Determination of Pathophysiological Mechanisms of ALPPS in an Animal Model Using Molecular and Functional Imagings

Background: ALPPS (associating liver partition and portal vein ligation for staged hepatectomy) is a two-stage strategy to induce rapid hypertrophy of future liver remnant (FLR) to increase hepatic tumor resectability and reduce postoperative liver failure rate. Rigorous and accurate determination of the pathophysiological mechanisms involved in hypertrophy of FLR in ALPPS is essential to ensure a good success rate in the second stage operation. Methods: An ALPPS model was established in rabbits with liver VX2 tumor. The pathophysiological mechanisms after the rst stage of ALPPS in the FLR and tumor were assessed by multiplexed positron emission tomography (PET) tracers and dynamic contrast enhanced magnetic resonance imaging (DCE-MRI). Results: The tumor volume in the ALPPS model increased signicantly from post-stage 1 ALPPS day 14 compared to control animals. The 18 F-FDG uptake increased signicantly from day 7 onwards in the ALPPS model. A Valid Volumetric Function measured by 18 F-FCH showed good values in accurately monitoring dynamics and time window for functional liver regeneration (days 3 to 7). DCE-MRI revealed changes in the vascular hyperpermeability function, with a peak on day 7 for tumor and FLR. Conclusions: Molecular and functional imagings are promising and non-invasive methods to investigate the pathophysiological mechanisms of ALPPS with good potentials in clinical application to improve surgical successful outcomes.


Introduction
Liver cancer is a common malignant tumor globally [1]. Although many treatments are currently available, surgical resection remains the main treatment aiming at cure for liver cancer patients with preserved liver function [2]. The volume and functional reserve of future liver remnant (FLR) are critical for clinical outcomes after extensive liver resection, as post-hepatectomy liver failure (PHLF) is an important cause of morbidity and mortality [3]. While portal vein ligation (PVL) or portal vein embolization (PVE) can be used to induce hypertrophy of FLR, these methods require long waiting intervals for adequate liver hypertrophy in FLR for the second stage operation, thus carrying a signi cant risk of tumor progression leading to unresectability during the wait [4]. Schnitzbauer et al rst reported the novel two-stage hepatectomy procedure which was named later as "Associating Liver Partition and Portal vein ligation for Staged hepatectomy (or ALPPS in short)" [5]. The rst stage of ALPPS involves portal vein ligation and in situ splitting of liver parenchyma, whereas the second stage consists of completion hepatectomy after adequate hypertrophy of FLR. ALPPS results in more rapid liver hypertrophy than PVE or PVL, thus it can provide an opportunity for patients with primarily unresectable liver tumors to undergo curative resection [4].
Liver function and tumor behavior are the two most important factors in uencing the timing of the second stage of ALPPS and the outcomes of treatment. However, the pathophysiological mechanisms after the rst stage of ALPPS are poorly understood. Although previous clinical studies have utilized measurement of FLR volume by computed tomography (CT) as the standard to select patients for ALPPS, volume changes do not necessarily re ect hepatic function [6]. Moreover, whether the rst stage of ALPPS promotes tumor progression remains debatable. In addition, it is necessary to investigate the correlation between liver function of the FLR and tumor behavior which primarily determine the clinical outcomes of ALPPS [7,8].
Molecular and functional imagings are valuable in diagnosing liver diseases, studying non-invasively pathophysiological molecular mechanisms, and evaluating therapeutic responses. Positron emission tomography (PET), when combined with CT scan, provide both anatomical and molecular information. Furthermore, multiparametric dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) can quantitatively re ect vascular permeability of tissue angiogenesis [9,10].
Among the PET radiotracers used in clinical research, glucose analog 2-deoxy-2-18 F-uoro-D-glucose ( 18 F-FDG) is widely used to detect a variety of cancers based on the accelerated rates of glycolysis in tumor cells. 18 F-uorothymidine ( 18 F-FLT) has also been used as it accumulates in proliferating cells by binding to thymidine kinase 1 (TK-1). Moreover, as the liver is a central organ for choline metabolism and as hepatic dysfunction causes important changes in phospholipid metabolism, the 18 F-labeled choline analog, 18 F-uorocholine ( 18 F-FCH), has also been used as a PET biomarker to detect various liver diseases [11][12][13][14][15].
In this study, an ALPPS model was established in rabbits with VX2 liver tumor. The pathophysiologic mechanisms of ALPPS were investigated by monitoring volume changes, choline and glucose metabolisms, proliferation, and vascular responses in FLR and tumor with multimodal molecular and functional technologies. This study lays novel empirical groundworks in designing effective treatment strategies for liver diseases (Fig. 1).

Experimental Design
Two weeks after VX2 tumor implantation, the tumor-bearing rabbits were randomly divided into the ALPPS group (with the left branch of portal vein being ligated and the liver parenchyma being split between the left medial lobe and right lobe) and the Sham group (with the hepatic artery and portal vein being dissected without ligated). Three rabbits were selected from each group to have a DCE-MRI scan and a PET/CT scan with 18 F-FDG, 18 F-FCH, and 18 F-FLT on days 0, 1, 3, 7 and 14 after surgery. After imagings, the rabbits were sacri ced for future pathological studies.
The ALPPS model in rabbits with liver VX2 tumor This study was approved by the Animal Care Committee of Harbin Medical University. The rabbit liver VX2 tumor (Guang Zhou Jennio Biotech Co., Ltd) model was used in this study. Adult male New Zealand White rabbits, weighing 2.0-2.5 kg (provided by Harbin Medical University) were anesthetized with an intramuscular injection of 50 mg/kg zolazepam (Zoletil; Virbac, Carros, France). After a midline subxyphoid abdominal incision, and crushing of VX2 tumor into pieces, approximately 1 mm 3 of tumor was directly implanted into the left lobe of liver. The VX2 tumor was allowed to grow for two weeks. The largest tumor, which measured in average 1.5 cm in diameter by baseline T2-weighted axial MRI, was selected to establish the ALPPS model. The tumor-bearing rabbits were anesthetized by intramuscular injection of 50mg/kg zolazepam and maintained on a mixture of 1-3% iso urane (RWD Life Science Co, Ltd) and 50% oxygen and air (0.8-1.5 L/min). The animals were xed in the Trendelenburg position on the operating table. After a midline laparotomy, the left branch of portal vein was dissected, marked, and ligated with 4-0 silk (Ethicon, Somerville, NJ). The ischemic line between the left medial lobe (LML) and the right liver (RL) then became prominent, and the liver parenchyma was then split along the ischemic line [16]. All operations were performed under an operating microscope (Binocular Operation Microscope; Type XTS-4A; Jiangsu Zhenjiangzhongtian Optical Instruments Co, Ltd. China) ( Fig.2A).
The ALPPS model in rabbits with liver VX2 tumor This study was approved by the Animal Care Committee of Harbin Medical University. The rabbit liver VX2 tumor (Guang Zhou Jennio Biotech Co., Ltd) model was used in this study. Adult male New Zealand White rabbits, weighing 2.0-2.5 kg (provided by Harbin Medical University) were anesthetized with an intramuscular injection of 50 mg/kg zolazepam (Zoletil; Virbac, Carros, France). After a midline subxyphoid abdominal incision, and crushing of VX2 tumor into pieces, approximately 1 mm 3 of tumor was directly implanted into the left lobe of liver. The VX2 tumor was allowed to grow for two weeks. The largest tumor, which measured in average 1.5 cm in diameter by baseline T2-weighted axial MRI, was selected to establish the ALPPS model. The tumor-bearing rabbits were anesthetized by intramuscular injection of 50mg/kg zolazepam and maintained on a mixture of 1-3% iso urane (RWD Life Science Co, Ltd) and 50% oxygen and air (0.8-1.5 L/min). The animals were xed in the Trendelenburg position on the operating table. After a midline laparotomy, the left branch of portal vein was dissected, marked, and ligated with 4-0 silk (Ethicon, Somerville, NJ). The ischemic line between the left medial lobe (LML) and the right liver (RL) then became prominent, and the liver parenchyma was then split along the ischemic line [16]. All operations were performed under an operating microscope (Binocular Operation Microscope; Type XTS-4A; Jiangsu Zhenjiangzhongtian Optical Instruments Co, Ltd. China).
PET/CT and MR imaging protocol PET/CT imaging was performed on the tumor-bearing rabbits on days 0, 1, 3, 7, and 14 after surgery. All animals were fasted for at least 6 h before 18 F-FDG PET/CT imaging. 18 F-FDG, 18 F-FLT or 18 F-FCH (37 MBq/kg) was injected into each rabbit through the marginal ear vein at one separate day individually; PET/CT (GE Discovery PET/CT Elite, American) was performed 50-60 min after tracer injection. The rabbits were xed in the supine position after anesthesia with intramuscular injection of zolazepam. The CT was performed before the PET. The CT parameters were: 120 kV, 10 mA, and a 3.33 mm pitch. Emission data were acquired at 3 min per bed position. The images were reconstructed using an aniterative reconstruction algorithm to obtain CT, PET, and PET/CT fusion images.
MRI examinations were performed on a 3.0T MR system (GE Discovery MR750W American). After anesthetized, the rabbits were placed in a prone position within the eight channel knee coil (3T HD T/R Knee Array, GE Healthcare). DCE-MRI was performed using a three-dimensional T1-weighted fatsuppressed fast spoiled gradient-echo (3D-FS-FSPGR): TR/TE of 7.7/2.1 ms; FOV of 170×170 mm; imaging acquisition matrix of 272×160, ip angle of 20°, Slice thickness of 2.5mm,gap of 1.0mm. In total, 60 phases were acquired. An intravenous bolus of gadodiamide (GE Healthcare AS) at a dose of 0.1 mmol/kg was administered by manual injection through the marginal ear vein at 4 ml/s and was immediately followed by a 5 ml saline ush after the rst 12 phases. For T1 mapping, four pre-contrast images were acquired with the same imaging parameters by using different ip angles (5°, 10°, 20°, and 30°). It took approximately 10-11min to complete a DCE-MRI sequence with 60 phases, with each phase taking 9s. The total acquisition time of each examination was approximately 15 min. DCE-MRI data were translated into a quantitative software (Omni-Kinetics, GE Healthcare). The results in the liver were calculated based on dual-input extended Tofts mode [19]. The input function was drawn on the abdominal aorta and the portal vein. The ROI was placed manually in the largest slice of FLR, which outlined the liver shape, excluding the major vessel. A single-input Tofts model was used to t the tumor [20]. The ROIs were used to delineate the entire tumor and avoid the necrosis area. The enhancedtumor on all tumor-containing image slices were selected. The K trans (volume transfer constant) measures the e ux of the contrast agent from the intravascular space to the extravascular extracellular space (EES). Based these kinetic models, the K trans of tumor and FLR were calculated.
Liver volume and volume increase rate measured by MR imaging FLR volume (FLRV MRI ) was calculated on a workstation (AW5.0; GE Medical Systems). On T 2 WI, the FLR area was manually delineated with a freehand ROI on all FLR containing images. The total FLRV MRI was calcu lated using the following equation: FLRV MRI =ΣFLR area on each FLR-containing slice×(Slice thickness + Gap) The percentage FLR volume increase after ALPPS was based on the FLRV MRI data and calculated using the formula.

PET/CT data analysis
For the result of 18 F-FDG, 18

Valid volumetric function
The functional liver is discernible as the region of apparent 18 F-FCH uptake. To represent the functional liver region (FLR), a Valid Volumetric Function (VVF) was de ned as follows: Valid Volumetric Function (VVF) = FLR SUV mean × FLRV FCH The rate of VVF increase after ALPPS was calculated using the formula Liver volume measurement by water displacement method Three rabbits were selected to be sacri ced. The liver was fully dissected and placed into a bottle lled with water. Liver volume (FLRV water ) was equal to the nal water volume (with liver) subtracted by the baseline water volume (without liver).

Histopathologic analysis
After euthanasia, the liver and tumor tissues were xed in 10% buffered formalin, embedded, and sectioned for future histopathological analysis. Hepatocyte and tumor cell proliferation was measured by immunohistochemistry with Ki67 antibody (ab15580, Abcam, UK). The sections were also stained with the CD31 antibody (ab199012, Abcam, UK) to measure the angiogenesis and stained with a-SMA (ab7817, Abcam, UK) to access vessel maturation. The microvessel density MVD marked with CD31 was quanti ed by using the Weidner method as follows: three areas of highest MVD were selected for scoring under high magni cation (×200) within each slide, and then the average value of these three areas was recorded as the MVD [21]. Any brown-staining endothelial cell or endothelial cell cluster was considered as a single countable microvessel. The percentage of a-SMA positive stained area was determined by analysing three randomly selected elds at high magni cation (×200) from each tumor and FLR section using ImageJ (NIH, Bethesda, MD, USA). The glucose transporter 1 (ab128033, Abcam, UK) was stained to re ect the glucose metabolism. The Ki67-positive rate was quanti ed with Image J software (National Institutes of Health, Bethesda, MD). The scores of GLUT1 were estimated according to staining intensity and number of stained cells, as previously described [22]. We also performed hematoxylin and eosin (H&E) stainings to observe the histopathological.

Statistical analysis
Data were expressed as mean ± standard error of the mean (SEM). Differences among each time point and differences between groups were analyzed by two-way analysis of variance. Further data comparison between the two groups at each time point was performed using the variance homogeneity test. The independent or paired t-test was used for equal variance, and the Pearson or Spearman correlation test was used to evaluate correlations between parameters of imaging and histopathologic analyses. All statistical tests were performed with two-tailed distribution and a P<0.05 was considered statistically signi cant. Statistical analyses were performed with GraphPad Prism8.0 (GraphPad Software, San Diego, CA).

Results
The tumor-bearing rabbit ALPPS model mimicked patients with imaging features The rst stage of ALPPS was performed in rabbits with VX2 liver tumor under an operating microscope (n = 30). Most rabbits survived the procedure and had a smooth recovery. The hepatic artery was the major blood supply of VX2 tumor. Tumor size in the ALPPS group increased more rapidly (from day 14) than in the Sham group (n = 30) (Fig. 2B). Although hardly any intercellular substances could be detected by H&E staining at early time points in the tumor tissues of the two groups, from day 7 onwards more severe necrosis developed in the tumors in the ALPPS group ( g. S1A). FLR hypertrophy became prominent in the ALPPS group from day 7 onwards, but not much changes were detected in the Sham group. H&E staining of the liver tissue architecture revealed some degree of balloon degeneration, edema, steatosis, and congestion in the FLR of the ALPPS group from day 3 onwards ( g.S1B).
When 18 F-FDG, 18 F-FCH, and 18 F-FLT in the same VX2 tumor-carrying rabbit were compared using day 7 as reference (Fig. 2C), 18 F-FDG PET/CT successfully delineated tumors (with an average SUV max of 6.67 ± 1.05) and necrosis areas. However, very low radioactive concentration was detected in the tumors by 18 F-FCH and by 18 F-FLT PET/CT imagings (Table 1 and g.S2A). In addition, 18 F-FCH successfully delineated the whole liver at all time points in both the ALPPS and Sham groups. 18 F-FDG or 18 F-FLT showed visually appreciable low contrast in the liver, which was almost identical to the contrast obtained in the muscle ( g. S2 B and C). On 18 F-FLT PET imaging, a higher uptake was observed in the bone marrow of the VX2 tumor rabbit model ( Fig. 2C and g.S2C), which was consistent with the results in previous reports on animal models and humans. The 18 F-FDG tumor uptake increased in the ALPPS group from day 3 onwards and was signi cantly higher than in the Sham group on day 7. The uptake remained relatively stable thereafter (Fig. 3A). There was little uctuation in the tumor 18 F-FDG SUV max in the Sham group at all the time points (Fig. 3B). The 18 F-FDG SUV max in FLR did not change signi cantly in either the ALPPS or Sham group ( g.S3A and Table 2). The metabolic tumor volume (MTV) as measured by 18 F-FDG PET/CT increased more rapidly and became higher in the ALPPS group than in the Sham group on day 3, 7, and 14.The difference only became signi cant on day 14 ( Fig. 3C and Table 2). To further investigate the mechanisms of altered tumor 18 F-FDG uptake in the ALPPS group, the GLUT1 expression was evaluated by histological examination. Tumors in the ALPPS group had a higher glucose metabolism rate when compared with the Sham group, reaching a peak on day 7 ( Fig. 3D and g. S3B). The GLUT1-positive staining in the tumors correlated positively with 18 F-FDG SUV max ( g.S3C). However, the GLUT1 expression levels in the FLR did not change signi cantly in either of the two groups (Fig. 3E).  18 F-FCH detected proliferative function associated with choline metabolically active hepatocytes Although 18 F-FCH PET/CT imagings fail to detect the tumors( g.S3D), they successed in liver( Fig. 4A and Table 3).As the liver is a central organ for choline metabolism, the region of apparent uptake of 18 F-FCH should re ect metabolically active hepatocytes. To assess the feasibility of using 18 F-FCH PET as a noninvasive method to visualize and quantify FLR, the SUV max and SUV mean in this region were calculated. In the ALPPS group, both the 18 F-FCH SUV max and SUV mean in FLR increased signi cantly from day 1 onwards, reaching a peak on day 7, and then began to decrease gradually, although they still remained at high levels on day 14 when compared with the Sham group ( Fig. 4A-C). There was no obvious uctuation in 18 F-FCH uptake in the FLR of the Sham group at different time points. To further investigate the mechanisms of altered 18 F-FCH uptake as observed in the ALPPS group, Ki67 immunocytochemistry staining was performed on FLR and tumor histological sections. A signi cant increase in the number of proliferative hepatocytes was detected in FLR of the ALPPS group from day 1 to day 14, but not in the Sham group ( Fig. 4D and E). Moreover, strong correlation was found between the Ki67-positive rate with 18 F-FCH SUV max ( g. S4A) and with 18 F-FCH SUV mean ( g. S4B). In tumors, Ki67 was highly expressed both in the ALPPS and the Sham groups.   S4D). However, FLRV MRI was consistently higher than FLRV FCH in the ALPPS group (Fig. 5C). As the FLR volume as measured by CT or MR is typically used to identify patients for liver resection, the percentages of FLR volume increase based on MR imaging data were analyzed (Table S1).
The concept of valid volumetric function (VVF, product of mean SUV and FLRV FCH ) is to more accurately re ect the entire functional capacity of FLR. The VVF of FLR in the ALPPS group increased signi cantly from day 1, to reach a peak on day 7, and then decreased gradually to day 14 (Fig. 4D) (Table S2).

DCE-MRI detected vascular response in tumor and FLR after ALPPS
As changes in parameter K trans are associated with alterations in hyperpermeable vasculature, DCE-MRI was used to assess vascular responses (Fig. 6.A and B).In the ALPPS group, K trans increased suddenly in tumors on day 7 (P < 0.05), and then decreased slightly, but remained higher than the Sham group (Fig. 6C). In contrast, for FLR in the ALPPS group, K trans increased gradually, peaked on day 7 and then declined; while in the Sham group, K trans remained at the baseline level at all time points (Fig. 6D).
Angiogenesis in tumor and FLR was then assessed with CD31.In the ALPPS group, the CD31 MVD counts in tumor increased rapidly until day 7 (P < 0.05) and then continued to increase more gradually. In the Sham group, CD31 MVD counts in tumor increased slowly at all time points (Fig. 6E). In FLR in the ALPPS group, CD31 MVD counts increased signi cantly when compared to the Sham group, reaching a peak on day 7, and remained elevated until day 14 (P < 0.05). For the Sham group, MVD counts maintained at the baseline level at all time points (Fig. 6E). These results strongly correlated with the K trans measurements as described above ( g.S6, A and B).
Vessel maturation in tumor and FLR was assessed with α-SMA immunostaining. In the ALPPS group, the α-SMA-positive area (%) increased rapidly to reach a peak on day 7 in tumor (P < 0.05). However, in the Sham group it increased gradually (Fig. 6F). Similarly, in the ALPPS group the α-SMA -positive area (%) in FLR increased signi cantly to peak on day 7, then decreased slowly, but it still remained signi cantly higher than the Sham group on day 14 (P < 0.05) (Fig. 6F). These results strongly correlated with the K trans measurements ( g.S6, C and D).

Discussion
To conduct ALPPS safely, the rst stage of ALPPS should have su cient function and volume in FLR to prevent PHLF [23], and the second stage of ALPPS should be performed as soon as possible to limit tumor progression [24]. Thus, both the pathophysiological mechanisms of the tumor and FLR should be monitored simultaneously after the rst stage of ALPPS to determine the optimal time for the second stage of ALPPS. Over the past decades, technologies for molecular and cellular functional imagings have been increasingly used to elucidate biological processes and pathophysiology. In this study, a comprehensive assessment was conducted on volume, glucose metabolism, choline metabolism, proliferation, and vascular functional changes in tumors and FLR after the rst stage of ALPPS in the rabbit liver VX2 tumor model using multiplex molecular and functional imaging methods. The result was validated ex vivo using histological staining.
Previous studies in ALPPS have been conducted on models in mice [25], rats [26],pigs [27], and on an ALPPS model in tumor-free rabbits [16]. However, researches using these models showed only changes in FLR, and hence the in uence of ALPPS on tumors remains unclear [25,27]. In this study, an ALPPS model was established in rabbits with liver VX2 tumor to study the pathophysiological changes in tumor and FLR after the rst stage of ALPPS. The results revealed that ALPPS on the tumor-bearing rabbit model showed similar pathological features as in patients, including tumor growth with necrotic regions, and with the hepatic artery being the major blood supply to tumors. Moreover, as longitudinal imaging studies could easily be performed with this model and changes in tumor volume has been used as a standard therapeutic response marker to estimate tumor growth [28], this study showed that the metabolic tumor volume (MTV) was signi cantly higher in the ALPPS model than in Sham animals, suggesting that ALPPS accelerated tumor growth. These results are consistent with previous data obtained after PVL and PVE [29,30]. Importantly, the 18 F-FDG uptake in tumor revealed that alterations in metabolism preceded volume changes.This nding was con rmed by immunocytochemistry with GLUT1. Furthermore, K trans DCE-MRI showed that tumor vascular permeability increased signi cantly from day 7 after ALPPS to reach a high level, and Ki67, CD31 and α-SMA immunostainings showed proliferation and angiogenesis.
Thus, in ALPPS, quantitative SUV max 18 F-FDG PET/CT and K trans DCE-MR imagings revealed tumor progression before volume changes became apparent. Also, tumors in the ALPPS group had larger necrotic regions than those in the Sham group after day 7. Moreover, the normal vascular fraction was destroyed, even though neovascularization was still occurring on day 14. On the other hand, in Sham animals these were not seen. Previous studies reported that after blocking portal blood ow to tumors led to increase in arterial blood supply as a compensatory mechanism, with a consequential improvement in tumor nutritional supply which ultimately promoted tumor growth and progression [31]. Although our data supported these changes as tumor progression was detected to accelerate on day 7 in the ALPPS model, it is still di cult to determine which is the cause and which is the consequence, or whether these two processes interact. This observation can be explained by high levels of endogenous thymidine in rabbit serum when compared with human, which would compete with 18 F-FLT to bind to TK-1 to inhibit 18 FFLT uptake [32].
Another possible explanation is that with increased glycosidic phosphorylation, human liver has a higher uptake of 18 F-FLT which is a compensatory mechanism that other mammals lack [33]. Also, 18 F-FCH was also found to have a negligible uptake in VX2 tumors in the ALPPS and control animals. The possible explanation is that as VX2 tumors do not have a hepatic origin, they do not have choline metabolism.
The liver is an important organ of choline metabolism and has an enormous proliferative capacity. During cell proliferation, hepatocytes synthesize phosphatidylcholine (PtC), the most abundant membrane phospholipid. As membrane phospholipid synthesis is linked to cell cycle [11,12], radiolabeled choline can be used to re ect proliferation by highlighting membrane lipid synthesis [34,35]. Indeed, our data showed that the uptake of 18 F-FCH (SUV max and SUV mean ) in FLR signi cantly increased after the rst stage of ALPPS, and a high level of Ki67 expression was detected by immunocytochemistry.The 18 F-FCH PET technology could measure functional volume (FLRV FCH ), a volume parameter, with better accuracy than FLRV water or FLRV MRI . However, the changes in FLRV FCH and choline metabolic function in the ALPPS model were not synchronous, as 18 F-FCH uptake in FLR began to decrease after day 14, yet FLRV FCH continued to rise. For these reasons, a novel evaluation reference, VVF, was introduced to more accurately assess the functional capacity of FLR. By incorporating the parameters of SUV mean and FLRV FCH to determine VVF, the level of active choline metabolism within the total volume of the regions of interest was accounted for. Our data showed that the VVF of FLR in the ALPPS group increased signi cantly from day 1, to reached a peak on day 7 and then decreased from day 14, while the FLRV FCH remained elevated. Despite our attempts to assess glucometabolic function in FLR with 18 F-FDG PET/CT imaging, a low level of GLUT1 expression in liver cells resulted in a continuous and low 18 F-FDG SUV max uptake in both the ALPPS and control animals. Thus, the pathophysiological mechanisms after the rst stage of ALPPS do not include accelerated glucometabolism relatated with GLUT1 in liver cells [36,37].
Quantitative DCE-MRI showed that vascular permeability in FLR increased rapidly after portal vein ligation, which then decreased. These changes correlated well with changes in VVF in the ALPPS model.
Combining the results obtained by histological staining in FLR in ALPPS with Ki67 and angiogenesis immunostaining, the vascular functional changes could be interpreted as neovascularization, as there were increase in MVD counts and reduced pericyte coverage of surviving vessels. Neovascularization is required in postoperative liver regeneration to meet the increased needs of hepatic functional reserve [38,39]. Importantly, our results revealed a reduction in hepatic functional reserve and vascular maturation after day 7, suggesting there is a critical time window for liver regeneration. A recent study suggests that partial transection of at least 50% of liver parenchyma could induce a liver hypertrophy level comparable to those obtained with complete transection (137% vs 156%), the latter being the standard procedure for the rst stage of ALPPS [40]. As VVF takes into consideration both the volumetric and hepatic functional reserve factors, 18 F-FCH VVF should allow an accurate monitoring of dynamics and time window of functional liver regeneration of FLR. In the ALPPS model, the VVF growth rate reached approximately 200% on day 3 and 336% on day 7, with no signi cant differences between these time points. Notably, tumor pathophysiologic processes were found also to accelerate as early as on day 7. Thus, taken together, our data strongly suggests that between days 3 to 7 is the optimal time for the second stage of ALPPS to be carried out in the ALPPS model.
The pathophysiological mechanisms after the rst stage of ALPPS are complex, as they are in uenced by many factors including acceleration of tumor growth and progression, liver hypertrophy, and functional liver regeneration. Our study combined the molecular and functional imaging technologies to uncover the pathophysiological mechanisms of tumor and FLR in ALPPS, and to uncover the optimal time window for the second stage of ALPPS in the rabbit liver VX2 tumor model. In addition, 18

Conclusion
In summary, 18 F-FCH PET provided a highly accurate assessment of liver morphology and regeneration and revealed the optimal time for second stage of ALPPS in rabbit models, which should be between 3 to 7 days. 18 F-FDG PET/CT and K trans DCE-MR imaging detect tumor progression after rst stage operation in ALPPS earlier than volume changes. The molecular and functional imaging is a suitable non-invasive method to accurately visualize and quantify liver regeneration and to assess the in uence of surgery on tumor growth and progression. This study will help guide clinical treatment of liver cancer for surgery.