

No

No

No

Alive

M/76

1.08

8

NA

10

Bisegm.

No

No

No

Alive

M/75

0.33

8

NA

10

Wedge

No

No

No

Alive

M/63

0.34

8

NA

10

Segm.

No

No

No

Alive

M/76

0.37

8

21.3

11

Wedge

No

No

No

Alive

M/50

0.84

8

26.3

11

Wedge

No

No

No

Alive

F/65

0.95

9

NA

12

Wedge

No

No

No

Alive

M/61

1.04

8

NA

12

Wedge

No

No

No

Alive

M/62

0.54

8

17.6

12

Wedge

No

No

No

Alive

M/65

0.68

8

11.0

13

Trisegm.

No

No

No

Alive

M/50

0.78

7

25.1

13

Wedge

No

No

No

Alive

M/57

0.45

6

6.5

14

Wedge

No

No

No

Alive

M/48

1.66

11

21.1

14

Wedge

No

No

Recurrence

Alive

F/51

0.48

9

21.8

15

Wedge

No

No

Recurrence

Alive

M/62

0.44

7

23.6

17

Wedge

No

No

No

Alive

F/65

0.68

9

21.8

17

Wedge

No

Sepsis

No

Sepsis

M/65

0.95

9

14.1

10

Segm.

Yes

No

No

Alive

M/40

0.71

11

26.3

11

Bisegm.

Yes

No

Recurrence

Alive

M/62

0.87

7

15.5

11

Trisegm.

Yes

No

No

Alive

F/64

0.82

9

24.0

11

Segm.

Yes

No

Recurrence

Alive

M/61

0.59

7

16.0

11

Bisegm.

Yes

No

No

Alive

M/66

0.75

8

32.4

12

Bisegm.

Yes

Yes

No

Alive

M/75

1.69

11

35.8

12

Wedge

Yes

Yes

No

Alive

M/50

0.55

9

30.5

14

Segm.

Yes

No

Recurrence

Liver failure

F/71

1.62

10

48.0

14

Segm.

Yes

No

No

Alive

M/53

0.62

8

26.3

14

Wedge

Yes

No

No

Alive

F/60

0.81

9

39.1

16

Segm.

Yes

No

No

Liver failure

M/55

0.73

10

46.4

16

Wedge

Yes

No

No

Alive

M/56

0.76

8

38.6

16

Segm.

Yes

No

No

Alive

M/56

2.05

11

42.7

16

Segm.

Yes

No

Recurrence

Alive

M/56

1.72

10

46.4

16

Wedge

Yes

No

No

Alive

23 M/66

1.14

9

15.2

17

Wedge

Yes

No

No

Alive

F/71

1.58

10

30.8

18

Wedge

Yes

No

No

Alive

LS=liver stiffness; waiting-list for liver transplantation includes causes of listing.

24 Table 4. Studies published during the last two decades reporting data on HVPG in surgical series.

Author, year Institution Bruix, 1996 [10]

Inclusion period

Study design

1991-

Prospective

Patients with

No. of surgical patients HVPG≥10 mmHg 29

1994

15 (51.7%)

BCLC group, University of Barcelona, Spain

Llovet, 1999 [11]

1989-

Prospective

43

1997

21 (48.8%)

BCLC group, University of Barcelona, Spain

Boleslawski, 2012 [8] Univesity of Lille, France

20072009

Prospective

40

18 (45.0%)

Main end-points

Main findings

Appearance of unresolved decompensation after 3 months

Eleven patients (39.7%) had unresolved hepatic decompensation 3 months after surgery all of them with HVPG ≥10mmHg. Preoperative HPVG of patients with unresolved decompensation was 13.9 versus 7.4 mmHg of patients without it.

Appearance of unresolved decompensation

The best candidates for resection were those without clinically relevant portal hypertension or HVPG <10mmHg. The OR for decompensation of patients with HVPG≥10mmHg was 13.1 with very wide confidence interval (1.6 – 103).

50-50 criteria

Post-operative complications occurred in 29/40 (72.5%) of patients; 90-days mortality was 8.6%. A raised HVPG was associated with post-operative liver dysfunction (median 11 and 7 mmHg in those with and without dysfunction respectively; P=0.017) and 90-day mortality (12 and 8 mmHg in those who died and

Dindo-Clavien complications 90-day mortality

25 survivors; P = 0.026). Llop, 2014 [12]

2007-

Prospective

46

2011

10 (21.7%)

BCLC group, University of Barcelona, Spain

Present study University of Bologna, Italy

Usefulness of liver stiffness. Post-operative complications

20092014

Prospective

70

34

PHLF defined by

(48.6%)

ISGLS 90-day mortality Detailed clinical evaluation after 3 months

Thirty-six (78.3%) patients had HVPG <10mmHg; 10 additional patients, with HVPG between 10.5 and 12.5 mm Hg and a favorable location of the tumor, were resected. Three of them, all with HCC recurrence developed ascites at a median of 8 months. Postoperative and 90day mortality were null even in the 34 patients (48.6%) with HVPG ≥10mmHg The cut-off of 10mmHg excludes 24.2% of patients who will benefit from surgery. Median MELD score returned to preoperative values of 8 in patients with an HVPG<10mmHg and of 9 in patients with an HVPG ≥10mm. The HVPG accuracy was not dissimilar to that of liver stiffness or MELD score.