With this study we were able to explore the time course of GLP-1 and GLP-2 in a yet unexplored clinical context, which expands our current knowledge of their “incretin” properties of regulating energy substrate resorption, trafficking and homeostasis postprandially [15]. We uncovered GLP-1 and GLP-2 being dynamically regulated in human liver regeneration after hepatic resections, independent of oral food intake. Intriguingly, when liver regeneration happens to be impaired and PHLF occurred, postoperative GLP-1 and GLP-2 plasma concentrations were subjected to even more pronounced changes, compared to the dynamics seen in patients who did not develop PHLF. We could show, that these variations in plasma GLP levels were not solely associated with the volume of lost liver tissue, or the development of severe postoperative morbidity, as could have been assumed, considering previous studies which reported GLP-1 elevation in patients with sepsis [16]. Our data delineated a tremendous increase in GLP-2 plasma level, with a concomitant pronounced decline in GLP-1 concentrations in PHLF patients. This converse postoperative dynamic was unexpected, given the fact that GLP-1 and GLP-2 are released in parallel, in equimolar concentrations from enteroendocrine L-cells, triggered by the same stimulus [15].
We initially hypothesized that postoperative alterations of GLP-1 and GLP-2 concentration might be caused by changes in their degradation kinetics. DPP4 is an exopeptidase, expressed on the cell surface of many different cell-types including hepatocytes [17], but it also occurs in an enzymatically active soluble form, cleaved from the cell surface. Interestingly degradation of GLPs happen so fast after secretion, that only 25% of active GLP-1 leaves the gut and another 50% is degraded in the portal venous system with only 10–15% leaving the liver [15], underlining the major role of the liver in GLP clearance.
Given the fact, that we detected hepatic DPP4 expression on hepatocytes, in immunohistochemical staining, primarily on the bile canaliculi directed surface, primarily in the perivenous portal zone, and not in the sinusoidal space, we speculated that the tremendous degradation of GLPs on their passage through the liver is facilitated via the soluble form and not the membrane bound form of DPP4 in the portal circulation. Strikingly, we observed a significant decrease of DPP4 postoperatively in general, but the resection extend solely did not account for any differences in plasma DPP4 level, assuming a different primary source of soluble DPP4 e.g. adipose, visceral adipose tissue and also hematopoietic cells have been discussed in the literature [18]. Acknowledging these facts, postoperative GLP-1 and GLP-2 alterations in PHLF are not explainable solely by circulating DPP4 level.
Since GLP-1 and GLP-2 plasma level dynamics cannot be explained by perioperative changes of their degrading enzyme, we thought to focus on activators of GLP secretion, previously described. Bile acids are known to activate L-cells via binding to the extracellular G-protein-coupled bile acid receptor TGR5, leading to GLP secretion [19]. In liver regeneration bile acids play a critical role by inducing the cell-cycle key regulator FoxM1b [20] as well as in regulation of multiple metabolic pathways via interaction with the farnesoid X receptor (FXR) [21]. Vice versa, GLP-2 regulates hepatocyte intracellular bile acid synthesis and export into the bile fluid and systemic circulation [22]. In contrast to our observations, it has been reported, that as a consequence of liver resection, plasma bile acid level increase proportionally to the extent of resected liver volume[23]. In PHLF patients, plasma bile acid concentrations were not subjected to alterations, and did not correlate with GLP levels postoperatively, attributing them to a minor role in promoting GLP secretion in PHLF. It has to be mentioned, that the complexity of bile acid metabolism and their variety of involved regulative pathways involving cells of the entero-hepatic axis being exposed luminally and basolaterally to a plethora of bile acid derivates, might not be accurately reflected by the simple correlation analysis of GLPs and total bile acid plasma concentrations.
Moreover, L-cell activation via IL-6 might be of greater relevance [24], particularly in situations of significant trauma as hepatic resection. IL-6 is a proinflammatory cytokine with a crucial function in the priming phase of liver regeneration [25], secreted by several cell types like fibroblasts, immune cells and endothelial cells upon various signals, due to activation via tumour necrosis factor (TNF)-α or Interleukin-1 in the acute phase response [26]. Following liver resection, bacterial constituents like lipopolysaccharides (LPS) translocate into the portal venous system, resulting in toll-like receptor (TLR)-4 mediated activation of liver resident Kupffer-cells to induce TNF-α release, leading to a rapid and pronounced IL-6 secretion within the liver [27].
In that context, as expected, we observed a postoperative rapid and significant increase in circulating IL-6, but differences in IL-6 levels regarding the resection extent could not be documented. If the volume of resected liver tissue is connected with IL-6 alterations, or associated with organ dysfunction is controversially discussed in the literature [28]. IL-6 is a potent activator of L-cells [24]. In our cohort PHLF was associated with elevated IL-6 level in the perioperative course, likely explaining our observed correlation with plasma GLP-2 level during the perioperative observational period. Interestingly, GLP secretion is also stimulated via basolateral contact of LPS with the TLR-4 [29] per se and both, GLP-1 and GLP-2 are known to reduce gut permeability to prevent further bacterial translocation [29, 30] strongly suggesting a attenuating role in hepatic inflammation.
Given our current knowledge, liver regeneration is crucially dependent on lipid supply [2–4]. Both, GLP-1 and GLP-2 facilitate critical, but opposing regulatory effects in lipid absorption, trafficking and metabolism. In fact, GLP-1 mediates a decrease of chylomicron production and reduces plasma triglyceride levels, termed “fasting dyslipidaemia”, in different clinical contexts [31, 32]. On the contrary, GLP-2 is known to increase postprandial triglyceride-rich chylomicrons primarily via upregulation of intestinal apoB48 synthesis and release from preformed stores, confirmed in mice and hamsters [31, 33] as well as in humans [34]. Also by enhancing intestinal lymphatic flow, net chylomicron and triglyceride output is increased [35]. Additionally, GLP-2 acts on hepatic VLDL production and lipogenesis, as well causing elevated plasma triglyceride level and consecutively leading to hepatic steatosis [7]. Lipogenic effects are not restricted to the gut-liver axis, Ejarque et.al. reported an elevation of hormone sensitive lipase and adipocyte triglyceride lipase expression in subcutaneous and visceral adipose tissue after GLP-2 administration in mice, indicating a lipid mobilising effect [36].
The opposing postoperative GLP-1 and GLP-2 dynamics in PHLF, as we have documented, presumably mirror a pro lipogenic GLP constellation in terms of massive energy requirements of the regenerating liver. In fact, this might explain why GLP-1 administration in partially hepatectomised rats affect liver regeneration negatively [11], while regeneration in mice, receiving GLP-2 prior to partial hepatectomy was improved [10].
Given the fact that we observed the strongest pro-lipogenic GLP-1/GLP-2 constellation in patients with more pronounced postoperative declines in parameters of lipid metabolism, we hypothesise that GLPs are regulated according to the increased energy demands, representing a rescue signal in patients with impaired or exhausted liver regeneration. Interestingly, triglycerides did not display that association, assuming their homeostasis might still met in an early stage after the operation. Of note, in patients with pre-existing hepatic steatosis, histological markers of liver inflammation, NAFLD or obesity, GLP-1 baseline level were elevated, possibly reflecting a metabolic prerequisite, unfavourable for liver regeneration in that context.
At this point it has to be mentioned, that most of our knowledge about the biological actions of GLPs on lipid metabolism are derived from studies with long term GLP administration or animal gene knockout models, especially in the postprandial setting, as the scope of incretin research, up to date, has been focusing mainly on (chronic) metabolic diseases. Given the fact, that systemic GLP actions seem to change during chronic exposure [31], immediate or short notice effects might not be reflected in those studies and extrapolating their findings into our setting should be interpreted with caution.
So far, it is unclear how endogenous GLP-1 and GLP-2 precisely impact liver regeneration, but this study provides evidence that the endogenous GLP system appears to be involved. A recovering role in lipid metabolism seems likely, also Psichas et.al. reported basolateral sensing of free fatty acids in L-cells [37], which provides a hint towards a direct regulatory mechanism in facilitating lipid homeostasis. Additional GLP-2 mediated effects on liver regeneration are also considerable. An increase in mesenterial blood-flow via a nitric oxide (NO) dependant mechanism was described [38], or a negative regulatory feedback loop involving gut permeability, bacterial translocation, activation of Kupffer-cell facilitated IL-6 release and consecutive GLP secretion (discussed above) as well as GLP-2 mediated blunting of inflammatory macrophage (M1) activation [39] might also contribute and provide intriguing perspectives on the gut-liver axis in the matter of post hepatectomy liver regeneration (Fig. 5).
If these mechanisms could be addressable as therapeutic targets, has to be clarified in further studies. Our study provides a new body of evidence for endogenous GLPs being involved in human liver regeneration and offer a promising new perspective to pursue potentially addressable targets to prevent, or even treat PHLF.