Introduction

Clinically significant post-operative complications occur in approximately 10 % of bariatric surgical patients, increasing hospital stay [1]. Up to 18 % of patients with pre-existing liver disease have progressive hepatic dysfunction following bariatric surgery, with severity ranging from transient elevation of liver transaminases [2] to organ dysfunction [3] and rarely liver failure causing death [4]. Although bariatric surgery leads to long-term reduction in liver steatosis in a majority of patients, inflammatory changes and fibrosis scores either increase or are unchanged in at least 35 % of patients [5]. Progression of fatty liver disease to steatohepatitis and fibrosis occurs due to oxidative stress and chronic inflammation [6]. Steatotic hepatocytes are more vulnerable to toxic insults and hypoxia, being less able to tolerate the accompanying mitochondrial dysfunction and buffer the increased oxidative stress. Thus, potential antioxidant and anti- inflammatory effects of NAC are an attractive treatment for patients with NAFLD [7].

During laparoscopic bariatric surgery, the left lobe of the liver is retracted to expose the stomach, causing ischaemia. Release of the retractor at the end of surgery can cause ischaemia-reperfusion injury. Following gastric bypass, a 6-fold rise in aspartate transaminase and alanine transaminase from baseline levels, causing a liver injury that may contribute to post-operative inflammatory response [2, 8]. Although usually self-limiting, a larger post-operative inflammatory response may predispose the patient to a worse response to further insults, such as anastomotic leakage or post-operative infective—a so-called “second hit” [9].

Non-alcoholic fatty liver disease (NAFLD) has a number of structural and metabolic differences to non-steatotic liver [10] and are more likely to die, more readily through necrosis than apoptosis [11]. As a consequence, tolerance to ischaemia during bariatric surgery is reduced. The common inflammatory processes implicated in the pathogenesis of insulin resistance, obesity and fatty liver disease are also of great relevance to the pathological effects of ischaemia-reperfusion injury (IRI), especially leptin, adiponectin and resistin [12, 13].

A number of animal models of IRI have shown beneficial effects of NAC, including increased bile flow, improved sinusoidal blood flow, decreased post-operative liver transaminases and reduced evidence of injury on liver histopathology [14] In nine clinical trials in human orthotopic liver transplantation, NAC was administered either to donors before organ retrieval or to recipients before graft revascularisation. Three showed a significant reduction in post-operative liver enzymes, although the clinical significance in terms of complications and graft survival could not be evaluated in these small studies [15]. In an RCT of NAC in liver resection, Robinson et al. found no differences in clinical outcomes between groups; though on day 3, alanine transaminase (ALT) levels were lower in the treatment group [16].

The aim of this study was to assess the effect of bariatric surgery upon hepatocellular injury and the impact of the administration of NAC upon this. Amelioration of the unavoidable post-operative inflammatory response may reduce the subsequent impact of other complications. The study hypothesis was that NAC infusion would reduce hepatocellular injury and associated inflammatory response.

Methods

Trial Design and Recruitment

This study was approved by the Local Research Ethics Committee and the Medicines and Healthcare-related products Regulatory Agency of the United Kingdom (UK). Each patient gave fully informed, written consent prior to enrolment into this study. All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards.

Patients were recruited from the obesity clinic at Kings College Hospital and randomised on the day of surgery into a treatment group (NAC group) or control group, receiving standard care.

Inclusion criteria were patients aged 18 to 75 years inclusive, meeting the published National Institute of Clinical Excellence indications for weight loss surgery, that is they must have BMI >40 or >35 kg/m2 with obesity-related complications and are undergoing laparoscopic sleeve gastrectomy [17]. Patients were excluded if they had a history of chronic liver disease, abnormal hepatitis serology or autoantibody screen, previous liver surgery, a history of active psychiatric illness, a known bleeding tendency or prescribed anticoagulant medications or a known allergy to NAC or related compounds.

Subjects randomised to the treatment group received N-acetyl cysteine infusion (Aurum Pharmaceuticals Ltd, Romford, UK) at a standard 150 mg/kg in 200 ml 5 % dextrose over 15 min at induction of anaesthesia followed by an infusion of 50 mg/kg in 500 ml of 5 % dextrose during surgical retraction of the liver for a maximum of 4 h, together with standard anaesthetic medications. The maximum dose was limited to 16.5 g for loading dose and 5.5 g for infusion, making a maximum total dose of 22 g per study participant. The control subjects received standard anaesthetic medications alone.

Treatment of Patients

Patients were treated as per routine practice. The LSG was performed by a single experienced surgeon (AP) for all patients. Six trocars were utilised and LSG was performed with standard techniques using a 38-French bougie. A 12 mmHg carbon dioxide pneumoperitoneum was created. A liver biopsy from the left lateral section of the liver was taken using a spring-loaded core biopsy cannula. A fixed Nathanson liver retractor was placed under the left lateral section of the liver to facilitate clear visualisation of the hiatus. After devascularisation of the greater curvature using ultrasonic shears, the LSG commenced 4–6 cm proximal to the pylorus on the greater curvature and continued towards the angle of His. The completed staple line was reinforced with absorbable sutures. A further liver biopsy from the left lateral section was taken before release of pneumoperitoneum. Patients received low molecular weight heparin and wore compression stockings for thromboembolism prophylaxis.

Post-operative diet consisted of fluids only for 4 weeks, puréed consistency foods for a further 2–4 weeks, followed by soft foods for an additional 4 weeks, followed by a gradual return to normal foods in reduced portions.

Sample Collection

Serial fasting serum and plasma samples were taken before surgery, immediately following surgery and on post-operative days 1, 2, 3 and 4. Samples were sent for routine biochemistry and haematology analysis in the Department of Pathology using an automated multi-analyser. The remaining samples were stored in aliquots at −80 °C for batch analysis of other markers.

TNF(tumour necrosis factor)α, interleukin (IL)-6, IL-10 and the adipocytokines: leptin, resistin and adiponectin were measured using a bead-based multiplex array (Fluorokine® MultiAnalyte Profiling Kit (R&D Systems Europe Ltd, Abingdon, UK) and the results were read on a Luminex® Analyzer (Luminex B.V., Oosterhout, The Netherlands).

Enzyme-linked immunosorbent assay (ELISA) were used for cytokeratin (CK)-18 M65 and M30 (M65® Classic and M30 Apoptosense®, PEVIVA AB, Bromma, Sweden, supplied through BIOAXXESS® UK, Malvern, UK), TNF-related apoptosis-inducing ligand (TRAIL) and FasLigand (FasL) (Quantikine ELISA, R & D Systems Europe Ltd, Abingdon, UK). The manufacturers’ protocols were followed.

Efficacy Variables

The following patient specific data were collected as part of the study: demographic details, medical history, body mass index, operating time, time of liver retraction, blood loss, length of stay and perioperative complications.

Primary outcome variables were alanine transaminase (ALT) and aspartate aminotransferase (AST). Secondary outcome measures included WCC, platelets (PLT), C-reactive protein (CRP), inflammatory cytokines: TNFα, IL-6 and IL-10 and adipocytokines: leptin, adiponectin and resistin and markers of liver cell death: CK-18, TRAIL and FasL.

Histopathology

NAFLD Activity Scoring (NAS) was undertaken blind, according to the criteria set out by Kleiner and Brunt [18], defined as the unweighted sum of the scores for steatosis (0–3), lobular inflammation (0–3) and ballooning (0–2); thus ranging from 0 to 8. Scores of ≥5 are compatible with a diagnosis of steatohepatitis. Fibrosis was scored 0–4 according to the Kleiner-Brunt definitions [18].

For assessment of hepatocellular injury a local liver injury score was developed for this study, based on the non-alcoholic steatohepatitis (NASH) Clinical Research Network Scoring System Definitions set out by Kleiner et al. [18], with additional modifications to include other potentially pertinent histological changes (see Table 1).

Table 1 Parameters assessed in liver biopsy to create liver injury score
Full size table

Sample Size Calculation

In a previous study, the AST increased significantly from baseline (24 ± 6) and peaked at 24 h after laparoscopic (152 SD ± 102 U/L) and open (231 SD ± 518 U/L) RYGB but there was no significant difference in AST levels between study groups (Nguyen et al. 2003 [2]).

In order to conduct a study with 80 % power to detect a 50 % reduction in the post-operative rise in AST in relation to placebo in the population of laparoscopy patients, assuming an increase in AST level of 128 U/L and a pooled standard deviation of 100 U/L, the 50 % reduction corresponds to an effect of size 0.64, yielding a sample size of 40 patients in each of the two groups. Interim analyses were planned after 2 years. Clinical outcomes were monitored by the principal investigator and the co-investigators and overseen by the institution’s clinical trials monitoring team and the Medicines and Healthcare-related products Regulatory Authority (UK).

Randomisation and Blinding

A computer-generated randomisation sequence was used, in blocks of 10, in a 1:1 ratio. Allocation concealment was performed using sequential numbered sealed envelopes, produced by an independent statistician. Randomisation took place on the morning of surgery. Participants were blinded to treatment allocation. The clinical team was not blinded, as there was no placebo infusion used. Biomarker laboratory analysis was performed blinded to treatment group.

Statistical Analysis

Statistical analyses were performed using SPSS 20 (SPSS, USA). Demographic data and intraoperative variables were compared between groups using independent samples t test and Fisher’s exact test. Complications between groups were compared using Pearson’s chi-square test. A secondary analysis of changes in parameters over time within groups was performed using repeated measures ANOVA. Post hoc comparisons between different time points were performed after applying a Bonferroni correction. Total area under curve (AUC) for each parameter was also calculated to give a more general measure of post-operative change.

To assess the interaction between demographic and intraoperative factors and the primary outcome measures, correlations were performed using Pearson’s correlation, including log10 transformation for non-parametric variables. Clinical significance was determined if p < 0.05.

Results

Twenty patients participated in this study from July 2009 to August 2012 (see Fig. 1 for CONSORT diagram) [19]. The study was stopped after a planned interim analysis of 20 patients. Both clinical and laboratory markers displayed a wide variation in data, with large confidence intervals and standard deviations. Such variance was likely to be due to multiple factors but included deficiencies in study design. It was noted that large differences in the timing of liver retraction and its apparent effect on the liver function were also evident. Taken together with the slow rate of recruitment, the investigators felt that it was unlikely that continuation of the study to full recruitment was impractical and unlikely to reach a more reliable and valid conclusion. Therefore, the study was terminated.

Fig. 1
figure 1

Flow diagram of patients through the trial (according to the Consolidated Standards of Reporting Trials Statement) [19]

Full size image

The baseline characteristics, including age, BMI and co-morbidities were similar in both groups. All patients had minimal regular alcohol intake, with one patient having given up heavy alcohol intake more than a decade previously. Only one patient was a smoker. Operating times, liver retraction times and recorded estimated blood loss were all similar. A summary of the characteristics of the study population and the intraoperative parameters is given in Table 2.

Table 2 Characteristics of study population and intraoperative parameters
Full size table

Adverse Events

Of 20 patients, 10 patients had significant complications, meeting criteria for serious adverse events, including two patients with complications of Clavien-Dindo grade 3 or worse [20]. One patient in the treatment group had an increase in airway pressure, on induction of anaesthesia, leading to abandonment of surgery. There were no sequelae and only the demographic details are included in this analysis, as no samples were taken.

One patient had a post-operative haemorrhage, then a staple line leakage requiring two further surgeries, one patient required a two-unit blood transfusion post-operatively and one patient with pre-existing obesity hypoventilation syndrome required non-invasive ventilator support for post-operative respiratory difficulties. Four patients were readmitted within 30 days of discharge. Complications are summarised in Tables 3 and 4.

Table 3 30-day complications, classified using Clavien-Dindo grading [20]
Full size table
Table 4 Baseline and post-operative changes in clinical liver and inflammatory markers
Full size table

Primary Outcome Measures

There was a rise in ALT and AST at the end of surgery and on post-operative day 1 (p < 0.001) but no difference between treatment groups was detected (Fig. 2). Wide variations in post-operative transaminase rises between subjects in both groups were found.

Fig. 2
figure 2

Change in post-operative ALT over first four post-operative days. ALT alanine transaminase, AST aspartate aminotransferase. Values given as mean ALT (95 % confidence intervals) IU/L. No significant differences between time points (p > 0.05)

Full size image

Secondary Outcome Measures

There were no significant differences between treatment groups for changes in AST, WCC, CRP and platelets (see Fig. 3). There was a significant change in WCC between baseline, end of surgery and day 1 (p < 0.001). CRP lagged behind and rose significantly on days 1 and 2 (p < 0.001), when it reached a plateau, before falling off more slowly. Platelet count was depressed on day 1 and day 2 from baseline (p < 0.01).

Fig. 3
figure 3

Changes in WCC, CRP and platelets over four post-operative days. WCC white cell count, CRP C-reactive protein, PLT platelets. Values given as mean (95 % confidence intervals). Units: WCC, ×109/L; CRP, mg/L; platelets, ×109/L. No significant differences between time points (p > 0.05)

Full size image

Liver Histopathology Scores

The majority of patients had evidence of significant liver pathology, including 11 patients with NASH. Assessment of liver histology showed no significant differences between treatment groups, with only 5/19 patients having minimal signs of steatosis.

The mean difference in liver injury score between treatment groups of the start and end of surgery injury scores were not significantly different (mean difference in scores control 0.50 ± SEM 0.52 versus NAC 1.56 ± 0.58, p = 0.19).

Plasma Cytokines

There are large changes in cytokines following surgery (Fig. 4), with peak changes occurring within an hour of surgery in IL-6 and IL-10 (p < 0.01). TNFα levels are not significantly changed after surgery until day 2, when they fall to <50 % of baseline (p = 0.001). Changes in adipocytokines are more difficult to interpret and there are no significant differences within 24 h of surgery or between treatment groups (see Figs. 5 and 6).

Fig. 4
figure 4

Graphs of cytokines IL-6, IL-10 and TNFα following surgery. IL interleukin, TNF tumour necrosis factor. No significant differences between treatment groups. Values shown are mean (95 % CI) pg/ml

Full size image
Fig. 5
figure 5

Adipocytokine changes after bariatric surgery. Values of leptin, resistin and adiponectin are given as mean (95 % CI) ng/mL. Note the different order of magnitudes used in the y-axes as adiponectin is >1000× more abundant than resistin. No significant differences between treatment groups (p > 0.05)

Full size image
Fig. 6
figure 6

Cytokeratin-18 M30 and M65 levels increase after surgery. Values depicted are CK-18 fragments − mean (95 % CI) IU/L. No significant differences between treatment groups but significant increase between day 1 from baseline values (M65 p = 0.003, M30 p = 0.03)

Full size image

Cytokeratin-18 Fragments

There is a large rise in CK-18 post-operatively, indicating significant cellular damage but there were no significant differences between treatment groups. Both necrosis and apoptosis occurs. TRAIL and FasL were not affected by NAC infusion and these data suggest they may not play a significant role in the cellular damage that occurs in this scenario.

Correlations

NAS score at the start of operation strongly correlated with the AUC of ALT and AST (**p = 0.003 and p = 0.008 respectively). Blood loss and time of liver retraction correlated with WCC and CRP (*p < 0.05). Operation time was strongly associated with longer liver retraction times and greater blood loss (p < 0.001). Operation time correlated strongly with ALT 24 h fold change (Pearson’s r = 0.556, p = 0.017).

Initial increase in M30 fragments correlates strongly with ALT 24 h fold-change (r = 0.578, p = 0.024) and with BMI (r = 0.613, p = 0.012). There are no correlations with TRAIL and FasL levels.

Discussion

The Effect of NAC on Liver Function and Histology

During laparoscopic bariatric surgery, the combination of pneumoperitoneum and mechanical liver retraction causes hepatocellular injury. This study shows that by the end of surgery, a 5-fold increase in ALT occurs. The injury peaks by the first post-operative day. The elevation is associated with a concomitant elevation in WCC and CRP.

This is the first report of the use of NAC or any other antioxidant as a method of attenuating intraoperative liver injury in the morbidly obese. The rationale for this study was based on evidence in our unit that NAC leads to improvement in measures of quality and metabolic function of hepatocytes isolated from steatotic livers [21]. The administration of NAC in the perioperative period did not have a significant effect on post-operative parameters. There were no significant differences between treatment groups in ALT, AST, WCC, CRP and platelet counts.

The locally developed liver injury score was used to demonstrate that histological changes did occur during surgery, and the change in injury scores due to surgery was statistically significant, although the rise in score was only on average 1 point. There are no other reports of serial intraoperative liver biopsies in humans. Although the left liver lobe looks discoloured and bruised after the removal of the liver retractor, these results demonstrate that histopathological changes at that point are limited. These data suggest that histological changes due to pneumoperitoneum are subtle. Given that markers of liver injury and inflammation continue to rise on post-operative days 1 and 2, it is likely that further histopathological changes may become discernible later. It would neither be safe or ethical to obtain liver tissue after the patient leaves the operating theatre in a clinical setting. Therefore, circulating markers of liver injury are more useful in this context.

NAS score correlated strongly with the AUC of ALT and AST, confirming that patients with higher grades of NAFLD/NASH are vulnerable to more hepatocellular damage [10, 22]. Operating and liver retraction times did not directly correlate with ALT or AST levels but a correlation between log10 transformed operating time and the 24 h ALT fold change was noted, in keeping with other studies [23, 24]. Liver retraction time correlated with AUC-CRP, indicating that inflammatory response is associated with liver injury.

Clinical Outcomes

The study groups were well matched, with similar demographic parameters. There was a high incidence of other obesity-related complications, especially diabetes, hypertension, hypercholesterolemia and obstructive sleep apnoea. Operating times, intraoperative blood loss and baseline outcome measures were similar between the groups.

The trial was not powered to detect differences in clinical outcomes. BMI >60 kg/m2 is associated with higher risk of complications following bariatric surgery [25]. The rate of life-threatening complications was low, although there were four readmissions and one patient with staple line leakage. Infusion of NAC was potentially associated with one adverse event of increased airway pressure, leading to termination of anaesthesia and abandonment of surgery.

The Effect of NAC on Cytokines

IL-6 levels peaked by the end of surgery and began to return to baseline on post-operative day 1. Bariatric surgery is associated with significant falls in IL-6 over 6–24 months [26]. IL-6 rises after any surgery but the magnitude of rise after laparoscopic surgery is smaller [27]. In turn, it stimulates production of CRP by the liver as part of the inflammatory response [28]. This study confirms this relationship, as 24 h fold change of IL-6 correlated strongly with peak CRP fold change from baseline (p = 0.016).

IL-10 underwent a mean 193-fold change after surgery (95% CI 37–350). This dramatic increase may reflect the fact that most of the patients were super-obese (BMI > 60 kg/m2), diabetic and had NASH—all factors associated with elevated IL-10 [29, 30]. IL-10 has traditionally been viewed as an anti-inflammatory cytokine [31].

Changes in adipocytokines, leptin, resistin and adiponectin were not significant over the first two post-operative days and do not support the existing literature suggesting that these hormones play an important role in the acute phase response [32]. Patients with NAFLD have evidence of chronic low-grade inflammation and altered innate immunity which contributes to progression of their liver disease [33]. Post-surgical stress and metabolic changes may affect the liver [34]. At present these mechanisms are poorly understood and modulation of the immune response by targeting specific pro-inflammatory cytokines has had little clinical success [35].

Apoptosis Markers

There is only one longitudinal study showing that CK-18 fragment levels fall 6 months after surgery in patients with NASH [36]. Our data is the first of its kind measuring CK-18 fragments in the first 4 days after bariatric surgery. It shows that proportional levels of apoptosis do not increase significantly after surgery. Similar to the elevation in transaminases, this study shows that both M30 and M65 fragment concentrations increase dramatically on the first post-operative day, before returning to near baseline by day 4.

Limitations of Study

The interpretation of the results are hampered by the small sample size and premature closure of the trial.

The major drawback of the study was a lack of a standardised, reproducible toxic insult to the liver during intraoperative liver retraction with the Nathanson liver retractor. The pressure applied to liver varied from patient to patient depending on their body habitus, intraabdominal dimensions and size and texture of the liver. There are presently no routine clinical methods of measuring tissue oxygen tension or pressure within the liver. This lack of uniformity lead the investigators to conclude that a major study redesign would be required to reach an robust conclusion, even with full recruitment to the planned total sample size of 80.

Patients were not stratified by their other co-morbidities and allowed to continue their regular medications. Eleven (58 %) of patients completing the study were taking cholesterol-lowering medications, including various statins and fenofibrate. These drugs are known to cause liver dysfunction [37], although it should be noted that all patients had normal liver function at baseline.

In this study, the timing of the injury to the liver was intraoperative. Taking the results presented here which show peak transaminase rise on post-operative day 1 together with other evidence from the ischaemia-reperfusion injury literature, it is likely that the effects of liver retraction, due to compression and ischaemia, might persist for up to 12 and perhaps more than 24 h after surgery [38]. It could be argued that a one-off infusion of NAC is insufficient to counteract this injury. Many animal models investigating NAC and IRI have used a continuous infusion lasting for greater than 6 hours [39]. In effect, the methods use herein describe an in vivo model of hepatocellular injury in fatty (and indeed non-fatty) liver, and a similar methodology may be used to test other potential therapeutic interventions.

Conclusions

This study shows that bariatric surgery results in ischaemic-reperfusion injury to the liver. This can be measured with biochemical markers and histopathological scoring systems. N-acetylcysteine did not affect this outcome.