Liver Regeneration in Surgical Animal Models – A Historical Perspective and Clinical Implications

Modern liver surgery has seen the development of split-liver grafting [1] and more aggressive, multimodal treatment of primary and secondary liver malignancies, increasing the possibilities for resection [2]. Despite continuous improvement in the surgical technique and perioperative intensive care, some patients still experience deficient regeneration and functional failure in the so-called SFSS occurring after liver transplantation if the graft is of marginal size (graft weight/body weight ratio <0.8%) [3], or if the liver remnant is too small after extended hepatectomy (<25% functionally normal liver remaining) [2]. Postresectional liver dysfunction is also a problem with the increasing use of neoadjuvant chemotherapy for colorectal metastasis [4]. The vast amount of research performed on liver regeneration to date has had relatively little practical consequences for the patient with a failing liver except for the development of liver support systems such as the molecular adsorbent recirculating system, bridging the time to transplantation, re-transplantation or as a support during recuperation of the native liver [5]. Contemporary liver surgery is therefore in need of a better understanding of the mechanisms controlling liver regeneration in order to design new treatment strategies to support the functionally deficient and failing organ. At the same time, strategies are needed to enhance its regenerative capacity be it a small-for-size graft or a failing remnant after hepatectomy with or without neoadjuvant chemotherapy. The purpose of this literature review is therefore to summarize previous experimental in vivo research on liver regeneration in animals, beginning with the Eck fistula model in 1877 up to present-day investigations, focusing on how this field has developed as a result of the interplay between clinical challenges and preclinical surgical research (see table 1 for a chronological overview of the studies discussed below). The review is organized according to past and present general themes on liver regeneration and their development over time. For each theme, we aim to highlight residual controversies, formulate new hypotheses and suggest novel experimental models in which they could be tested.

By writing this review we aim to give the reader a historical overview of the major trends in animal research on liver regeneration with emphasis on the importance of in vivo models, the value of translational research, and the necessity of increased collaboration between the basic laboratory scientist and the clinician. In our opinion, these measures are necessary if we are to make any further progress in aiding the patient with a failing liver due to insufficient regeneration.

The study of liver regeneration was largely triggered by Eck’s seminal paper on complete PCS (Eck fistula) in dogs in 1877 [6] which led to the belief that the liver’s homeostasis was not dependent upon portal blood perfusion. However, in 1893 this was contested by Hahn et al. [7] whose dogs did poorly with the Eck fistula, showing signs of liver atrophy, weight loss and encephalopathy. The changes incurred by PCS were for many ensuing years thought to be the result of a lack of sinusoidal distension and/or lack of portal flow through the liver (as opposed to a lack of the substances transported to the liver in the portal blood). The theory of sinusoidal distension was corroborated by Grindlay and Bollman [8] who in 1952 observed that the liver regenerated after a 70% PHx in dogs when constricting the vena cava above the liver, and hence increasing the hepatic venous pressure (in a Budd-Chiari-like manner). The theory of liver volume and functional maintenance by sinusoidal flow per se received much support due to Child’s model of portocaval transposition in 1953, where, after a 70% PHx in dogs, the portal vein and vena cava inferior were switched surgically, resulting in the liver remnant receiving only systemic blood from the caudal stump of the vena cava inferior, and all portal blood being diverted to the cranial stump of the vena cava inferior [9]. The observation that the remnant liver (in the portocaval transposition group) regenerated by 50% seemed to support the theory that sinusoidal flow in itself was adequate to initiate and support liver regeneration, irrespective of the quality of the perfusate. Conceptually, the observations in both studies could have been the result of a systemic overflow of growth-stimulating factors from the upper gastrointestinal tract away from the portal circulation and back to the liver via the hepatic artery in Grindlay’s experiments, and via the vena cava inferior in Child’s experiments, but this possibility is not discussed in either article. Further solid support to the ‘flow theory’ came from several canine experiments conducted in the 1950s and early 1960s with PVA (after PCS) showing that this maneuver would not only arrest the changes incurred by the Eck fistula, such as ‘meat intoxication’, weight loss of the animals and liver atrophy [10,11,12], but also allow liver regeneration to occur after a 42% PHx [13]. However, several of these [11,12] and other studies [14,15,16] also reported the development of vasculitis, periportal fibrosis, intima proliferation, lipid infiltration and the development of cirrhosis in long-term (5-year)-arterialized livers [12]. This would indicate the unphysiological nature of PVA. A note of interest at this point is the preceding work of Rous and Larimore [17] in 1920 which illustrated what they coined the phenomenon of ‘parenchymal shift’. Upon ligating one of the portal vein branches in the rabbit, they observed ipsilateral atrophy and contralateral hypertrophy of the liver, postulating ‘the liver is wholly a portal organ, finding its reason for being in the substances carried to it in the portal blood’. This study had clearly implicated the importance of the humoral effect of splanchnic inflow on liver homeostasis; however, the flow theory prevailed unchallenged until the mid 1960s.

With the advent of auxiliary liver grafting in the 1960s unveiling the phenomenon of graft atrophy due to the portal steal effect came the realization that there must be certain substances delivered to the liver in the portal blood only, upon which the organ is dependent to regenerate and/or maintain its volume and function [18]. One could no longer regard liver homeostasis as a result of mechanical portal flow stimuli (which was hypothesized by Rous and Larimore [17] 40 years earlier). A period of intense investigation followed from 1965 to 1978 with a large body of experiments revealing the importance of the hormonal and nutritional effect of portal blood on liver regeneration, in particular insulin (table 2). This period commenced with canine models of split portocaval transposition (one portal branch perfused with blood from the vena cava inferior and the other portal branch perfused with portal blood, with similar flow rates and oxygen tension). This eliminated the possible confounding effect that graft rejection could have had in causing graft atrophy in previous auxiliary transplantation experiments. After 3 months of split portocaval transposition in dogs, Marchioro et al. [19] observed hypertrophy with glycogen deposition, increased DNA synthesis and mitosis on the side receiving splanchnic blood and atrophy on the side receiving systemic blood. The importance of portal blood supply to the liver’s homeostasis in auxiliary grafting was consequently corroborated the same year (1965) by Tretbar et al. [20], Thomford et al. [21] and Halgrimson et al. [22]. Marchioro et al. [23] also substituted the blood flow in one of the portal vein branches over 3 months, observing that increasing the flow and oxygen supply to one side of the liver could not compensate for the qualitative loss of the portal blood stimulus.

Given that the trophic substances seemed to be found in the portal blood, subsequent investigations were designed to disclose their origin. Separating the portal inflow coming from the upper gastrointestinal tract (distal stomach, duodenum, pancreas and spleen) from that originating from the small intestine (model of splanchnic flow division), Marchioro et al. [24] transplanted auxiliary liver grafts supplying them with portal blood from the small intestine only, while the native liver received pancreaticogastroduodenosplenic blood. This resulted in atrophy, centrilobular necrosis, cytoplasmic fat deposits in hepatocytes, hemosiderin deposits in Kupffer cells and collapse of the reticulin framework in the grafts [24]. Further research in the 1970s utilized various canine models of splanchnic evisceration: Price et al. [25,26 ]eviscerated dogs, simultaneously performing portocaval transposition (maintaining liver perfusion with systemic blood), with or without glucagon infusion and suggested that glucagon was a main factor as it could reverse the many effects of portal blood deprivation. This was challenged by Starzl et al. [27] a year later in experiments of splanchnic flow division with diversion of the pancreaticogastroduodenosplenic venous blood to one half of the liver and blood from the small intestine to the other half resulting in atrophy of liver parenchyma not receiving pancreaticogastroduodenosplenic venous blood, postulating that the substance in question was insulin. This theory was later corroborated by the same investigator upon constructing an Eck fistula and infusing insulin into one of the portal vein branches, which limited the atrophy (on the side receiving insulin) caused by PCS [28] and further fortified by observing liver atrophy and lack of regeneration after PHx in alloxan-induced diabetes mellitus in dogs [29]. However, contrary to the findings of Starzl et al. [30], Duguay and Orloff [31] did find some additive effect of glucagon infusion in addition to insulin in 1976. At this stage, it became apparent that the liver regeneration observed in Child’s model of PHx and portocaval transposition, once thought to be the result of flow stimulus, was in fact the result of redirecting the portal stimulants via the PCS to the systemic circulation and back to the liver via the vena cava inferior [32]. Further experiments in dogs with various degrees of splanchnic evisceration and PHx followed by portal infusion of insulin and/or glucagon confirmed the importance of insulin in the process of liver regeneration but also demonstrated that this hormone could not compensate for total evisceration [33]. This hormonal effect was also demonstrated in the liver after splanchnic evisceration and PCS and intraportal insulin infusion, preventing atrophy and glycogen depletion and promoting DNA synthesis [34].

With the importance of insulin for liver maintenance and regeneration firmly established, research evolved to screen for other potential hepatotrophic substances in the portal blood. Francavilla et al. [35] utilized a canine Eck fistula model in 1991, infusing T₃, glucagon, prolactin, angiotensin II, vasopressin, norepinephrine, estradiol, IGF-II, HSS, TGF-α, HGF (also termed hepatopoietin-A), EGF, TGF-β, tamoxifen, IL-1, IL-2 and insulin into one of the detached portal vein branches above the shunt [35]. Insulin and partly T₃, IGF-II, HSS, TGF-α and HGF inhibited liver atrophy. The remaining substances were inert. Interestingly, TGF-β increased atrophy, but this effect was reversed upon concomitant insulin infusion. This is an important study because it illustrates the importance of performing in vivo studies when studying a complex and integrated process such as liver regeneration: several of the above substances were found to enhance hepatocyte replication in in vitro studies (EGF [36], angiotensin II [37] and norepinephrine [38]) but they were not active when placed in context in a living, biological system, as was done in the above study. Certainly, cell culture models have their advantage in that one may study signaling pathways in individual cell lines without the confounding effects of different cell types, but the cross-talk between the extracellular matrix and different cell types in the liver parenchyma known to be particularly important to liver regeneration is missed [39]. This leaves us with the classic scientific paradox of the investigator changing the things he aims to observe by his act of intervention when utilizing the more ‘mechanistic’ model of cell culture. In contradistinction to this, animal models provide a more integrated and realistic means to study the highly coordinated process of liver regeneration.

Despite the numerous surgical models with splanchnic vascular manipulation and liver transplantation over a period of approximately 100 years implying the dominant role of humoral regulative mechanisms initiating liver regeneration and maintaining liver homeostasis, new studies appeared suggesting that increased sinusoidal flow after PHx could play a role after all. In the late 1990s, Sato et al. [40] and Niiya et al. [41] suggested that the acute portal hypertension and increased shear stress over the sinusoidal endothelial surface caused by PHx (and increased flow per gram remaining liver) triggered the regeneration cascade. In the same period, increased endothelial shear stress was found to modulate the endothelial production of nitric oxide and influence the hepatic vascular bed [42,43,44]. Later, liver regeneration after PHx in rats was shown to be inhibited by the administration of the NO antagonist N^G-nitro-L-arginine methyl ester, and restored by the NO donor 3-morpholinosydnonimine-1 [45,46]. This potential renaissance of the flow theory was challenged by Mortensen et al. [47] in a porcine model of PHx and gene expression analysis. By increasing the degree of liver resection (and consequently the rise in portal pressure and flow per gram remaining liver tissue) they observed a switch of the genetic response in the liver remnant from one of primarily cell cycle propagation (after 62% PHx) to that of modulation of the intracellular redox status and the caspase cascade (after 75% PHx) [47]. The different genetic response was proposed to be either due to the differences in sinusoidal pressure/shear stress and flow per gram remaining tissue, or due to differences in the amount of portal hepatotrophic substances delivered to the remnant. This was further investigated by constructing an aortoportal shunt from the aorta to the left portal vein, selectively increasing the flow to segments II, III and IV to the same flow levels as that seen after a 75% PHx (2.89 ml/g/min). The investigators observed that these segments remained unchanged over 3 weeks, whilst the right side of the liver, receiving only portal blood in the same period, hypertrophied. The augmented flow seemed to inactivate the hyperperfused segments, as was reflected by the general down-regulation of transcriptional activity (according to microarray analysis). This suggested once more that increased flow in itself is not an adequate stimulus to trigger either hypertrophy or hyperplasia of the liver – the flow must be of splanchnic origin [48]. At the same time, the importance of the remnant perfusate quality (vs. quantity) was further illustrated by Ladurner et al. [49] who performed a 75% PHx in pigs assigning one group to receive a side-to-side portosystemic H shunt decompressing the portal system. As expected, the portal vein flow (to the liver remnant) in the H-shunt group was significantly lower than in animals without the shunt. However, the livers in both groups showed no differences in regenerative response, again providing evidence in support of the dominant stimulatory role of portal blood constituents.

To conclude, the flow theory seems again less credible although the optimal amount of portal and sinusoidal flow in the liver remnant seems to be undetermined. However, what is established is the damage caused by too much sinusoidal flow, as observed in the clinical scenarios of what has been termed SFSS [3]. Could it be that this syndrome is not only the result of the well-recognized sinusoidal congestion and endothelial damage, but also due to the lack of liver regeneration? If so, this would indirectly be an argument for the role of the flow theory. The observation of the fact that a graft weight/body weight ratio <0.8% predisposes to SFSS has resulted in the assumption that portal hyperperfusion is the main culprit as the flow per gram liver tissue through the liver sinusoids increases to a harmful level. Evidence that this is the case is seen in preclinical large-animal models with portal vein decompression by portosystemic shunting in transplantation of small-for-size grafts [50,51,52,53,54,55] and in combined models of hepatectomy with marginal liver remnants and portosystemic shunting [56,57]. Portal vein decompression, a normalization of the portal vein pressure and portal flow improves liver regeneration. This has been confirmed in clinical (human) studies [58,59,60,61]. As a decrease in portal inflow results in a reciprocal increase in the hepatic artery flow due to the hepatic arterial buffer response [62], one could speculate that one of the reasons for the improved graft function and regeneration with portosystemic shunting is the increased oxygen tension in the regenerating liver, which brings us to the next topic.

In a canine Eck fistula model in 1952, Cohn and Herrod [10] observed that arterialization of the portal vein stump over 4 months prevented liver atrophy. The oxygen tension in the hepatic veins was similar to control groups, indicating a pronounced oxygen extraction by the liver deprived of its portal blood supply. Consequent long-term canine experiments performed the next 10 years in animals with Eck fistula and portal vein stump arterialization showed that this altered hepatic vascularity was compatible with life, although many reported the development of vascular damage and liver fibrosis progressing to cirrhosis [11,12,14,15,16]. However, in 1954, Fisher et al. [13] did report superior liver regeneration after a 42% PHx in a dog model with Eck fistula and arterialization of the portal vein stump (controls regenerated to 80% of original volume vs. 103% in the arterialized group) and hypothesized that the increased oxygen delivery (and preserved flow) contributed to this difference. Furthermore, in 1967 Mito et al. [63] observed growth of a partial heterotopic homograft arterialized with an aortoportal shunt (grading the pressure in the shunt to 25–35 mm Hg by a Teflon cuff) over a period of 16 days. Notably, this occurred despite the fact that the native liver received all the portal flow. In 1975, Horak et al. [64 ]also noted similar protective effects of PVA in Eck fistula dogs over a period of 10 weeks (also grading the arterial shunt flow with flow probes to equal flow in the portal vein before establishment of the Eck fistula) and, recently, Ott el al. [65] observed increased regeneration of the porcine liver remnant after PHx upon PVA compared to pigs with portal perfusion of the liver remnant. Not all investigators have reported beneficial effects of PVA though: in a study by Marchioro et al. [23], the arterialized side of the liver atrophied after 60 days in a canine model of split portocaval transposition and portal vein branch arterialization. This, however, occurred in conjunction with undisturbed portal flow to the contralateral side of the liver which hypertrophied, in effect, taking over the liver function (not unlike the ‘parenchymal shift’ described by Rous and Larimore 47 years earlier [17]).

In the clinical setting, PVA has been found useful in counteracting the portoprival state of the liver and hepatic encephalopathy in cirrhotic patients with Eck fistula [66,67], but later experience with PVA of liver grafts has varied from being problematic due to histological changes in microsteatosis and fibrosis [68] to unproblematic in other series with good long-term liver function [69].

How can we explain the apparent beneficial short-term effects of PVA on regeneration and liver homeostasis and what is relevant to research on liver regeneration? As obvious as it may seem, rodent models of PHx from the 1970s and canine models from the 1990s have shown that the capacity of the liver remnant to regenerate after PHx is dependent upon an increased supply of energy [70,71,72]. After PHx in rats, Yoshioka et al. [73] showed that oxygen supply to the liver increases by increased hepatic artery flow. Simultaneously, the hepatic oxygen extraction rate increases, while the total energy load decreases along with increased DNA synthesis. Arterialization of the liver remnant leads to improved survival in rats after extended hepatectomy [74,75,76], and this has also been shown to be beneficial in humans after extended hepatectomy [77]. While investigating the mechanisms behind SFSS, Smyrniotis et al. [78] studied the hemodynamic changes in differently sized liver grafts in pigs and disclosed that while the portal pressure and flow per gram liver increased inversely with graft size, hepatic artery flow decreased. However, the hepatic arterial buffer response was preserved, even showing an increased response with decreasing graft size. One could therefore hypothesize that a graded PVA could prove beneficial for the function and regeneration of the marginal liver remnant and the small-for-size liver graft, as arterialization potentially leads to an optimal oxidative status and energy charge in the hepatocytes. Accordingly, a surgical model of extended hepatectomy with arterialization of the functionally small and deficient remnant with observations of energy charge and histological signs of regeneration could cast light on this aspect and potentially be used as a bridge to complete regeneration in patients with small-for-size grafts. To avoid the deleterious effects of long-term arterialization in the patient, the end-to-side shunt should be embolized upon completed liver regeneration or normalization of liver function.

The preceding sections have focused on how liver homeostasis and regeneration is influenced by the amount, pressure and the composition of its blood supply and drainage, but the liver itself is also a source of growth factors and cytokines which play a vital role in regeneration. In 1939, Brues et al. [79] found that a liver extract from adult and embryonic porcine liver would inhibit growth of cultured fibroblasts, with the process being reversible, suggesting that the liver was a source of growth inhibition. However, in 1952, Glinos and Gey [80] found the serum of PHx rats to exert a growth-promoting action on fibroblasts in tissue culture. Around the same time, Bucher et al. [81 ]and Wenneker and Sussman [82] reported an increased number of mitoses in the non-hepatectomized partner in parabiotic rats with cross circulation indicating the presence of growth-stimulating factors in the effluent from the liver remnant [81,82]. This hypothesis was corroborated in in vivo rodent models by several investigators who observed increased liver cell mitosis in intact animals injected with serum from hepatectomized counterparts [83,84]. To circumvent the changes in portal hemodynamics caused by PHx, Sigel et al. [85] conducted canine experiments in the early 1960s with autotransplanting of small liver grafts to the jejunal mesentery, later randomizing the animals to a 70% PHx of the native liver. In contrast to control groups, the autografts in the animals with 70% PHx did not undergo atrophy, indicating again a growth stimulus from the resected liver to the autografts via the systemic circulation [85]. Similarly, in another experiment, autografts transplanted to the neck did not undergo atrophy, tentatively stimulated by the native liver manipulated with an Eck fistula (in contrast to animals without an Eck fistula) [86]. Thomford et al. [21] showed similar results in 1965 in dogs with heterotopic allografts, where the grafts did not suffer from atrophy when the native liver, receiving all the portal blood, was resected, again indicating a growth-stimulating effect from the liver effluent after PHx. Fourteen years later, Starzl et al. [87] extracted cytosol from hepatectomized canine livers (48 and 72 h after PHx) injecting it into the portal vein stump of Eck fistula dogs and observed a proliferative response. A year later, it was observed that the growth-stimulating factor in the cytosol extract from regenerating canine livers (termed HSS) was organ specific in that it did not stimulate any glomerular proliferative activity when injected into the renal artery. Additionally, HSS was not found in liver extracts from animals that underwent splanchnic evisceration synchronously with PHx, indicating that its synthesis in the liver probably was a result of splanchnic ‘collaboration’ [88]. Investigating how factors in the recipient liver influenced the action of HSS, Terblanche et al. [89] injected regenerative liver extract into the portal vein perfusing normal canine livers without any response. However, an augmented proliferative response was seen upon injecting the extract into the portal vein of the liver remnants 48 and 72 h after PHx. Further investigations of possible growth-stimulatory substances in the liver effluent from PHx pigs were performed by van Hoorn-Hickman et al. [90] in 1981 by cross circulation with recipient animals or exchange perfusion. Increased thymidine kinase activity and mitotic indices in biopsies from PCS (recipient) pigs corroborated Starzl’s previous observations in dogs. Kahn et al. [91] also showed in 1982 that a stimulatory substance was transferred from a transplanted PHx liver to the host liver (which had PCS), stimulating a proliferative response in the latter (as judged by increased thymidine kinase activity and mitotic indices). The authors discussed whether this phenomenon could be clinically useful in aiding liver regeneration in the host liver in patients with liver failure treated by auxiliary liver grafting, leading to the question of whether the development of an acute or acute-upon-chronic liver failure in a large-animal model is useful to test the effect of injecting serum extracts of liver hepatotrophic substances from resected livers on the regeneration of damaged liver. Is it possible that infusion of a concentrate of the patients’ own serum in the portal vein may assist in the regeneration of the remnant liver after an extensive liver resection and could a small-for-size graft procured from living donor split-liver grafting profit or in some way be supported by the stimulus that the serum of the donor could offer?

During the last 20 years, the focus of research on liver regeneration after PHx has turned from examining extrinsic hepatic factors, such as portal and hepatic arterial blood flow and its content, to the intrinsic consequences these changes have in the extracellular matrix, the intracellular signal transduction mechanisms and genetic response in the liver. Studies in various cell culture models, stem cell transplantation, microarray analysis in rodent and porcine models, the impact of the immune system, blood platelets and serotonin, the complement system, cytokines and the interaction between the many different cell types now known to regulate the regenerative process have all unquestionably added much knowledge to the research on liver regeneration. However, at the same time, these studies have made the picture complex and seemingly increasingly intangible when it comes to the clinical application of the knowledge gained.

Although this review comprises traditional surgical animal models, novel genetic engineering with rodent knockout models warrants mention because this area of research will most likely become increasingly dominant in future investigations of liver regeneration (table 3). This technique is unquestionably very powerful, in a deterministic sense, and has illuminated the roles of several important genes and gene products concerned with the priming of hepatocytes (IL-6, TNF, C3a, C5a, MyD88 and TIMP3), extracellular matrix remodeling (uPA, MT and MMP9), intracellular signal transduction (C/EBP-β, IGFBP-1, HB-EGF, OSM, TGF-α, Met and caveolin-1), lipid metabolism (PPAR-α) and termination of liver regeneration (TGF-β). The challenge, on a larger scale, however, lies in linking all the bits of information together.

Extensive reviews of the vast amount of more contemporary published literature on the molecular control of liver regeneration in the past 20–30 years have been written by authorities on liver research and require mention here. In 2004, Taub [92] outlined the ‘cytokine’ and ‘growth-factor’ pathways eventually leading to the activation of downstream intracellular signaling substances promoting cell cycle entry and mitosis, and the (relatively few number of) molecules arresting liver growth, maintaining a constant liver/body mass. The review concludes that what remains unclear is how the size of the liver is determined, ‘how the known molecular pathways necessary for liver regeneration are altered in human disease’, and that ‘greater insight will be required to develop improved pharmacological therapeutics and surgical approaches’. There is no mention of any clinical application of the knowledge gained so far [92]. In 2006, Fausto et al. [93] reviewed research in liver regeneration and also described the ‘cytokine’ and ‘growth-factor’ pathways and their interaction and, additionally, the influence of metabolic stimuli in regeneration. They concluded that the study of liver regeneration affords a unique model to study signal transduction mechanisms and cell cycle events and emphasized the importance of understanding the process of liver regeneration for the appropriate management of acute liver failure and liver cirrhosis. They added that what is needed is a more ‘rigorous effort to apply the knowledge gained in experimental work to solve clinical problems’. However, the review does not mention if and how all this scientific work has resulted in a single therapeutic procedure [93]. In his published Rous-Whipple Award Lecture from 2010, Michalapoulos [39 ]summarized the control of liver regeneration by the so-called ‘complete mitogens’ that are mitogenic in hepatocyte cell cultures (HGF and receptor c-MET and ligands of the EGF, TGF-α, HB-EGF and AR), and by ‘auxiliary mitogens’, the ablation of which delay the regeneration process (amongst others, norepinephrine, TNF, IL-6, VEGF, bile acids, serotonin, complement proteins, leptin, FGF-1, FGF-2 and insulin). A redundancy exists in all these different pathways as there seems to be a considerable overlap between the many signaling cascades – blockage of one route is compensated by another, leading to completion of liver regeneration. In this comprehensive review, the author concludes that ‘liver failure, essentially a failure of regeneration, should be subject to mechanistic analysis based on knowledge already gained on regeneration, and perhaps therapeutic interventions may be designed with impact on human liver disease’ [39]. Hence it seems clear that the authors are aware of the gap between basic laboratory investigation and clinical appliance in this field.

How then has research on liver regeneration in the past 133 years benefited the patient with liver cirrhosis, acute liver failure or liver metastasis? The recognition of the importance of portal blood to liver homeostasis and regeneration was obviously crucial to the pioneers of liver transplantation as they observed how the auxiliary graft would undergo atrophy without portal blood constituent stimulus [18]. The earlier canine and porcine PVA experiments [10,11,12,13,14,63,64,65] also illustrated to the transplantation surgeon that the auxiliary graft could be perfused by PVA as an option to leave the hilus of the native liver untouched and also in cases of portal vein thrombosis [94,95,96,97]. Could PVA of the small-for-size graft be an option in the future to avoid the small-for-size syndrome?

Furthermore, as part of the emerging multimodal three-stage treatment of colorectal metastases [2], surgeons may now embolize the portal vein before performing large resections in order to stimulate liver hyperplasia in the remnant to be. By this, one may avoid postoperative liver failure, acknowledging that diverting portal flow away from one side to the other results in the ‘parenchymal shift’ described by Rous and Larimore [17] in 1920.

The surgical principles and practice of preoperative portal vein embolization to induce hyperplasia of the remnant liver after PHx, and portal vein decompression by portosystemic shunting to reduce sinusoidal congestion in the case of SFSS after liver transplantation are well established. Apart from these, there are no novel patient therapies available to aid and augment the process of liver regeneration after extended liver resections, toxic liver insults or in the cirrhotic patient. This is in spite of all the modern technological advances and the knowledge gained on the microscopic and molecular aspects of liver regeneration in the past 20–30 years. We suggest that it is time to turn back to the systemic large-animal surgical research on liver regeneration as it offers a more integrated, systemic biological understanding of this complex process, and that a more clinically relevant progression could possibly be made with a closer collaboration between the hepatologist, liver surgeon/transplant surgeon and the laboratory scientist.

The authors have no conflict of interest.

Abbreviations: AR = Amphiregulin; C/EBP-β= CCAAT enhancer-binding protein-β; C3a = complement factor 3a; C5a = complement factor 5a; EGF = epidermal growth factor; FGF = fibroblast growth factor; HB-EGF = heparin-binding epidermal growth factor; HGF = hepatocyte growth factor; HSS = hepatic stimulatory substance; IGFBP-1 = insulin-like growth factor binding protein-1; IGF = insulin-like growth factor; IL = interleukin; iNOS = inducible nitric oxide synthase; Met = hepatocyte growth factor receptor; MMP9 = matrix metalloproteinase 9; MT = metallothionein; MyD88 = myeloid differentiation factor 88; OSM = oncostatin M; PCS = portocaval shunt; PHx = partial hepatectomy; PPAR-α = proliferator-activated receptor-α; PVA = portal vein arterialization; SCF = stem cell factor; SFSS = small-for-size syndrome; T₃ = triiodothyronine; TGF = transforming growth factor; TIMP3 = tissue inhibitor of metalloproteinase 3; TNF = tumor necrosis factor; uPA = uroplasminogen activator; VEGF = vascular endothelial growth factor.