Introduction

The expanding resectability criteria for liver malignancies has made partial liver resection a standard procedure for an increasing number of patients [1, 2]. This procedure is offered even to elderly patients, including those over 80 years old, without increased liver related-morbidity and mortality [3].

Low future liver remnant volume (FLR-V) and underlying parenchymal disease have been identified as important risk factors for post-hepatectomy liver failure (PHLF) and impaired liver regeneration after partial liver resection [4–6]. An FLR-V > 25 % of the total functional liver volume is considered to indicate a safe and feasible partial hepatectomy [7–9]. Selective embolization of the right portal branch (PVE) has been shown to increase the FLR-V and thereby minimize the PHLF risk [6, 10]. However, to our knowledge, the process of liver function shift after PVE has not been described before.

If there is dysfunction of the liver parenchyma, such as that caused by steatosis, cirrhosis, or preoperative chemotherapy, the FLR-V alone is insufficient to assess the individual PHLF risk [11–14]. Thus, the future liver remnant function (FLR-F) should be used to predict the post-hepatectomy outcome [4, 15]. Different methods of investigating the FLR-F have been investigated, including the indocyanin green retention test [16], 99mTc-labelled galactosyl-human serum albumin scintigraphy [17–19], and 99mTc-labelled mebrofenin hepatobiliary scintigraphy [18, 20, 21]. Mizuguchi et al. [22] recently reviewed the literature on currently available methods of preoperative liver function assessment. We propose a new diagnostic approach to estimate FLR-F by using the LiMAx test and CT scan-assisted volume/function analysis.

The LiMAx is a 13C-based breath test that assesses actual liver function, the diagnostic power of which has been proven in several clinical settings. In particular, the LiMAx test has been shown to predict postoperative morbidity and mortality after partial liver resection when performed on postoperative day (POD) 1 [4]. Recently, our group proposed guidelines for using the LiMAx test with CT-assisted volumetric analysis to predict FLR-F [5]. The aim of this preliminary study was to analyze the changes in liver function after PVE and prove the feasibility to predict FLR-F by the LiMAx test and CT-scan assisted volume-function analysis in patients undergoing extended hepatectomy either after PVE, or without any preconditioning.

Methods

This prospective observational study was performed in the Department of General, Visceral and Transplantation Surgery at the Charité Berlin, Germany, between January 2005 and December 2009. The study was performed in cooperation with the Department of Diagnostic and Interventional Radiology at the Charité Berlin. The study protocol received prior approval from the local ethics committee and informed consent was obtained from all subjects before their inclusion. The trial was performed in accordance with the precepts established by the Helsinki Declaration. All patients evaluated to undergo extended right hepatectomy for primary or metastatic liver malignancy were screened for the exclusion criteria for this study, as follows: a history of liver surgery, apart from cholecystectomy; severe infectious disease such as HIV; and mental health disorders. Thereafter, the decision was made as to whether the patient would undergo portal vein embolization (PVE) to improve the FLR-V or whether they would undergo the hepatectomy directly (NO-PVE). This decision was made by the surgeon in charge according to the initial FLR-V volume. Postoperative outcome was assessed by 30-day mortality and complications were evaluated according to the Clavien/Dindo classification [23]. Liver function was measured using the LiMAx test consecutively in all patients until discharge and on POD 85. Liver histology was evaluated semi quantitatively according to the Batts and Ludwig scoring system [24]. Fibrosis was staged on a 0–4 scale as follows: F0, no fibrosis; F1, portal fibrosis without septa; F2, portal fibrosis and a few septa; F3, numerous septa without cirrhosis; and F4, cirrhosis.

Portal vein embolization and hepatectomy

PVE was performed in an analogous manner by two experienced interventional radiologists, via a percutaneous transhepatic ipsilateral approach. A catheter was inserted into the right portal vein by a trans-hepatic CT-guided puncture of the right portal branch. Briefly, after establishing percutaneous ipsilateral transhepatic access, direct portography was performed to visualize the portal vein anatomy. Branches of the right portal vein were then catheterized selectively, followed by embolization with PVA particles (500–710 μm Contour, Boston Scientific, Natick, MA, USA) until complete stasis was achieved. If segment IV portal vein branches arose from the left portal vein, these branches would not have been embolized selectively. The success of portal vein embolization was controlled by direct portography. Patients were discharged 1 day after PVE and then readmitted after an interval of 4 weeks.

On readmission, they underwent standard preoperative work-up, including contrast-enhanced four-phase CT scan. Every patient underwent extended right hepatectomy, including segments V–VIII of the liver and partial or complete resection of segment IV and segment I resection (en bloc). Portal vein resection and biliary anastomosis was performed for all Klatskin tumors. Liver pedicle clamping (Pringle maneuver) was performed if there was profuse parenchymal bleeding, and this decision was made by the surgeon intraoperatively.

Dynamic liver function test

Liver function was assessed with the LiMAx (MAximum Liver function capacity) test, as described previously (9, 20). Based on the hepatocyte-specific metabolism of the 13C-labelled substrate Methacetin (Euriso-top, Saint-Aubin Cedex, France) by the cytochrome P450 1A2 enzyme, which is ubiquitously active throughout the liver, 13C-methacetin is instantly metabolized into acetaminophen and the demethylated 13C-group after its intravenous injection at 2 mg/kg. This is converted into 13CO2 and is exhaled, leading to a significant alteration of the regular 13CO2/12CO2 ratio in the expired breath. A suitable device connected to the patient measures this change (FLIP®, Humedics company, Berlin, Germany) and analyzes the breath automatically. Liver function capacity is calculated from the kinetic analysis of the 13CO2/12CO2 ratio over a maximum of 60 min.

Volumetric analysis

Patients in the PVE group had a multiphase contrast-enhanced CT scan done directly before PVE and before hepatectomy. Patients in the NO-PVE group were assessed directly before resection, at which time a volumetric analysis was performed using Amira® software (Visage Imaging GmbH, Richmond, Australia, Fig. 1). This software was validated previously in an experimental study [25] and is now used routinely by radiologists in our hospital. Total liver volume and tumor volume measurements were done manually by the delineation of margins in every CT slide by an experienced radiologist in collaboration with the attending liver surgeon. Since tumor volume is regarded as a non-functional tissue, total functional liver volume was calculated as follows:

$${\text{Total}}\;{\text{functional}}\;{\text{liver}}\;{\text{volume}} = {\text{total}}\;{\text{liver}}\;{\text{volume}} - {\text{tumor}}\;{\text{volume}}$$

Resected liver volume was measured by an intraoperative water displacement method. The measured compound volume was multiplied by a factor of 1.15 to compensate the blood volume [25]. The future liver remnant volume (FLR-V) was thus calculated as follows:

$${\text{FLR}} - {\text{V}} = {\text{total}}\;{\text{functional}}\;{\text{liver}}\;{\text{volume}} - {\text{resected}}\;{\text{liver}}\;{\text{volume}} \times 1. 1 5$$
Fig. 1
figure 1

Computed tomography (CT) scans before (a, c) and after (b, d) portal vein embolization (PVE). Volumetric analysis was performed manually by a trained physician (supervised by a second physician), who delineated the margins in every CT slide with a handheld cursor. The volumes were calculated automatically according to the delineations and slice thickness. Arrows show the right portal branches before PVE (a) and the non-perfused areas after PVE (b). Grown liver segments II and III as well as a hyper-perfused left portal branch can be seen in picture ‘d’

Full size image

Calculation of FLR-F

Future liver remnant function (FLR-F) was calculated based on a FLR-V as:

$${\text{FLR}} - {\text{F}} = \frac{{{\text{FLR}} - {\text{V}}}}{{{\text{total}}\;{\text{functional}}\;{\text{liver}}\;{\text{volume}}}} \times {\text{LiMAx}}$$

Additionally, the function of liver segments II and III was calculated according to the volume of those segments as:

$$\begin{aligned} &{\text{Segment II}} + {\text{III}}\;{\text{function}} \\ &\quad = \frac{{{\text{Segment II}} + {\text{III}}\;{\text{volume}}}}{{{\text{Total}}\;{\text{functional}}\;{\text{liver}}\;{\text{volume}}}} \times {\text{LiMAx}} \end{aligned}$$

Statistical analyses

Continuous quantitative variables are expressed as means ± 95 % confidence interval if not noted otherwise. Group comparisons were performed with the Chi-quadrant, unpaired student t test, the Mann–Whitney U test, or the Wilcoxon test, in accordance to date scale and distribution. Moreover, an ROC analysis was performed to assess the predictive value of preoperative parameters (area under the curve) for liver-related death. Statistical significance was accepted at a p value of <0.05. Calculations were performed with SPSS 19.0 (SPSS Inc., Chicago, IL, USA).

Results

Thirty-one patients underwent extended right hepatectomy for primary or secondary liver malignancies. Table 1 provides detailed demographical and clinical data about the cohort. A group of 19 patients (59 %) received PVE prior to the operation, whereas 12 patients (31 %) underwent surgery without prior embolization (NO-PVE). There were 12 male patients in the PVE group (63 %) and 7 (58 %) in the NO-PVE group (male vs. female p = 0.043), All of the PVE patients and five of the NO-PVE patients had a Klatskin tumor, while two in the NO-PVE group had cholangiocarcinoma (17 %) and three (25 %) had metastases of colon adenocarcinoma (p = 0.019). Liver cirrhosis (F4) was found in the pathological workup of specimens from the two of the NO-PVE group patients, both of whom had a positive alcohol anamnesis (Table 1). None of the patients in our cohort had viral hepatitis positivity. There were no other significant differences in demographical, clinical or laboratory data between the study groups. Data regarding portal resection and the duration of hepatectomy did not differ between the groups (Table 1). Although liver-related postoperative complications such as bile leakage and bilioma tended to be more common in the PVE group, the difference was not significant. The mortality rates were 17 % in the NO-PVE and 12 % in the PVE group (p = 0.630). In the NO-PVE group, two patients died of portal vein thrombosis followed by liver failure. In the PVE group one patient died of intestinal perforation with peritonitis. Another patient in the PVE group had leakage of the biliary-enteric anastomosis, resulting in peritonitis and associated intestinal perforations. Overall, the three deaths were classified as “liver related”. Table 4 summarizes the clinical characteristic of the patients who died. An ROC analysis of the liver-related deaths and the associated parameters; namely, FLR-V, FLR-F, and LiMAx on POD 1 was also carried out (Fig. 2).

Fig. 2
figure 2

Predictive values for liver related death. Four patients died: three, of liver failure (liver-related death (Table 4). The area under the curve (AUC) from the receiver operating curve was calculated for the FLR-V (% of total functional liver volume) and FLR-F (µg/KG/h) before resection, as well as for the LiMAx on POD 1 (µg/KG/h) for liver-related death

Full size image
Table 1 Characteristics of the patients undergoing extended right hepatectomy
Full size table

The volume of liver segments II and III increased from 17.3 to 23.6 % after PVE. The measured liver-remnant volume after surgery was 34.7 % in the NO-PVE group and 36.6 % in the PVE group, without a significant difference. Table 2 summarizes further data on the liver volume measurements.

Table 2 Liver volume parameters
Full size table

Table 3 lists the liver function parameters. The initial LiMAx value was within the normal range [>315 (µg/kg/h)] and did not differ between the groups [344 vs. 401 (µg/kg/h), p = 0.326]. The initial FLR-F before PVE also did not differ between the groups. However, four weeks after PVE, the FLR-F increased to 141 µg/kg/h, with subsequent outbalance of the PVE group over the NO-PVE group (141 vs. 114 µg/kg/h, respectively; p = 0.023). Liver function measured on POD 1 did not differ significantly between the groups [NO-PVE 86 vs. PVE 95 (µg/kg/min), p = 0.215]. An overestimation of the planned FLR-F was 32 and 45 (µg/kg/min) for the NO-PVE and PVE groups, respectively (p = 0.391). The FLR-F and LiMAx values on POD 1 correlated significantly (Fig. 3) in the PVE (r = 0.585, p = 0.011) and NO-PVE groups (r = 0.608, p = 0.036). The postoperative regeneration of liver function after resection, quantified by the LiMAx test in both groups was similar. The LiMAx value was higher in the NO-PVE group only on POD 10 (p = 0.05, Fig. 4).

Table 3 Liver function parameters
Full size table
Fig. 3
figure 3

Correlation between predicted and measured liver function on POD 1. Future liver remnant function (FLR-F) can be predicted before resection by using CT-scan volumetric analysis and liver function assessment with the LiMAx test. FLR-F and LiMAx on POD 1 correlated significantly (r = 629, p < 0.001). The FLR-F was slightly overestimated in this clinical trial (adjusted line). This might be based on an additional intraoperative liver function loss caused by vascular clamping

Full size image
Fig. 4
figure 4

Liver regeneration after PVE and extended right hepatectomy. Liver regeneration was assessed by the LiMAx test. Liver function developed similarly after partial liver resection in both groups. On postoperative day 10, liver function in the NO-PVE group appeared to be higher, although this difference had diminished by 3 months postoperatively

Full size image

Discussion

Insufficient remnant liver function after hepatectomy leads to a posthepatectomy liver failure (PHLF), which increases the risk of postoperative mortality. We conducted the present study to predict FLR-F, using volume-function analysis by combining the LiMAx liver function test with CT/MRT liver volumetric analysis.

FLR-F is a crucial predictor of post-hepatectomy complication and mortality rates [26]. Different approaches, such as 99mTc-GSA scintigraphy, have been used to assess FLR-F in patients undergoing extended hepatectomy with or without prior PVE, but none has proven effective and limited data have been published [20, 27, 28]. One of the reasons for this might be that nuclear imaging techniques are cost- and facility-intensive. A volume function analysis using the LiMAx test and CT/MRT volumetric analysis was published recently by our group. We found an excellent correlation between planed FLR-F and morbidity and mortality (Table 2, page 142 [5]). Moreover, a high predictive power of the LiMAx test on POD 1 for liver failure and liver-related death was shown (AUC 0.99 and 0.99 respectively; Table 4, page 125 [4]). The findings of the present study are consistent with those of the previous reports and they also present the feasibility of volume function analysis to predict FLR-F in patients undergoing preoperative PVE.

Table 4 Characteristics of the patients who died after hepatectomy
Full size table

An important parameter influencing the FLR-F calculation is intraoperative liver injury. After intraoperative hepatic pedicle clamping with ensuing liver ischemia, an unknown liver function loss may occur [29]. Another disturbing factor is the presence of non- or hypo-perfused liver areas after hepatectomy. In the present series, all Klatskin tumors were resected according to the no-touch technique by Neuhaus et al. [30], which involves portal vein resection, accompanied by some liver ischemia. Assessing the hypo-perfused areas is difficult and requires a CT\MRT scan directly after hepatectomy, which was not performed in our study. Thus, the feasibility of volume function planning to predict the FLR-F is limited in patients undergoing complex hepatectomy. An over-calculation of FLR-F with a respective safety range is needed for those patients.

Another interesting finding of this study relates to the liver function distribution after PVE. Since blood supply to the right liver lobe is depleted after PVE, a loss of function may result; however, our data showed the opposite. First, we did not see any loss of function directly after PVE (Table 3; Fig. 4). Thus, we believe that loss of portal vein blood does not lead to an acute function alteration, as long as the arterial perfusion is preserved. Through a hepatic artery buffer response an arterial hyper-perfusion takes place in those portal hypo-perfused areas [31]. This hypothesis is supported by the fact that PVE, in contrast to arterial embolization, does not lead to necrosis or to a loss of hepatocyte mass. Second, the underestimation of the FLR-F did not differ between the PVE and NO-PVE groups in this study. This finding also suggested similar distribution of liver function in patients with and those without PVE. We assume that the shift in liver function from the right to the left lobe is a slow remodeling process, which includes proliferation and atrophy of the right liver lobe. Thus, the liver function distribution does not change at each time point after PVE and the data presented here support this hypothesis.

We observed an insignificant increase in the LiMAx value, from 360 to 401 µg/kg/h, between PVE and hepatectomy. This slight increase was supposedly due to improved therapy for cholestasis and cholangitis, which are known to impair liver function [32, 33]. In our department, biliary stenting is usually performed directly before PVE in patients with central cholangiocarcinoma. In the present series we noted a decrease in the total bilirubin serum level from 5.3 (2:7) (µg/dl) before PVE to 2.1 (1:3) (µg/dl) before hepatectomy (p < 0.001).

We presented data from a single-center prospective observational study of patients undergoing standard treatment for liver malignancy in our department. Every patient who fulfilled inclusion criteria was evaluated so there was no bias of patient selection criteria in relation to initial liver function. Portal vein embolization (PVE) was performed by the same method in all patients, thereby eliminating bias related to this issue in our trial [7]. However, this study has several limitations which need to be addressed in further research studies. First, the limited number of patients and too few “hard” end points such as liver failure-related death make it impossible to perform logistic regression in this selected population. Furthermore, none of the patients in this clinical trial had a history of liver diseases, such as hepatitis or cirrhosis, or any marked comorbidity impairing global liver function, as well as FLR-F. On the other hand, the LiMAx could be carried out easily in a common clinical situation, by using a mobile device. In our experience, the LiMAx test might even be used in patients with acute cholestasis, which is known to affect the results of other liver function tests, such as the ICG-PDR [34].

Conclusions

The LiMAx test combined with CT-assisted volume/function analysis allows for a reliable calculation of the future liver remnant function (FLR-F) directly before extended right hepatectomy, or after portal vein embolization (PVE). An overestimation of the FLR-F with a respective safety range is needed for complex hepatectomy to compensate for the intraoperative liver injury caused by vascular clamping or the presence of non-functional (hypoperfused) liver areas. The prediction of early postoperative liver function by the preoperative calculation of FLR-F could be used to stratify the risk of postoperative liver failure. Liver function distribution does not change after PVE showing a slow and homogeneous remodeling process.