Depleted polymorphonuclear leukocytes in human metastatic liver reflect an altered immune microenvironment associated with recurrent metastasis

Abstract

Background

Hepatic immunity, normally protective against neoplasia, is subverted in colorectal liver metastasis (CRLM). Here, we compare the inflammatory microenvironment of CRLM-bearing liver tissue to donor liver.

Methods

Twenty-five patients undergoing resection for CRLM were recruited, 13 of whom developed intrahepatic recurrence within 18 months. Biopsies were obtained from tumour and normal liver tissue adjacent to and distal from, the tumour. Donor liver biopsies were obtained during transplantation. Biopsies were cultured and conditioned media (CM) screened for 102 inflammatory mediators. Twelve of these were validated by Luminex assay. Transwell assays measured cancer cell chemotaxis. Polymorphonuclear leukocytes (PMN) and lymphocytes were quantified in H&E sections.

Results

Fewer periportal tissue-resident PMN were present in metastatic liver compared to donor liver. Patients with the fewest PMN in liver tissue distal to their tumour had a shorter time to intrahepatic recurrence (P < 0.001). IL-6, CXCL1, CXCL5, G-CSF, GM-CSF, VEGF, LIF, and CCL3 were higher in liver-bearing CRLM compared to donor tissue. Consequently, cancer cells migrated equally towards CM of all regions of metastatic liver but not towards donor liver CM.

Conclusions

The local inflammatory environment may affect both immune cell infiltration and cancer cell migration contributing to recurrence following resection for CRLM.

Introduction

Colorectal cancer (CRC) is the third most common cancer worldwide, with approximately 1.3 million new cases per year [1]. Metastatic disease remains the chief determinant of mortality and the liver is the most common site of metastatic spread [2–5]. Surgical resection of the tumour mass remains the best treatment option for these patients [6]. However, even following curative hepatectomy, metastases recur in as many as 60% of patients [7]. Cancer immunotherapies have dramatically improved the prognosis of many malignancies [8]. However, their optimum application remains to be elucidated [9], particularly in patients with CRLM.

A favorable microenvironment is critical to the recruitment, development, and effector function of resident hepatic immune cell populations. Microenvironmental homeostasis in the liver is maintained by a network of cytokines, chemokines, and growth factors [10]. However, tumour-associated inflammation disrupts this delicate homeostasis which, in turn, compromises local tumour surveillance mechanisms [11]. Molecular cross-talk between tumour-infiltrating immune cells, local tissue-resident immune cell populations, stromal cells, and malignant epithelial cells determines the success of metastatic disease [12]. Defining the complex molecular environment of the liver and how it changes with metastases may provide useful prognostic tools, highlighting targets for more effective therapy in metastatic liver disease.

Cytokines and chemokines control angiogenesis and tissue remodeling via progenitor cell migration and activation [13]. While known to play an active role in the hepatic tumour surveillance mechanism, specific cytokines can accelerate growth in already established hepatic malignancies [14]. We have previously found increased levels of IL-12 and IL-10 in hepatic parenchyma adjacent to metastatic deposits [15, 16] and demonstrated secretion of multiple pro-inflammatory mediators from primary colorectal tumour explants [17, 18]. Here, we extend these observations and present an in-depth analysis of the inflammatory environment of CRLM combined with histological analyses of the PMN and lymphocytic infiltrates in these tissues. We show that changes in the relative frequencies of PMN and lymphocytes in metastatic liver correlate with time to recurrence. We also demonstrate an altered cytokine microenvironment in metastatic liver compared to normal donor liver. The conditioned media generated from tumour bearing liver act as a potent chemoattractant for the CRC cell line SW620 in vitro. This suggests that the altered inflammatory environment of metastases-bearing liver may contribute to tumour recurrence both by acting to attract circulating tumour cells and disrupting local immune activity.

Materials and methods

Patient recruitment

Patients attending the National Liver Unit at St. Vincent’s University Hospital for hepatic metastasectomy secondary for tumours of colorectal origin were eligible for inclusion in this study. Twenty-five consecutive patients were recruited preoperatively (Table 1), 10 females and 15 males. Of these, 19 patients received chemotherapy within the 6 weeks prior to their surgery. All patients received leucovorin/5-flouracil; this was combined with oxaliplatin in 13 patients and irinotecan in 6 patients, while 15 patients received bevacizumab.

Table 1.

Characteristics of patients undergoing resection for CRLM

Patient demographics
Number of patients	25
Median patient age, year (range)	64 (39–86)
Male/female	15/10
Preoperative chemotherapy (n)
No treatment	6
Treatment partial response	10
Treatment no response	9
Operative approach (n)
Planned one-stage hepatectomy	23
Planned two-stage hepatectomy	2
Repeat resection	5
Primary tumour still in situ	5
Disease burden
Number of metastases, median (range)	2 (1–10)
Median tumour size, mm (range)	32 (2–420)
Satellite lesions (n)	4
Lymphovascular invasion (n)	6
Tumour Biology (n)
Mucinous adenocarcinoma	3
Poorly differentiated adenocarcinoma	2
Moderately differentiated adenocarcinoma	20
Histology of background liver (n)
Normal	8
Steatosis	16
Fibrosis	1

Liver biopsies were also obtained from donor organs prior to implantation in the setting of liver transplantation. These biopsies were obtained from deceased brain-dead patients prior to circulatory arrest. Two donor livers were included for analysis: one a 46-year-old female in whom cerebral haemorrhage following a head injury was found to be the cause of death and the other a 50-year-old male who suffered diffuse axonal injury following a road traffic accident. Neither donor had any evidence of liver disease on formal histological analysis.

Tissue sampling

Fresh tissue was obtained from surgical resection specimens and from donor organs. Biopsies were taken from the tumour, tumour adjacent macroscopically normal liver, and macroscopically normal liver at the most distant surgical resection margin [median distance, 40 mm from the metastatic deposit (range 15–96 mm). A wedge biopsy of normal liver was obtained during orthotopic liver transplantation, after implantation of the donor organ. Tissue biopsies were divided into two: one piece was formalin-fixed, paraffin-embedded, and used for histological analysis; the second piece was placed directly into complete RPMI 1640 medium, supplemented with 10% fetal calf serum and 1% penicillin/streptomycin (Gibco, Wicklow, Ireland)] and used to generate conditioned media (CM).

Histological analysis: granulocyte and lymphocyte estimation

Routine histological analysis was performed on all specimens from each region of the resected liver (tumour, normal liver tissue both adjacent to and distal from the tumour). Donor liver was assessed in a similar fashion. Formalin-fixed paraffin embedded tissue was sectioned and stained with haemotoxylin and eosin (H&E). Lymphocytes and PMN were counted in ten distinct 40x high-power fields (HPF) (for both lobular and portal regions). The same experienced pathologist (NN) examined all samples.

Generation of conditioned media (CM)

Conditioned medium was prepared following ex vivo culture of biopsies, adapting a previously described protocol [18]. Briefly, biopsies were cut into equal sized pieces of approximately 5 mm³ and cultured (in 24-well plates) in complete RPMI 1640 medium. After 72 h in culture, the CM was collected, centrifuged, aliquoted, and stored at −20 °C until used for analyses. Donor liver biopsies were cultured in the same fashion and CM samples stored at − 20 °C.

Analysis of inflammatory mediators

An initial screen of cytokines present in liver CM was performed using the Proteome Profiler Human XL Cytokine Array (R&D Systems; Abingdon UK). Samples of CM from two metastatic livers and two donor livers were analyzed according to the manufacturer’s instructions. Densitometry analysis of the array was performed using the ImageJ software (U.S. National Institute of Health, Bethesda, Maryland, USA, http://imagej.nih.gov/ij/). Raw signal intensity was normalized to background from negative control spots. The mean signal of each technical replicate was displayed as a heatmap (R studio, Version 1.0.136 & R Version 3.2.1).

Measurement of cytokine levels in liver-conditioned media (CM)

Levels of CXCL5, IL-6, CXCL1, VEGF, GM-CSF, LIF, CCL3/4, G-CSF, CCL5, CXCL10, CCL2, and IL-8 protein were determined in CM from 25 metastatic livers using the Milliplex Map Kit Human Cytokine/Chemokine Magnetic Bead Panel (Merck Millipore). Fluorescent signals were measured with a Luminex 200 reader (Luminex Corp., Austin, TX, USA). Data were calculated by generating a calibration curve obtained using recombinant cytokines specified above. Concentrations of cytokines were calculated using a five-parameter logistic curve-fitting method. Cytokine and chemokine concentrations in all samples were then normalized to the total protein content determined for each sample using a BCA assay (Pierce).

Effect of conditioned media (CM) on chemotaxis of metastatic cells

Metastatic CRC (SW620) cells were seeded at a density of 2.5 × 10⁴ cells in the upper compartment of Matrigel-coated inserts (8-mm pore size; BioCoat, BD Biosciences, Erembodegem-Dorp, Belgium) in serum-free medium. Liver-conditioned medium was used as chemoattractant in the lower chamber. After 48 h of incubation, cells remaining in the upper chamber were removed with a PBS-soaked cotton swab. Cells on the lower face of the chambers were fixed using 1% glutaraldehyde, stained with 0.1% crystal violet, and imaged under a light microscope at a magnification of 10x, averaging three random fields per insert. Crystal violet staining was quantified using the ImageJ software, and cell invasion was expressed as percentage compared to control inserts with no chemoattractant.

Statistical analysis

Statistical analyses of the frequencies of immune cells in donor and metastatic liver were carried out using the Mann–Whitney U test (GraphPad Prism software, version 6.0, La Jolla, CA, USA). Kaplan–Meier survival curves for time to recurrence were generated for patients who had neutrophil counts < 4 (N = 13) and > 4 (N = 11) (SPSS, Version 21.0 (2012), IBM Corp, New York, USA). Levels of cytokines present within different regions of metastatic liver were compared using one-way ANOVA with Tukey’s multiple comparison test (Graph Pad Software, version 6.0). Pearson correlations were calculated using SPSS and data inputted at https: discover.nci.nih.gov/cimminer/ to draw correlation matrices.

Results

Histological analysis of normal liver parenchyma distal from the metastatic deposit in CRLM

Hematoxylin and eosin (H&E) stained tissue sections of normal liver, distal from the tumour deposit, in CRLM cases (n = 25) and liver donor biopsy controls (n = 13) were examined. Lymphocytes and PMN were counted in ten separate 40× HPF per slide for both lobular and portal areas (representative images shown in Fig. 1a, b). Lymphocytes and PMN were found in all tissue sections examined. Significant inter-individual variation was noted in both the PMN and lymphocyte counts in the periportal and lobular regions from both healthy and metastatic liver [Fig. 1c (i) and (ii)]. Lymphocyte counts were significantly increased in the periportal and lobular regions of metastatic liver when compared with healthy donor tissue. Numbers of PMN were similar in the parenchyma of healthy and metastatic liver tissue; however, they were significantly decreased in the periportal regions of metastatic liver when compared with healthy donor tissue. When these results were expressed as ratios (neutrophil-to-lymphocyte ratio, NLR), even more dramatic differences were observed between healthy and tumour bearing liver [Fig. 1c (iii)].

Fig. 1.

Reduced hepatic PMN counts correlate with the recurrence of CRLM. a Representative images of donor healthy liver and b distal metastatic liver, haematoxylin-eosin staining; 40× magnification. c Neutrophil counts (i), Lymphocyte counts (ii), and neutrophil lymphocyte ratio (iii) comparing distal metastatic liver and donor liver stained with haematoxylin–eosin; 40× magnification. d Neutrophil counts (i) and Lymphocyte counts (ii) comparing donor liver, distal metastatic liver of patients that were recurrence-free and distal metastatic liver of patients that recurred stained with haematoxylin–eosin; 40× magnification. Results were analyzed with Mann–Whitney U test. The horizontal bars represent the mean of each of the groups, with standard deviation denoted by perpendicular bars, *p < 0.05; **p < 0.01 ***p < 0.001. e Kaplan–Meier curve illustrating time to recurrence in both neutropenic group (neutrophils < 4/HPF) and neutrophilic group (neutrophils > 4/HPF), significant difference calculated using log-rank test p = 0.028

Hepatic PMN counts correlate with metastatic recurrence following curative hepatectomy

Baseline histological analysis of hepatic parenchyma revealed higher neutrophil counts and lower lymphocyte counts in donor liver when compared with distal metastatic liver (Fig. 1c). On follow-up of our patient cohort over a median of 18 months, 52% (13 of 25) developed recurrent hepatic metastasis. Patients who demonstrated recurrence had fewer neutrophils at the distal resection margins, both in portal and lobular regions than patients who remained in remission [Fig. 1d (i)]. These patients also had significantly increased lymphocyte counts in both regions [Fig. 1d (ii)]. Depleted neutrophil counts (< 4/HPF) at the distal resection margin were associated with hepatic metastatic recurrence (p = 0.015). Furthermore, patients with neutropenic hepatic parenchyma had a significantly shorter disease-free survival (p = 0.028) (Fig. 1e).

Metastatic liver has a characteristic microenvironment that differs from healthy human liver

The cytokine profile of the CM from 2 donor livers was analyzed by protein array. Of 102 cytokines measured, 25 were highly expressed in donor liver CM (Fig. 2a). These included cytokines of hepatic origin such as retinol binding protein, platelet factor 4, and C-reactive protein. Other inflammatory mediators including: CCL5, CCL2, IL-8, and ICAM-1, resistin, lipocalin, macrophage migration inhibitory factor (MIF), and dipeptidyl peptidase-4 (DPP4) were also highly expressed. This suggests the presence of a constant state of dynamic benign inflammation in the healthy hepatic microenvironment.

Fig. 2.

Healthy human liver has a characteristic microenvironment that differs from metastatic liver. a Cytokine profile of conditioned media acquired from biopsies of healthy donor liver was analyzed by protein array. 102 cytokines were measured; signal at each capture spot was quantified using ImageJ software (U.S. National Institute of Health, Bethesda, Maryland, USA, http://imagej.nih.gov/ij/). Signal intensity was normalized to background (negative control spots) and displayed by heat map (R studio). Signal intensity is represented by colour legend (red-high, orange-intermediate, yellow-low/absent). b Conditioned media obtained from metastatic tumour, tumour adjacent liver, and distal metastatic liver were analyzed for the same 102 cytokines as previously described. Twelve cytokines demonstrated significant variation from donor liver and signal intensity at each capture spot quantified using ImageJ software as before

Next, we repeated this analysis using CM generated from livers bearing CRLM (n = 2). Biopsies from the tumour, normal tissue adjacent to or distal from the metastatic deposit were cultured ex vivo and CM collected. Eight cytokines, CXCL5, IL-6, CXCL1, VEGF, GM-CSF, LIF, G-CSF, and CCL3, were elevated in CM obtained from normal liver tissue distal from the metastatic deposit compared to donor liver CM (Fig. 2b). While the distal and adjacent tissue CM from patient 1 contained CCL3/CCL4, that of patient 2 did not. Lower levels of CCL5 were seen in tumour CM of both patients when compared to CM from adjacent and distal metastatic liver and, indeed, donor controls. The converse was observed with CXCL10, where lower levels were demonstrated in CM of donor and distal metastatic liver but elevated in tumour. Data presented as a heatmap show the cytokine profile of both donor livers clustered together despite demographic dissimilarity in both patients. Furthermore, both tumour CM exhibited similar cytokine profiles and thus clustered together. However, this pattern was not observed with adjacent or distal CM, suggesting that whilst a similar cytokine signature exists in the tumour microenvironment, increased inter-individual variation exists in the cytokine profile of parenchyma adjacent and distal to metastases.

The hepatic cytokine milieu is perturbed in the presence of metastasis

The cytokine profile of tissue CM from three distinct areas (tumour tissue, tumour adjacent normal tissue, and normal liver tissue at the resection margin) of 25 metastatic livers was quantified by multiplex ELISA. Figure 3a displays correlation matrices summarizing the degree of co-expression of inflammatory mediators at each site: (i) distal from the tumour, (ii) adjacent to the tumour, and (iii) tumour. CXCL5 and CXCL10 display the weakest correlation with other cytokines across all tissue sites. LIF and VEGF correlate weakly in the distal tissue but show a strong correlation in tumour.

Fig. 3.

Hepatic cytokine milieu is perturbed in the presence of metastasis. a Matrix of Pearson correlation coefficients of cytokine values in CM generated from liver tissue biopsies taken from distal (i) adjacent (ii) and tumour (iii). b Six cytokines of interest were quantified by multiplex assay to establish levels of cytokine expression across metastatic liver. Results were analyzed using one-way ANOVA followed by Tukey’s multiple comparisons test. The horizontal bars represent the mean of each of the groups, with standard deviation denoted by perpendicular bars, *p < 0.05; **p < 0.01; ***p < 0.001

Figure 3b shows the concentration of GM-CSF, LIF, CXCL10, VEGF, IL-6, and CCL5 in each individual patient at each site of liver sampled. Higher concentrations of CCL5, a lymphocyte chemoattractant, were found in both the CM of distal tissue obtained from metastatic liver, compared to tumour CM. In contrast, high concentrations of CXCL10 and VEGF were found in tumour CM but not in either metastatic or donor liver, while IL-6 is higher in tumour compared to distal, but not adjacent (Fig. 3b).

The cytokine microenvironment of metastatic liver promotes migration of metastatic CRC cells

To define functional effects of the altered cytokine profile of metastatic liver, a Boyden chamber migration assay was performed. SW620 cells (metastatic colorectal cell line) migrated towards the CM from all areas of metastatic liver (tumour, tumour adjacent, and distal liver) with similar affinity that was significantly increased compared to donor liver CM (Fig. 4).

Fig. 4.

Microenvironment of metastatic liver is chemotactic for the SW620 cells. Metastatic CRC cells (SW620) were seeded at a density of 2.5 × 10⁴ cells in the upper compartment of Matrigel-coated inserts; liver-conditioned medium was used as chemoattractant in the lower chamber. a After 48 h of incubation, cells on the lower face of the chambers were fixed using 1% glutaraldehyde, stained with 0.1% crystal violet, and imaged under a light microscope at a magnification of 10×. b Crystal violet staining was quantified using the ImageJ software and cell invasion was expressed as percentage compared to control inserts with no chemoattractant. Data were analyzed using ANOVA with Turkey’s multiple comparisons test, *p < 0.05; ***p < 0.001

Discussion

The tissue microenvironment of metastatic liver is characterised by increased levels of inflammatory cytokines and perturbation of chemokine expression. This increased local inflammation may disrupt the normal hepatic immune cell repertoire and subsequent tumour surveillance. We have found depletion of PMN in the hepatic parenchyma to be predictive of intrahepatic disease recurrence. Furthermore, CM from all areas of metastatic liver are highly chemotactic for metastatic colon cancer cells (SW620). Local hepatic cytokine and growth factor interactions are critical to the phenotype and function of tissue-resident immune cells [19, 20]. Ultimately, it is the altered cytokine landscape that drives changes in the immune repertoire, promoting malignant cell migration and ensuing disease recurrence.

Healthy human liver is a dynamic organ, undergoing constant inflammation in the maintenance of homeostasis [10]. Here, we demonstrate a broad range of inflammatory and immune regulatory cytokines and chemokines present in non-pathologic donor liver. Elevated levels of chemokines such as CCL5, CCL2, and IL-8 facilitate leukocyte accumulation [21–24] and correlate to the increased levels of innate T cells, NK cells, and monocytes seen in healthy liver [25, 26]. Furthermore, elevated levels of resistin, lipocalin, MIF, and DPP4 were observed, confirming continual active tissue remodeling in healthy human liver [27–29]. Successful establishment of metastatic disease relies on manipulation of this dynamic immune microenvironment.

Cytokines and chemokines are increasingly recognized as integral to malignant progression, not only for their ability to induce tumour cell migration but also in their ability to shape an inflammatory niche that is tumour protective [30]. Chemokines produced by epithelial cancer cells are known to attract both PMN and lymphocytes to the tumour microenvironment [31, 32]. Our analysis shows increasing concentrations of CXCL10 with proximity to tumour deposits. A known chemoattractant for T cells, these data echo recent findings of increased T cells at the invasive margin of CRLM, orchestrated by CXCL10 [33]. Further analysis of the hepatic parenchyma of metastatic liver revealed increased levels of CXCL5, IL-6, CXCL1, G-CSF, GM-CSF, and CCL3 when compared with healthy donor liver. These factors involved in neutrophil differentiation [34], migration [35, 36], and activation [37] provide evidence that in the presence of malignancy, the whole hepatic immune landscape undergoes remodeling [38, 39]. Density and type of tumoural immune cell infiltrate have demonstrated prognostic significance in CRC [40], and its metastastases [41, 42]. Indeed, our examination of histologically uninvolved hepatic parenchyma of CRLM revealed significant alterations in immune cell ratios with an elevated lymphocyte and decreased granulocyte count relative to donor healthy liver (Fig. 1d). This increase in lymphocyte density is likely driven by changes in chemokine gradients across the hepatic parenchyma, notably the increased expression of CCL5 in distal liver. While CD8+ T cells have been linked to improved outcomes in primary colorectal cancer, the phenotype and function of tumour-infiltrating lymphocytes in the metastatic setting is less clear. Though likely a complex mixture of lymphocyte phenotypes including a high proportion of T_reg cells, detailed T-cell phenotyping was beyond the scope of this present work, PMN plays a key role in host defense by mobilizing to a site of insult to phagocytose invading organisms. In inflamed tissues, neutrophils have been shown to engage in bi-directional interactions with macrophages, lymphocytes, and mesenchymal stem cells, affecting cell proliferation, differentiation, and survival [37, 43]. While increased densities of PMN are linked to tumour growth and migration [44], in the setting of CRC, peritumoural PMN have also been associated with increased sensitivity to 5-Fluorouracil-based chemotherapy [45]. Conversely following surgical stress, neutrophil extracellular traps have been implicated in the formation of liver metastases [46]. The role of PMN in the context of malignancy remains complex and may confer a pro- or anti-tumourigenic effect [47]. Much like macrophages, Fridlender et al. have described plasticity in tumour-associated neutrophils, whereupon an N1 or N2 phenotype might prevail depending on host conditions [48]. Following cytokine activation, tumour cytotoxic N1 neutrophils have the potential to kill malignant cells, thus inhibiting further tumour growth [49, 50]. The data, which we present, reflect the interaction between metastasis and hepatic parenchyma post-establishment of the tumour and prior to surgical insult. In this setting, the increased frequencies of tissue-resident neutrophils may display a more ‘N2’ phenotype as a consequence of their exposure to the tumour. This altered phenotype may, in turn, dampen inflammatory activity in the cells that influx after surgical resection. Examination of our cohort demonstrated depleted neutrophil densities throughout metastatic liver. Moreover, intrahepatic metastatic recurrence was seen in those with significantly lower neutrophil densities (Fig. 1e). When patient groups were analyzed for confounders, both those that recurred and those that remained disease-free were similar in number, size, and grade of tumours. Similarly, lymphovascular invasion was seen in both groups and not contributory on analysis. This may implicate a somewhat protective role for hepatic neutrophils against recurrent metastasis.

We have found the cytokine milieu present in all areas of metastatic liver to be chemotactic for metastatic CRC cells in vitro (Fig. 4). Moreover, the most prevalent cytokines found exhibit a predilection for granulocyte migration and function. Overexpression of granulocyte-related cytokines across metastatic liver is reflected by changes in hepatic immune cell composition, with recurrent intrahepatic disease seen in those either unable to mobilize PMN to the hepatic parenchyma, or in whom granulocyte trafficking to metastatic deposits rendered the parenchyma depleted of PMN. Analysis of peritumoural immune infiltrate has been previously proven as an accurate prognostic predictor in CRC [51], such that routine measurement of tumoural immune infiltrates is recommended as an adjunct to current staging systems [52]. Immunoscores in metastatic colon tumours have shown consistent correlation to overall survival regardless of tumoural KRAS status [53]. Similarly, components of the systemic inflammatory response reflect tumour progression and metastasis [54]. Elevated neutrophil lymphocyte ratios, associated with a specific pro-inflammatory cytokine signature in plasma, correlate with poor disease-specific and overall survivals in both primary CRC [9] and CRLM [55]. Altered immune cell ratios influence the activity of tumour surveillance mechanisms present in metastatic liver, the effects of which extend beyond the tissue planes resected at “curative” hepatectomy. Targeting the complex–immune interactions in the hepatic parenchyma in this clinical setting may provide new therapeutic options for recurrent metastatic disease.

Abbreviations

CCL

Chemokine (C–C motif) ligand

CM

Conditioned media

CRC

Colorectal cancer,

CRLM

Colorectal liver metastases,

CXCL

Chemokine (C-X-C motif) ligand

DPP4

Dipeptidyl peptidase-4

HPF

High-powered field

LIF

Leukemia inhibitory factor

MIF

Macrophage migration inhibitory factor

NLR

Neutrophil-to-lymphocyte ratio

PMN

Polymorphonuclear leukocytes

VEGF

Vascular endothelial growth factor

Author contributions

FH: study design and conception, data collection, analysis and interpretation, drafted, reviewed, edited, and approved manuscript. CH: study design and conception, data analysis and interpretation, drafted, and approved manuscript. LAE: data collection and analysis, and reviewed and approved manuscript. FC: data collection and analysis, and reviewed and approved manuscript. AL: data analysis, and reviewed and approved manuscript. DM: provided clinical samples, and reviewed and approved manuscript. EH: provided clinical samples, and reviewed and approved manuscript. NN: data collection and analysis, and reviewed and approved manuscript. JGG: study design and conception, provided clinical samples, and reviewed and approved manuscript. EJR: study conception, design and supervision, data collection, and analysis, and drafted and approved manuscript. CO’F: study conception, design and supervision, and drafted and approved manuscript.

Funding

This work was supported by grants from the Health Research Board of Ireland (RP 2008/189) to C.O’Farrelly and HRA-POR-2013-281 to E.J. Ryan and a Science Foundation Ireland Investigator Award (132/IA/1667) to C. O’Farrelly.

Compliance with ethical standards

Conflict of interest

The authors declare they have no conflict of interest.

Ethical approval and ethical standards

All protocols were approved by St. Vincent’s University Hospital Ethics Committee in accordance with the ethical guidelines of the 1975 Declaration of Helsinki.

Informed consent

Informed consent was provided by all patients prior to undergoing resection. A patient information leaflet was provided and all queries addressed by F. Hand before consent was provided. Where donor liver was obtained, informed consent was given by the donor’s family prior to sampling. Donor families were approached by a dedicated transplant coordinator. All protocols and study information were provided to the family prior to consenting.