Ischemia-reperfusion injury and the risk of hepatocellular carcinoma recurrence after deceased donor liver transplantation

This study aimed to evaluate the effects of ischemia-reperfusion injury (IRI) on the risk of hepatocellular carcinoma (HCC) recurrence after liver transplantation. Data of 195 patients were retrospectively analysed. Post-reperfusion aspartate (AST), alanine transaminase, and lactate dehydrogenase (LDH) levels were the primary measures of IRI. Tumour recurrence was the primary endpoint. Post-reperfusion AST was a continuous risk factor for tumour recurrence in patients within Milan criteria (p = 0.035), with an optimal cut-off of 1896 U/L. Recurrence-free survival of patients within Milan criteria and post-reperfusion AST of <1896 and ≥1896 U/L was 96.6% and 71.9% at 5 and 3.7 years, respectively (p = 0.006). Additionally, post-reperfusion AST and LDH exceeding the upper quartile significantly increased the risk of HCC recurrence in patients within Milan criteria (p = 0.039, hazard ratio [HR] = 5.99 and p = 0.040, HR = 6.08, respectively) and to a lesser extent, in patients within Up-to-7 criteria (p = 0.028, HR = 3.58 and p = 0.039, HR = 3.33, respectively). No other significant IRI effects were found in patients beyond the Up-to-7 criteria and in analyses stratified for independent risk factors for recurrence: tumour number and differentiation, alpha-fetoprotein, and microvascular invasion. Thus, IRI exerts major negative effects on the risk of HCC recurrence after liver transplantation in patients within standard and extended criteria.

deceased donors and to a lesser extent, partial grafts procured from living donors may be more susceptible to IRI, the use of these grafts may increase the risk of post-transplant HCC recurrence. This hypothesis was subject to numerous studies with inconsistent results. Although transplantations of grafts procured from living donors or high-risk grafts procured from deceased donors after cardiac death or those who were older and had hepatic steatosis or other risk factors were reported to have adverse effects on outcomes after liver transplantations for HCC in several studies, the results of available studies are not completely consistent [17][18][19][20][21][22][23] . However, recent reports found that prolonged ischemic times, directly related to the magnitude of IRI, increased the risk of post-transplant HCC recurrence 24,25 . Nevertheless, data on the direct effect of the magnitude of IRI on the risk of HCC recurrence after deceased donor liver transplantation are limited. Therefore, this study aimed to evaluate the association between the degree of graft IRI as indicated by post-reperfusion transaminase and lactate dehydrogenase (LDH) levels and the risk of post-transplant HCC recurrence after deceased donor liver transplantation with respect to patients' initial risk profile.

Methods
This was a retrospective observational study. In total, 250 liver transplantations were performed for HCC patients between January 2001 and June 2016 at the Department of General, Transplant and Liver Surgery (Medical University of Warsaw). Patients with fibrolamellar HCCs and those with combined HCC/cholangiocarcinoma were not included. After exclusion of 55 patients with missing measurements of transaminase levels 2 h after reperfusion, the final study cohort comprised 195 liver transplant recipients. The study protocol was approved by the institutional review board of the Medical University of Warsaw. Informed consents were not obtained from the patients due to the retrospective nature of the study, which is in line with institutional review board and national regulations. All methods were performed in accordance with the relevant guidelines and regulations. No organs were procured from prisoners.
The degree of IRI was represented by three variables, namely, serum alanine transaminase (ALT), serum aspartate transaminase (AST), and serum LDH levels; each was assessed from a blood sample obtained 2 h after portal reperfusion. These variables were the primary factors of interest. Peak serum bilirubin concentration, international normalised ratio (INR), and gamma-glutamyl transpeptidase (GGTP) activity over the first 7 post-transplant days were additionally analysed as variables associated with IRI. The duration of cold and warm ischemia was defined as the time from clamping of the donor aorta until the removal of the graft from the preservation solution and that from the removal of the graft from the cold preservation solution until portal reperfusion, respectively. The sum of cold and warm ischemic times formed the total ischemic time. All grafts were procured from donors after brain death. Tumour recurrence over the 5-year post-transplant observation period was the primary end-point. Recurrence-free survival was calculated from the date of transplantation until tumour recurrence and censored at the date of last available follow-up, death for non-HCC related causes or 5 years post-transplantation (whichever occurred first). Details on the surgical technique, perioperative care, immunosuppression, and follow-up protocol are provided elsewhere 26,27 .
First, post-reperfusion ALT, AST, and LDH levels were assessed for their potential effect on the risk of post-transplant tumour recurrence in all patients. Other independent predictors of recurrence were also assessed, including peak post-transplant bilirubin concentration, INR, and GGTP activity. Furthermore, the analyses were adjusted for their potential confounding effects in bivariable analyses. Subgroup analyses were subsequently performed to determine the potential differences in associations between IRI degree and the risk of HCC recurrence according to patients' initial risk profile, based on fulfilment of selection criteria and established independent predictors of recurrence.
Continuous and categorical variables are given as medians (interquartile ranges) and numbers (percentages). The Kaplan-Meier method was used for survival calculations, and log-rank test was used for intergroup comparisons. A Cox proportional hazards regression analysis was performed to evaluate the associations between particular factors and the risk of recurrence. A multivariable model was created using forward stepwise method with p thresholds of 0.05 and 0.150 for inclusion and exclusion of variables, respectively. An additional series of bivariable analyses were performed to adjust the effects of IRI to potential confounding effects of independent risk factors for tumour recurrence. Spearman correlation coefficients were calculated to evaluate the associations between ischemic times and donor age and post-reperfusion laboratory measurements. Post-reperfusion AST, ALT, and LDH levels; peak post-transplant bilirubin concentration; and peak post-transplant GGTP activity were transformed to their natural logarithms prior to their analyses as continuous variables. Additionally, they were assessed as categorical factors using the upper quartile for division. Receiver operating characteristic (ROC) curves were constructed to determine the optimal cut-offs of continuous factors in predicting recurrence. Hazard ratios (HRs) and c-statistics were presented with 95% confidence intervals (95% CIs). Significance threshold was set to two-tailed p values of 0.05. Analyses were computed in STATISTICA version 13 (Dell Inc., Tulsa, USA) software. The datasets generated and/or analysed during the current study are available from the corresponding author on reasonable request.

Results
The characteristics of the 195 patients are shown in Table 1. Median AST, ALT, and LDH levels assessed 2 h post-reperfusion were 850 U/L (interquartile range: 486-1625 U/L; range 153-14375 U/L), 566 U/L (interquartile range: 304-935 U/L; range 102-9912 U/L), and 2240 U/L (interquartile range: 1322-4670 U/L; range 385-38207 U/L), respectively. Post-reperfusion AST and ALT levels were significantly, yet poorly correlated, with total (both p = 0.001) and cold ischemic times (both p < 0.001), whereas post-reperfusion LDH levels were poorly correlated with cold ischemic time (p = 0.031), intraoperative fresh frozen plasma transfusions (p = 0.002), and intraoperative packed red blood cell transfusions (p = 0.018, Table 2). Donor age and warm ischemic times were not correlated with post-reperfusion AST, ALT, and LDH levels. The median follow-up period was 37.5 months. A total of 27 patients developed HCC recurrence with recurrence-free survival rates of 90.8%, 83.4%, and 81.0% at 1, 3, and 5 years, respectively. Univariable analyses revealed that post-reperfusion AST (p = 0.521), ALT (p = 0.773), and LDH (p = 0.575) levels and peak 7-day post-transplant bilirubin concentration (p = 0.592), INR (p = 0.553), and GGTP activity (p = 0.534) were not significantly associated with recurrence in all patients (Table 3). There were also no significant differences in   Tables 4 and 5. Additionally, fulfilment of the Milan (p < 0.001; HR 0.17, 95% CI 0.07-0.43); University of California, San Francisco (UCSF, p = 0.009; HR 0.37, 95% CI 0.17-0.78); and Up-to-7 (p < 0.001; HR 0.24, 95% CI 0.11-0.51) criteria significantly reduced the risk of recurrence. For further analyses, the patients were divided into subgroups based on the fulfilment of selection criteria and independent predictors of recurrence. Cut-offs for tumour number of ≥3 and alpha-fetoprotein concentration of ≥48.3 ng/ml were derived from the corresponding ROC curves. As a continuous variable, post-reperfusion AST significantly influenced the risk of HCC recurrence only in patients within the Milan criteria (p = 0.035, Table 6) with the optimal cut-off of ≥1896 U/L. Additionally, post-reperfusion AST and LDH levels exceeding the upper quartiles were significantly associated with increased risk of recurrence in patients either within  Table 3. Analyses of risk factors for hepatocellular carcinoma recurrence after deceased donor liver transplantation. Hazard ratios for continuous variables are given per: 1 log e (U/L) increase for AST, ALT, LDH, and GGTP activity; 1 increase for INR; 1 hour increase for total and cold ischemic times; 10 minute increase for warm ischemic time; 1 year increase for recipient and donor age; 1 increase for tumor number; 1 mm increase for diameter of the largest tumor; 10 cm 3 increase for total tumor volume; 1 log e (ng/ml) increase for alpha-fetoprotein concentration; 1 point increase for Model for End-stage Liver Disease; and 1 unit increase for transfusions. HR -hazard ratio; 95% CI -95% confidence interval; AST -aspartate transaminase; ALTalanine transaminase; GGTP -gamma-glutamyl transpeptidase; INR -international normalized ratio; LDHlactate dehydrogenase; PRBC -packed red blood cells; FFP -fresh frozen plasma the Milan (p = 0.039 and p = 0.040, respectively) or Up-to-7 (p = 0.028 and p = 0.039, respectively) criteria.
The degree of IRI, as reflected by post-reperfusion AST, ALT, and LDH levels, did not significantly influence the HCC recurrence risk in patients within the UCSF criteria or in those beyond the Milan, UCSF, or Up-to-7 criteria. No other significant associations between post-reperfusion AST, ALT, and LDH levels and the risk of post-transplant tumour recurrence were observed in subgroups derived from divisions based on tumour number, alpha-fetoprotein concentration, presence of microvascular invasion, and degree of tumour differentiation. In contrast to the significant effects of IRI in patients within the Milan or Up-to-7 criteria, no effects were found for the duration of total ischemia (all p > 0.701), cold ischemia (all p > 0.417), warm ischemia (all p > 0.373), and donor age (all p > 0.276) in these subgroups (Table 7). No significant associations between peak 7-day postoperative bilirubin concentration (all p > 0.081), INR (all p > 0.205), and GGTP activity (p > 0.097) and HCC recurrence risk were identified in subgroup analyses (Table 8).

Discussion
In the era of donor shortage and increasing utilization of high-risk grafts to partly ameliorate its negative effects, the problem of potential association between the degree of IRI and the risk of HCC recurrence after liver transplantation is of utmost importance. According to the available results of experimental studies, hepatic IRI, universally present in the setting of liver transplantation, increases the risk of metastasis formation both within the ischemic and remote sites through changes in the local microenvironment, induction of inflammatory response, induction of metastatic potential of circulating cancer cells, and systemic release of pro-tumourigenic   [12][13][14][15][16] . Our study results demonstrate a major negative effect of IRI on the risk of post-transplant HCC recurrence, although limited to patients with low tumour burden. Importantly, initial analyses performed in all patients failed to reveal any significant associations between post-reperfusion AST, ALT, and LDH levels and HCC recurrence risk, irrespective whether the factors were analysed as continuous or categorical variables. However, the study cohort comprised patients with a wide range of tumour burden due to a liberal selection policy utilised in the authors' department before establishment of   Table 6. Subgroup analyses of the associations between post-reperfusion aspartate transaminase, alanine transaminase, and lactate dehydrogenase activity and the risk of hepatocellular carcinoma recurrence after liver transplantation according to fulfillment of selection criteria and independent risk factors. Q4 -fourth quartile; Q1-Q3 -first to third quartile; AST -aspartate transaminase; ALT -alanine transaminase; LDH -lactate dehydrogenase; UCSF -University of California, San Francisco; AFP -alpha-fetoprotein; MVI -microvascular invasion.
precise criteria 5 . Nevertheless, a major significant negative effect of post-reperfusion AST and LDH levels was observed for patients within the Milan criteria, which still determine the majority of liver transplant recipients 28 . Similar findings, although of remarkably lesser extent, were found for patients within the Up-to-7 criteria, whereas the magnitude of IRI did not influence the risk of recurrence in patients beyond the extended criteria. This indicates that the clinical relevance of IRI is limited to generally low-risk populations and diminishes with increasing tumour burden. This appears to be particularly importantly because it demonstrates the possibility of using high-risk grafts to expand the donor pool for high-risk HCC candidates in the context of discussion on widening the boundaries of existing selection criteria [2][3][4][5][6]29 . Notably, the safe use of extended criteria allografts preferentially for patients with advanced tumours was already reported 30 . Conversely, none of the subgroup analyses performed in high-risk patients, including those beyond particular selection criteria, with ≥3 tumours, alpha-fetoprotein concentration ≥48.3 ng/mL, or with tumours either poorly differentiated or with microvascular invasion, revealed a significant effect of IRI on the risk of HCC recurrence. Therefore, while these findings point toward the possibility of the utilization of grafts more prone to IRI for high-risk HCC patients, they also indicate limited clinical relevance of reducing IRI in these patients. In contrast to the use of post-reperfusion transaminases and LDH levels as surrogates of IRI degree in the present study, previous studies focused on the negative effects of prolonged graft ischemia or donor characteristics [17][18][19][20][21][22][23][24][25]31 . However, the degree of IRI is driven by the interplay of several donor risk factors, of which a single component may not necessarily be an adequate measure of IRI 32 . In the present study, the laboratory measures of graft ischemia were significantly, yet poorly correlated to graft ischemic times, which in fact is consistent with the results presented by other authors 25 . This may partly explain the inconsistent results of studies on the effect of duration of graft ischemia and particular donor factors on HCC recurrence risk, as these may not always accurately reflect the magnitude of IRI [17][18][19][20][21][22][23][24][25]31 .
In contrast to the significant effects of IRI limited to low-risk patients found in the present study, two previous analyses specifically aimed at the effect of ischemic times on tumour recurrence revealed the presence of significant associations particularly in high-risk HCC patients 24,25 . These populations were characterised by 18 F-fluorodeoxyglucose tumour avidness on pre-transplant positron emission tomography and vascular invasion, both of which are known surrogates of biological aggressiveness. Although positron emission tomography data were not available, categorization of patients based on pre-transplant alpha-fetoprotein concentration and tumour differentiation, which are important markers of tumour biology, did not reveal any significant effects of IRI and neither did the analyses stratified for microvascular invasion. The reason for this discrepancy is unclear, although it may be related to a wider spectrum of tumour burden in patients included in the present study. Of note, post-operative peak transaminases did not emerge as risk factors for HCC recurrence in these previous reports. However, we chose post-reperfusion AST, ALT, and LDH levels routinely assessed in our department and not peak levels over the postoperative period in order to minimise the effect of events other than IRI on these parameters.
The results of the present study point toward the importance of strategies aimed to decrease IRI particularly for patients within the standard selection criteria. A single retrospective study revealed decreased magnitude of IRI, as illustrated by low transaminase levels and decreased risk of HCC recurrence in patients receiving prostaglandin E1 analog alprostadil in the early period after liver transplantation 33 . The protective effects of ischemic preconditioning with respect to the development of metastases were also reported in a recent experimental   14 . The use of machine perfusion devices has also been shown to decrease the magnitude of IRI and recently even enabled the development of a strategy to practically eliminate its negative consequences [34][35][36] . Although the present study does not provide any evidence for the effects of these measures in liver transplantation for HCC, it provides a rationale for prospective trials aimed at addressing this issue. This study had several limitations besides those inherent to its retrospective nature. Donor characteristics other than a baseline variable of age were neither analysed for associations with post-reperfusion transaminase and LDH levels nor as predictors of tumour recurrence. However, such analyses were beyond the scope of this study, specifically aimed at the effect of IRI on post-transplant HCC recurrence rather than on its determinants. Because all recipients received grafts from donors after brain death, this study did not directly address the issue  Table 8. Subgroup analyses of the associations between peak 7-day postoperative bilirubin concentration, INR value, and GGTP activity and the risk of hepatocellular carcinoma recurrence after liver transplantation according to fulfillment of selection criteria and independent risk factors. Hazard ratios for continuous variables are given per: 1 mg/dL increase for bilirubin concentration; 1 increase for INR; 1 log e (U/L) increase for GGTP activity. Q4 -fourth quartile; Q1-Q3 -first to third quartile; INR -international normalized ratio; GGTPgamma-glutamyl transpeptidase; AFP -alpha-fetoprotein; MVI -microvascular invasion.
SCIEnTIfIC RePoRts | (2018) 8:8935 | DOI:10.1038/s41598-018-27319-y of using grafts from donors after cardiac death for HCC patients, which was recently shown not to increase the risk of post-transplant recurrence 23 . Although subject to additional warm ischemia and thus potentially increased magnitude of IRI, their use in HCC patients may be confounded by other factors, including but not limited to, non-random allocation and differences in other donor characteristics. Furthermore, the duration of warm ischemia was not identified as a significant predictor of HCC recurrence. Finally, the main findings of our study are based on the results of univariable subgroup analyses. Therefore, the findings may be confounded by the effects of other risk factors for tumour recurrence. Although there was no particular policy at the authors' department for the allocation of high-risk grafts to higher-risk HCC patients, the results may also be confounded by non-random allocation of grafts more prone to IRI to patients within the Milan or Up-to-7 criteria, yet at higher initial recurrence risk.
In conclusion, the magnitude of IRI is strongly associated with the risk of tumour recurrence in patients within the Milan criteria and to a lesser extent, in patients within the extended criteria. Available measures to decrease IRI should be evaluated as a method to prevent HCC recurrence after liver transplantation, specifically in patients with low tumour burden.