Use of N-acetylcysteine during liver procurement: A prospective randomized controlled study


  • Francesco D'Amico contributed to the study concept and design, the acquisition and analysis of data, and the supervision of the study and manuscript. Alessandro Vitale contributed to the analysis and interpretation of data, the drafting of the manuscript, and the statistical analysis. Anna Chiara Frigo contributed to the statistical analysis. Umberto Cillo contributed to the analysis and interpretation of data and the supervision of the study. Alessandra Bertacco contributed to the acquisition and analysis of data and the critical revision of the manuscript. Donatella Piovan, Domenico Bassi, Rafael Ramirez Morales, Pasquale Bonsignore, Enrico Gringeri, Michele Valmasoni, Greta Garbo, Enrico Lodo, Francesco Enrico D'Amico, Michele Scopelliti, Amedeo Carraro, Martina Gambato, Alberto Brolese, Giacomo Zanus, and Daniele Neri contributed to the critical revision of the manuscript for important intellectual content and the acquisition of data.

  • Francesco D'Amico had full access to all the data and takes full responsibility for the veracity of the data and the statistical analysis.

  • The present study is registered with, identifier NCT01394497.

  • See Editorial on Page 118

Address reprint requests to Francesco D'Amico, M.D., Ph.D., F.E.B.S., Department of Hepatobiliary Surgery and Liver Transplantation, University Hospital of Padua, Via Giustiniani 2, 35128 Padua, Italy. Telephone: +39 049 821-8624; FAX: +39 049 821-1816; E-mail:


Antioxidant agents have the potential to reduce ischemia/reperfusion damage to organs for liver transplantation (LT). In this prospective, randomized study, we tested the impact of an infusion of N-acetylcysteine (NAC) during liver procurement on post-LT outcomes. Between December 2006 and July 2009, 140 grafts were transplanted into adult candidates with chronic liver disease who were listed for first LT, and according to a sequential, closed-envelope, single-blinded procedure, these patients were randomly assigned in a 1/1 ratio to an NAC protocol (69 patients) or to the standard protocol without NAC [71 patients (the control group)]. The NAC protocol included a systemic NAC infusion (30 mg/kg) 1 hour before the beginning of liver procurement and a locoregional NAC infusion (300 mg through the portal vein) just before cross-clamping. The primary endpoint was graft survival. The graft survival rates at 3 and 12 months were 93% and 90%, respectively, in the NAC group and 82% and 70%, respectively, in the control group (P = 0.02). An adjusted Cox analysis showed a significant NAC effect on graft survival at both 3 months [hazard ratio = 1.65, 95% confidence interval (CI) = 1.01-2.93, P = 0.04] and 12 months (hazard ratio = 1.73, 95% CI = 1.14-2.76, P ≤ 0.01). The incidence of postoperative complications was lower in the NAC group (23%) versus the control group (51%, P < 0.01). In the subgroup of 61 patients (44%) receiving suboptimal grafts (donor risk index > 1.8), the incidence of primary dysfunction of the liver was lower (P = 0.09) for the NAC group (15%) versus the control group (32%). In conclusion, the NAC harvesting protocol significantly improves graft survival. The effect of NAC on early graft function and survival seems higher when suboptimal grafts are used. Liver Transpl 19:135–144, 2013. © 2012 AASLD.


alanine aminotransferase


aspartate aminotransferase


confidence interval


donor risk index




intensive care unit


international normalized ratio


ischemia/reperfusion injury


liver transplantation


Model for End-Stage Liver Disease




primary dysfunction


primary nonfunction


postoperative day

Liver transplantation (LT) is the treatment of choice for end-stage liver disease. Ischemia/reperfusion injury (IRI) during the conventional cold storage and preservation of donated livers is a key determinant of early graft function after LT.[1]

In recent years, suboptimal livers have been increasingly used because of organ shortages.[2] Unfortunately, the use of marginal livers may have negative effects on initial graft function,[3] with prolonged IRI being the main source of graft dysfunction.[1]

The pathophysiology of IRI is mainly due to the oxidative stress and cytotoxicity arising from the imbalance between oxidants and antioxidants.[4] Glutathione (GSH) is the main endogenous antioxidant, and the liver is the main organ producing GSH under normal conditions (normothermia and normoperfusion/oxygenation). Because of the donor's stay in the intensive care unit (ICU) and the period of ischemia, potential intrahepatic GSH depletion can lead to the liver being less able to overcome reperfusion oxidative stress. Hepatic GSH synthesis is a cysteine/methionine-dependent process strictly linked to the dietary introduction of its precursors. N-Acetylcysteine (NAC) is the acetylated precursor of both the amino acid L-cysteine and reduced GSH. Experimental animal models have shown a potential positive impact of NAC or GSH infusions in decreasing IRI during cold liver preservation.[4, 5]

NAC is commonly used in clinical practice as a mucolytic agent for chronic respiratory illness, as an antidote for hepatotoxicity due to acetaminophen overdose, and as prophylaxis against renal injury due to radiocontrast agents.[6] NAC has a very good safety profile, and there is no documented evidence of dose-dependent side effects.

The aim of this prospective, randomized study was to test the impact of systemic and locoregional infusions of NAC during liver procurement on post-LT outcomes.


Trial Design and Participants

This was a prospective, single-blinded, randomized phase 2 study of 2 parallel groups designed to evaluate the clinical benefits of NAC infusions during liver procurement versus a standard procedure.

All subjects who met the inclusion criteria were randomized in a 1:1 ratio to receive an organ perfused or not perfused with NAC during the harvesting procedure.

This study included all consecutive subjects with chronic liver disease who underwent LT for the first time with deceased donor livers at our institution between December 2006 and July 2009. The following subjects were excluded: (1) recipients with acute liver disease, (2) pediatric patients or adult patients receiving livers from pediatric donors, (3) patients undergoing multiorgan transplantation, (4) patients undergoing re-LT, and (5) patients undergoing living donor LT.

Data were collected during the organ procurement phase, surgical interventions, and follow-up medical visits performed at our facilities.

This study was reviewed and approved by our institutional ethics committee, which considered consent and authorization for research necessary only for the recipients.


In this study, the selection of patients for LT and the assignment of donor organs strictly adhered to our clinical practice and were thus based on a transparent and well-defined allocation protocol using the Model for End-Stage Liver Disease (MELD) score to stage the severity of liver cirrhosis.[7] This allocation scheme was established at our center 6 months before the initiation of this study.

All the recipients were at our institution (ie, this was a single-center study). Organ procurement, on the contrary, was performed throughout the Italian region of the center and sometimes also extraregionally. However, the surgical recovery team always belonged to our institution.

In Italy, organ procurement is performed exclusively with heart-beating donors.

Consecutive deceased donor livers with eligible potential recipients were randomly assigned to either the standard procurement procedure or the NAC protocol.

In addition to the standard procedure, the NAC protocol included a 15-minute systemic NAC infusion (30 mg/kg for a maximum dose of 3000 mg) diluted into a 5% glucose solution (500 mL) 1 hour before the initiation of liver harvesting and a locoregional infusion (150 mg/kg of estimated liver weight for a maximum dose 300 mg) into the portal vein just before cross-clamping. The protocol infusion doses were established according to the following criteria: (1) the systemic NAC dose (30 mg/kg) corresponded to the maintenance dose used for the treatment of acute hepatic failure, and (2) the locoregional NAC dose (150 mg/kg of liver weight) corresponded to the loading dose used for the treatment of acute liver failure.

As previously described,[8] the standard procurement procedure at our center was based on a modified double perfusion technique (aortic and portal cooling with tourniquet clamping of the splenomesenteric vein inflow). Single aortic perfusion was used only with split or multivisceral harvesting procedures. Donor livers were gravity-perfused in situ via the aorta and portal vein with Celsior solution at 4°C. Approximately 60% of the solution volume (30 mL/kg via the portal vein and 60 mL/kg via the aorta in the double perfusion procedure and 90 mL/kg via the aorta with the single aortic perfusion technique) was infused rapidly (10-15 minutes) after aortic cross-clamping. Perfusion was then slowed for the remaining 40% of the solution until the harvesting was completed (20-40 minutes). After hepatectomy, donor livers were further perfused at the back table with Celsior (700 mL via the portal vein and 300 mL via the hepatic artery) and then stored in conventional bags containing the same solution at 4°C until transplantation.

LT was always performed with the preservation of the retrohepatic vena cava (the piggyback technique) and without the use of a biopump. Transplant patients were given dual- or triple-drug immunosuppressive therapy, which included cyclosporine or tacrolimus combined with corticosteroids with or without mycophenolate mofetil. Liver biopsy was performed after transplantation only when it was clinically indicated.

Outcomes and Definitions

The primary endpoint of this study was graft survival.

The secondary endpoints were as follows: (1) patient survival; (2) the incidence of primary dysfunction (PDF); (3) differences in transaminase, international normalized ratio (INR), and bilirubin values on postoperative days (PODs) 2, 7, and 15; (4) lengths of hospital stays, blood-derived transfusions, and days of therapy with vasoactive amines; (5) postoperative complications (surgical and medical complications, including infections); and (6) acute rejection. We considered a complication to be any clinical adverse event graded as 3 or higher according to Common Terminology Criteria for Adverse Events version 4.0.

The quality of donor livers was described with the donor risk index (DRI).[3]

Graft PDF was defined according to Pokorny et al.[9] as the sum of primary nonfunction (PNF) and initial poor function. PNF was defined as non–life-sustaining function of the graft leading to death or retransplantation within 7 days. Initial poor function was defined as an aspartate aminotransferase (AST) level > 2500 IU/L and clotting factor support for more than 2 days during the first 5 days after the operation.

In this study, we sought maximal adherence to our normal clinical practice. Consequently, we did not perform protocol-specific biopsy at the time of LT or at 3 months, but we performed liver biopsy only when it was clinically indicated (to discriminate between acute rejection, hepatitis, and other kinds of liver damage).

Sample Size

The sample size was calculated on the basis of the following assumptions:

  • A type I error rate of 5%.
  • A power of 90%.
  • A 1:1 distribution ratio of enrolled patients in the 2 groups.
  • A 1-sided test.
  • An accrual period of 30 months.
  • A minimum follow-up period of 12 months for living enrolled patients.

To reach these goals, we decided to recruit 176 patients (88 per group).

An interim analysis was planned for 15 months after the initiation of enrollment.[10] No a priori criteria were set for an early end of the study.

Because of the monocentric and spontaneous nature of this study and the well-known positive safety profile of NAC, the inclusion of an external data safety monitoring board in the study design was not deemed necessary or required.


between December 2006 and July 2009, all potential grafts for recipients considered eligible for this study were randomly assigned to 1 of 2 arms: (1) deceased donors received systemic and locoregional infusions of NAC, or (2) the standard harvesting procedure (without NAC infusions) was performed. The designation of organs to either group was performed according to computer-generated block randomization (n = 10 per block). Random ranking per block was determined with the RAND function of Microsoft Excel, and it was used in the sequential attribution of treatments. Randomization was performed by the principal investigator immediately after he received the notification call from our organ-sharing organization for a potential liver from a deceased donor (this included organs that were deemed unsuitable for transplantation after an accurate evaluation), and this was communicated to the surgical team at the moment of organ retrieval. Patients were blinded to the type of preservation treatment that the organ had received, and at no point during the study were they unblinded.

Between December 2006 and July 2009, 214 potential grafts were included in the procedure (107 in the study group and 107 in the control group). In 74 cases (38 in the study group and 36 in the control group), the livers were subsequently considered unsuitable for transplantation because of high-grade steatosis (steatosis > 50% after biopsy; 10 cases), liver cirrhosis (13 cases), cancer in the donor (9 cases), liver fibrosis (13 cases), sepsis in the donor (4 cases), hepatitis C virus positivity (4 cases), hepatic trauma (2 cases), liver ischemic injury (6 cases), and other causes (13 cases).

The remaining 140 livers were transplanted into the recipients enrolled in this study.

Statistical Analysis

All demographic and baseline variables were described as follows: categorical data were described as frequencies and percentages, and continuous data were described as medians and ranges. Missing data were excluded from the final analysis.

For subgroup comparisons, quantitative variables were compared with the Student t test or the Wilcoxon rank-sum test, and categorical variables were compared with the χ2 test or Fisher's exact test as appropriate.

Enrolled patients were followed until April 2011 when the final data analysis was performed.

The follow-up and survival periods were expressed as medians and ranges. Survival curves were calculated according to the Kaplan-Meier method and were compared with a log-rank analysis. The Cox proportional hazards model was used to calculate univariate and multivariate hazard ratios and 95% confidence intervals (CIs). In the multivariate model, the effect of NAC on graft and patient survival at 3 and 12 months was adjusted for the following covariates: recipient age, recipient MELD score at LT, period of LT, and DRI. The study was dichotomized into nonoverlapping 15-month periods.

We used targeted subgroup analyses to obtain a more accurate interpretation of the results. For quantitative variables, the cutoff was the median value.

Statistical significance was set at P < 0.05. Calculations were performed with the JMP package (SAS Institute, Inc., 1989-2003).


Participant Flow, Recruitment, and Baseline Data

The enrollment period started in December 2006 and ended in July 2009 when the established enrollment period of 30 months was reached.

As shown in Fig. 1, 184 consecutive patients were assessed for eligibility; 44 were excluded according to the inclusion/exclusion criteria. One hundred forty patients were enrolled and randomized: 69 patients in the study group and 71 patients in the control group. All enrolled patients attained the primary endpoint (there were no losses or exclusions after randomization).

Figure 1.

Consolidated Standards of Reporting Trials (CONSORT) 2010 flow diagram.

A bar graph of patient accrual by 6-month intervals is presented in Fig. 2. The established sample size (88 patients per arm) was not met because of slower than expected accrual. Moreover, the encouraging results of the interim analysis[10] prompted us to not prolong the established enrollment period, although no a priori criteria for ending the study early were originally planned.

Figure 2.

Bar graph of patient accrual by 6-month intervals.

Tables 1 and 2 describe the baseline characteristics of the donors and recipients. As expected, there were no significant differences between the groups.

Table 1. Donor Characteristics of the NAC and No-NAC Groups
VariableNo NAC (n = 71)NAC (n = 69)P Value
  1. aThe data are presented as medians and ranges.
Age (years)a56 (18-86)54 (19-79)0.71
Partial grafts [n (%)]11 (15)5 (7)0.18
Body mass index (kg/m2)a24.7 (15.6-33.2)24.7 (18.4-40.0)0.64
Hemodynamic instability [n (%)]31 (44)24 (35)0.37
Ultrasound steatosis [n (%)]19 (27)21 (30)0.42
Sodium (mmol/L)a149 (122-167)147 (134-188)0.98
AST (U/L)a36 (7-1700)36 (4-293)0.64
ALT (U/L)a23 (4-933)31 (4-227)0.92
Bilirubin (mg/dL)0.8 (0.2-3.9)0.6 (0.2-7.2)0.08
Time from start of surgery (NAC infusion) to cross-clamping (minutes)a180 (100-385)195 (70-350)0.21
DRIa1.8 (0.9-3.0)1.7 (0.9-2.4)0.15
Table 2. Recipient Characteristics and Technical Aspects of the NAC and No-NAC Groups
VariableNo NAC (n = 71)NAC (n = 69)P Value
  1. aThe data are presented as medians and ranges.
Age (years)a58 (30-68)55 (27-68)0.17
Sex: female [n (%)]20 (28)14 (20)0.33
Main indication for LT [n (%)]  0.87
Hepatitis C virus13 (18)19 (28) 
Hepatitis B virus7 (10)3 (4) 
Alcohol4 (6)6 (9) 
Hepatocellular carcinoma36 (51)34 (49) 
Other11 (15)7 (10) 
Child-Pugh-Turcotte scorea8 (6-14)9 (6-14)0.32
MELD scorea17 (7-35)16 (7-37)0.85
Cold ischemia time (minutes)a500 (253-620)489 (140-605)0.12
Warm ischemia times (minutes)a30 (15-70)31 (15-85)0.32
Total LT time (minutes)a460 (310-855)420 (300-930)0.21
Cell saver (mL)a1000 (0-9000)755 (224-6875)0.24
Rapid infusion system (mL)a9669 (850-63,033)10,132 (1200-31,115)0.31
Red blood cells (mL)a1500 (0-7800)1350 (0-5700)0.91
Plasma (mL)a1600 (0-5000)2000 (0-12,000)0.13

The characteristics of the liver grafts showed a relevant prevalence of donor factors with a potentially negative impact on post-LT outcomes (Table 1). In particular, we observed a high median DRI value in both groups. A DRI value of 1.8 (corresponding to the median value in the control group) was used for subgroup analyses.

As for recipient characteristics, the prevalence of patients with hepatocellular carcinoma in both groups has to be underlined (Table 2).

Primary Endpoint

The median follow-up was 33.4 months (range = 0.1-52.8 months).

We recorded 38 graft failures (27%): 13 (34%) in the study group and 25 (66%) in the control group (P = 0.03). Six cases of graft failure did not result in patient death because of successful liver retransplantation.

Nineteen graft failures (14%) occurred within the first 3 months after LT; the main causes were PDF (8 cases or 42%), sepsis complicating initial PDF (5 cases or 26%), sepsis with regular initial liver function (3 cases or 16%), heart failure (2 cases or 11%), and portal/arterial thromboses (1 case or 5%).

A substantial proportion of graft failures (6.4%) also occurred 3 to 12 months after LT. Seven of these failures (78%) occurred in the control group: 3 due to hepatitis C virus recidivism, 2 due to delayed graft function/sepsis, 1 due to a de novo tumor, and 1 due to intra-abdominal bleeding after liver biopsy. We had only 2 failures in the same period in the study group (P < 0.05): 1 due to hepatitis C virus recidivism and 1 due to hepatocellular carcinoma recurrence.

The graft survival rates at 3, 12, and 24 months were 93%, 90%, and 86%, respectively, in the study group and 82%, 70%, and 67%, respectively, in the control group (P = 0.02; Fig. 3). The difference in graft survival at 3 months was 11% (95% CI = 2%-20%, P = 0.04), and although it was significant, the difference was lower than that assumed in the sample size determination. The calculated difference in graft survival at 12 months was 20% (95% CI = 9%-31%, P < 0.01), and the sample size determined the power of Fisher's exact test in estimating this difference to be 91%.

Figure 3.

Kaplan-Meier curves showing graft survival in the 2 groups. The log-rank test showed a statistically significant difference between the 2 curves (P = 0.0223).

To better understand the effects of NAC on graft survival at different time points, we used 2 separate Cox regressions at 3 and 12 months and considered both unadjusted and adjusted models (Table 3). This analysis confirmed an increasing impact of NAC on graft survival from 3 to 12 months that was not influenced by other recipient or donor variables.

Table 3. Cox Proportional Hazards Models Showing the Unadjusted and Adjusted Impacts of NAC on Graft and Patient Survival
Survival EndpointHazard Ratio (95% CI)/P Value
3 Months12 Months
  1. NOTE: In the multivariate model, the effects of NAC on graft and patient survival at 3 and 12 months were adjusted for the following covariates: recipient age, recipient MELD score at LT, period of LT, and DRI.
Graft survival  
Adjusted0.33 (0.10-0.91)/0.040.30 (0.12-0.70)/<0.01
Unadjusted0.35 (0.11-0.91)/0.030.31 (0.12-0.70)/<0.01
Patient survival  
Adjusted0.09 (0.00-0.46)/<0.010.18 (0.05-0.51)/<0.01
Unadjusted0.09 (0.00-0.45)/<0.010.20 (0.06-0.52)/<0.01

Secondary Endpoints

During the study period, we registered 32 patient deaths: 10 (31%) in the study group and 22 (69%) in the control group (P = 0.02). The patient survival rates at 3, 12, and 24 months were 99%, 94%, and 90%, respectively, in the study group and 86%, 75%, and 72%, respectively, in the control group (P = 0.01; Fig. 4).

Figure 4.

Kaplan-Meier curves showing patient survival in the 2 groups. The log-rank test showed a statistically significant difference between the 2 curves (P = 0.01).

Similarly to the graft survival analysis, the effect of NAC on patient survival was independent of other recipient or donor covariates (Table 3).

We recorded 14 PDF events in the study group (20%) and 16 PDF events (23%) in the control group. The difference did not reach statistical significance (Table 4).

Table 4. Secondary Endpoints
VariableNo NAC (n = 71)NAC (n = 69)P Value
  1. The data are presented as medians and ranges.
PDF [n (%)]16 (23)14 (20)0.84
PNF [n (%)]4 (6)3 (4)0.45
POD 2 AST (U/L)a705 (47-11,933)571 (65-10,160)0.85
POD 2 ALT (U/L)a670 (34-8632)538 (35-9600)0.89
POD 2 bilirubin (mg/dL)a2.9 (0.6-46.7)2.6 (0.4-19.8)0.49
POD 2 INRa1.5 (1.0-6.4)1.4 (1.1-3.6)0.52
POD 7 AST (U/L)a48 (12-2987)48 (17-1087)0.36
POD 7 ALT (U/L)a101 (9-1580)105 (21-1160)0.70
POD 7 bilirubin (mg/dL)a5.0 (0.7-37.4)5.1 (0.7-20.1)0.33
POD 7 INRa1.2 (1.0-2.1)1.2 (1.0-1.8)0.80
POD 15 AST (U/L)a38 (13-326)31 (15-150)0.02
POD 15 ALT (U/L)a68 (10-938)62 (12-720)0.45
POD 15 bilirubin (mg/dL)a2.7 (0.9-26.2)2.4 (0.9-23.0)0.78
POD 15 INRa1.1 (1.0-1.8)1.1 (0.9-1.9)0.71
Hospital stay (days)a19 (8-155)17 (7-95)0.07
Plasma in ICU (mL)a1650 (0-15,000)2250 (0-16,100)0.40
Vasoactive amines (days in ICU)a1 (0-55)0 (0-18)0.08
Postoperative complications [n (%)]36 (51)16 (23)<0.01
Acute rejection [n (%)]2 (3)14 (20)<0.01

Among the other secondary endpoints, we found the following significant or marginally significant differences (Table 4): (1) the POD 15 AST levels were 31 U/L (range = 15-150 U/L) in the study group and 38 U/L (range = 13-326 U/L) in the control group (P = 0.02), (2) the median hospital stay was 17 days (range = 7-95 days) in the study group and 19 days (range = 8-155 days) in the control group (P = 0.07), (3) the number of days of therapy with vasoactive amines during the ICU stay was lower in the study group [0 days (range = 0-18 days)] versus the control group [1 day (range = 0-55 days), P = 0.08], (4) the overall incidence of postoperative complications was 23% in the study group and 51% in the control group (P < 0.01), and (5) the incidence of acute rejection episodes was significantly higher in the study group [20% (14 patients)] versus the control group [3% (2 patients), P < 0.01]. In all, the diagnosis of acute rejection was histologically proven in 16 cases. Biopsy was performed in 8 other cases (all in the control group). The diagnoses were sepsis (6 cases) and ischemia/reperfusion damage (2 cases).

The numbers of septic events in the 2 groups were not significantly different (and thus did not explain the difference in the complication rates).

Subgroup Analysis

We evaluated the impact of suboptimal liver grafts (DRI > 1.8) on the incidence of PDF and early graft survival. Sixty-one patients (44%) received a suboptimal liver: 27 patients (39%) in the study group and 34 patients (48%) in the control group (P > 0.05).

The incidence of PDF events was clearly lower in the study group (15%) versus the control group (32%); the difference reached marginal statistical significance (P = 0.09).

Conversely, the incidence of PDF events was similar in the 2 groups when only optimal grafts were considered (DRI ≤ 1.8).

When the subgroup analysis was applied to the early graft survival endpoint, we found that the differences between the 2 groups were concentrated exclusively in patients receiving suboptimal grafts. In particular, we had 7 early graft failures (21%) in the control group and only 1 early graft failure (4%) in the study group when donors with DRIs > 1.8 were used (P = 0.05). When the specific causes of failure were investigated, we found that the difference was mainly due to the 5 cases of early graft failure due to sepsis following PDF (all in the control group). We found that the main factor influencing septic graft failure after LT was a previous PDF event.


To the best of our knowledge, no randomized clinical trials have evaluated the use of antioxidant agents infused during liver procurement and their impact on post-LT outcomes.

The main result of this study is that our NAC procurement protocol was able to significantly improve graft survival after LT with respect to the conventional technique (Fig. 2).

This result was confirmed both in the very early post-LT phase (3-month graft survival) and in the intermediate follow-up period (12-month graft survival). However, when this was analyzed in detail, the main difference between the 2 groups in specific causes of graft failure was related to the occurrence of sepsis complicating a PDF event. In particular, 5 cases of graft failure were clustered in the subgroup of patients receiving suboptimal grafts (DRI > 1.8). The subgroup analysis of the 61 patients receiving suboptimal livers also revealed that there was a significant upward trend for the incidence of PDF events in the control group versus the study group. On a purely speculative basis, one could interpret the observed positive effect of NAC on graft survival as the result of a protective effect of this antioxidant agent against IRI, especially when suboptimal livers are used. This effect may also be a likely explanation for other significant differences between the 2 groups with respect to secondary endpoints such as patient survival (Fig. 3), lengths of hospital stays, and postoperative complications.

Three to 12 months after LT, the control group had a higher incidence of graft loss due to delayed graft function or early and aggressive hepatitis C virus recidivism, and it is well known that both pathophysiological events may be influenced by the extent of IRI.[1, 2, 11] These results suggest a potential impact of the NAC protocol in the intermediate term as well.

When we analyzed the prognostic impact of all variables included in the DRI formula, only donor age had a significant impact (data not shown). When we separately considered the effect of donor age on the study group versus the control group, this variable maintained a significant impact on graft survival only in the latter group. This finding supports a potential protective effect of NAC on post-LT outcomes, particularly when older donors are used.

Although liver biopsy is important for directly estimating ischemia/reperfusion damage at the moment of LT, GSH reserves, and the quality of graft recovery 3 months after LT, we deliberately chose not to perform protocol-specific biopsy because we felt that adherence to our daily clinical practice was essential for limiting any potential bias in the selection and care of patients.

Moreover, it is possible for data on acute rejection to be biased by the biopsy rate. However, in our clinical practice, we have adopted an extremely aggressive approach to biopsy, which is immediately performed whenever liver functional tests show any alterations. This mitigates the potential bias related to the absence of protocol-specific biopsy samples.

It is very difficult to understand the higher incidence of acute rejection in the study group. Recent reports[11] suggest a potential interaction between IRI and the recipient's immune system, which could be the interpretative key to understanding the link between NAC and acute rejection.

In terms of donor and recipient characteristics (Tables 1 and 2), our population was similar to one reported in a recent Italian multicenter prospective study.[12] In particular, we had a donor population characterized by a higher median DRI versus a US population.[13, 14] It must be stressed that the median DRI of our 2 groups was also higher than the median DRI of the Italian Liver Match study cohort.[12] The poor results for the controls could be due to this high proportion of extended criteria donors (Table 1). The graft survival rates 3 and 12 months after LT were in fact 82% and 70%, respectively, in the control group. The survival figures were very similar to those described in a US population with a DRI of approximately 1.8.[3]

As for recipient characteristics, we compared the results of our study to those for a historical cohort from our center (January 2003 to December 2006). We selected an additional group of 223 consecutive patients who fulfilled the study's inclusion criteria but were not present in the original study. In 2006, our center introduced a MELD-based organ allocation system.[7] The overall graft survival rates for this retrospective cohort of patients were 88% and 85% at 3 and 12 months, respectively. These rates were intermediate between those of the study and control groups enrolled in our randomized clinical trial. The only relevant difference between these 2 populations was the median MELD score at LT, which changed significantly in non–hepatocellular carcinoma patients from 15 (range = 6-41) in the retrospective cohort to 22 (range = 9-37) in this study (data not shown). This trend is evident for all Italian centers with respect to historical cohorts.[12] It must be emphasized that the MELD score variable had a significant prognostic role in our Cox multivariate analyses at both 3 and 12 months.

MELD-based allocation alone may not explain our poor control results. However, clinical evidence from our center favors the hypothesis that the particular interaction between high MELD values and high-DRI grafts leads to a deterioration in post-LT survival. This phenomenon seems to have been confirmed by several Italian[12] and European centers.[15]

Taken together, these considerations of donor and recipient factors strengthen our conclusions regarding the excellent results obtained for the study group and thus the positive effects of the NAC protocol.

Although we did not reach the predetermined sample size and survival difference at the 3-month graft survival endpoint, the graft survival gain of 20% at 12 months was a very strong result, and Fisher's exact test showed statistical significance with a power of 91%. The final sample size was sufficient to determine a significant positive impact of the NAC protocol on 1-year graft survival.

Previous studies have thus far failed to show a relevant clinical impact of NAC on post-LT outcomes.[16, 17] The originality of our protocol was the systemic NAC infusion 30 minutes before the initiation of the harvesting procedure. This methodological aspect was observed in only 1 previous study.[18] Incidentally, that particular work is to date the only other study showing a positive impact of NAC on post-LT outcomes. All other studies known to us have administered NAC either in later phases of harvesting or during the recipient's operation. We think that an early systemic infusion of NAC is essential for the sought-after effect because it gives time for the donor's liver to replenish its metabolic reserve of GSH.[6] This may be particularly important when we are using extended criteria grafts, whose GSH reserves may be depleted.

NAC has a very good safety profile, and there is no documented evidence of dose-dependent side effects.[6] In our study, we did not record any drug-related adverse events. Moreover, a preliminary investigation of the outcomes of other organs harvested from donors into which NAC was infused did not show any significant adverse events that could be attributed to the use of NAC (data not shown). We are now performing further data analyses to determine whether the NAC protocol has beneficial effects on organs other than the liver.

The good safety performance of the NAC protocol, the very low costs of this antioxidant agent, and its strong effect on graft survival make the adoption of the NAC protocol highly cost-effective. For this reason, an ancillary study is now beginning at our institution with the aim of producing a detailed cost-effectiveness analysis of the NAC protocol.

Several limitations of this study must be noted. It was a monocentric experience; thus, our results should be considered with caution until larger, better designed, and externally validated studies become available. Although randomization was used, some trends toward differences in the characteristics of donors of the control and study groups could be observed (partial graft use, bilirubin levels, and DRI values). We cannot exclude the idea that these trends had an impact on the study results. This was not a double-blinded study with a risk of bias in either the selection of patients or their care. In this study, the selection of patients for LT was based on a transparent and well-defined allocation protocol essentially based on the MELD score.[7] In addition to allocation, with the randomization of the harvesting procedure and the study inclusion criteria, we have sought to limit the impact of any potential selection bias.

As for the care of the patients, although caregivers were not explicitly blinded for the purposes of this study, patients in both arms were treated according to the center's standard of care. The extremely standardized nature of the harvesting and transplantation procedures[8] and the specific lifesaving feature of LT with its ethical and clinical implications are such that the non–double-blinded design had limited influence on the physicians' care of graft recipients. In our assessment, unblinded caregivers had marginal or negligible effects on the results of this study.

The main problem of this study is that the significant difference in terms of graft survival (the primary endpoint) is not well justified by significant differences between conventional post-LT variables such as transaminase and lactate levels and PDF (Table 4). This fact may be due to at least 2 causes. First, it is well known that conventional variables are not accurate predictors of early graft function and IRI. Thus, the absence of a correlation between NAC administration and these conventional variables does not exclude a correlation between NAC and IRI. Second, the main methodological problem of this study is that we did not perform protocol biopsy to study IRI on a histological and molecular basis. A prospective study focused on this particular matter is currently under elaboration by our institution.

Finally, the impact of the NAC protocol on mid- to long-term outcomes (eg, hepatitis C virus recurrence) should be analyzed in greater detail.[14]

The question of the potential impact of this study on everyday clinical practice remains. Our study suggests that NAC could be an effective tool for reducing the risk of PDF and graft failure after LT, especially when suboptimal livers are used. The NAC procurement protocol could be selectively used when a potentially marginal donor organ is to be allocated.

NAC could be a complementary tool used in addition to other approaches for optimizing the post-LT outcomes of candidates receiving suboptimal grafts in a multimodal strategy of sorts. Our group has previously shown the importance of the donor harvesting technique.[7] The adoption of low cold ischemia times and routine biopsying are also advisable, especially in those cases involving potentially at-risk donors.[19] The use of machine perfusion methods for graft preservation during the procurement procedure may also improve the post-LT outcomes of patients receiving suboptimal livers.[20] Finally, an optimal donor-recipient match represents a crucial issue for overcoming the risk of post-LT graft failure.[21]

In conclusion, we propose that organ harvesting with the NAC protocol is one of the possible tools for improving LT results. This harvesting procedure should be used in the context of a multimodal strategy aimed at improving the outcomes of LT, increasing the use of suboptimal livers, and extending and standardizing the use of such relevant resources.


The authors thank Charles Miller, Myron Schwartz, Sukru Emre, and Gabriel Gondolesi for the interest and support shown to Francesco D'Amico during his stay as a research fellow at Mount Sinai Hospital of New York and to all the 2000 and 2004-2005 transplant fellows.

This article is dedicated to the memory of Augusto D'Amico.