Intraoperative simulation of remnant liver function during anatomic liver resection with indocyanine green clearance (LiMON) measurements

Abstract

Objective

Post-hepatectomy liver failure (PHLF) is the major cause of death following liver resection. The aim of this study was to evaluate the feasibility of an intraoperative simulation of post-resection liver function.

Methods

Intraoperative liver function was measured by indocyanine green (ICG) clearance using the LiMON™ technology. In 20 patients undergoing anatomic liver resection, ICG plasma disappearance rate (PDR (%/min) and ICG retention at 15 min (R₁₅) (%) were measured immediately after the induction of anaesthesia (t₀), after selective arterial and portovenous inflow trial clamping (TC) of the resected liver segments (t₁), after the completion of resection (t₂) and before the closure of the abdominal cavity (t₃).

Results

The median baseline (t₀) PDR was 16.5%/min. Trial clamping of the inflow (t₁) resulted in a significant reduction in PDR to 10.5%/min. Results under TC were similar to those obtained after resection (t₂) (median PDR: 10.5%/min). Linear regression modelling showed that post-resection liver volume could be accurately predicted by TC of liver inflow (P < 0.0001), but not by determining the resected liver volume. Simulated post-resection liver function under TC correlated well with PHLF and length of hospital stay.

Conclusions

Intraoperative ICG clearance measurements allow real-time monitoring of intraoperative liver function during surgery. Trial clamping of arterial and portovenous inflow accurately predicts immediate post-resection liver function. The intraoperative measurement of liver function and simulation of post-resection liver function may help to avoid PHLF.

Introduction

Although major liver resection has become safer as a result of advances in surgical techniques and perioperative management, the resection of a large amount of functional liver tissue continues to bear risk for post-hepatectomy liver failure (PHLF).^1,2 In individuals with normal liver parenchyma, the remnant functional liver mass can be as low as 20–25%. Pre-existing liver diseases (steatosis, fibrosis and cirrhosis) and intensified preoperative chemotherapy significantly reduce this functional liver reserve,³ even to the point at which liver resection seems impossible.

In order to prevent the occurrence of PHLF, various methods of determining liver function have been established, all of which measure the function of the entire liver preoperatively.^4–6 However, the surgeon's interest refers to the function of the remnant liver after the resection, rather than the function of liver that is to be resected. Therefore, predicting the function of the future liver remnant is important and all the more so because function cannot be assumed to be distributed homogeneously throughout the liver parenchyma.^7,8 Depending on entity, size and grade of differentiation, tumours may or may not contribute to liver function.^9,10 Tumours may also change the perfusion of healthy liver parenchyma by diverting blood flow and disturb biliary excretion by local compression.¹¹ Finally, liver resections along anatomic resection lines may differentially affect remnant liver function according to individual liver anatomy.⁹

To account for the inaccuracies of conventional methods for the preoperative determination of liver function, the present group has established intraoperative indocyanine green (ICG) clearance measurements for real-time measurements of liver function. This paper reports the results of tests to establish whether liver function under trial clamping (TC) of arterial and portovenous inflow predicts PHLF, thus providing a tool to guide and tailor liver resections ex ante.

Materials and methods

Patients

This pilot study prospectively included 20 patients undergoing anatomic liver resection (two or more segments) between February 2013 and February 2014. Another prerequisite for inclusion required the selective clamping of the arterial and portovenous inflow to be possible.

Surgery and intraoperative ICG pulse spectrophotometry

Liver resection was carried out using a standard open or laparoscopic-assisted ultrasound dissection technique along the anatomic borders defined by Couinaud's liver segments. The hepatic ligament was dissected to allow for the identification and selective clamping of the arterial and portovenous inflow of the liver segments scheduled for resection. Resection was performed without a Pringle manoeuvre.

Indocyanine green clearance was determined intraoperatively by non-invasive pulse spectrophotometry (LiMON™; Pulsion Medical Systems AG, Munich, Germany). For each measurement, an aqueous solution of ICG 0.25 mg/kg body weight was injected i.v. Baseline liver function was obtained after the induction of general anaesthesia (t₀). The next measurement was obtained under inflow TC in order to simulate the situation after resection (t₁). Post-resection liver function was measured after the completion of resection prior to the closure of the abdominal incision (t₂) (Fig. 1).

Figure 1.

Trial clamping in a right hemi-hepatectomy. This schematic drawing shows the inflow clamping of the right portal vein (pv) and the right hepatic artery (ha) in preparation for an intended right hepatectomy with a resection line slightly right of the middle hepatic vein. Indocyanine green clearance was determined prior to resection (t₀), under trial clamping (t₁) and after resection (t₂)

Study endpoints

Intraoperative ICG measurements of liver function were expressed as the plasma disappearance rate (PDR) (%/min) and the rate of retention after 15 min (R₁₅). Other markers of liver function, such as the international normalized ratio (INR), serum bilirubin (‘50–50 criteria’¹²) and the laboratory Model for End-stage Liver Disease (labMELD) score¹³ were recorded prior to surgery and on postoperative days (PoD) 3 and 5. Post-hepatectomy liver failure was defined as an increased INR with concomitant hyperbilirubinaemia from PoD 5 onwards as suggested by the International Study Group of Liver Surgery (ISGLS).¹³

Volumetric data were calculated from preoperative computed tomography (CT) scans. Preoperative total liver volume was measured using the volumetry function of the open-source software OsiriX® (Pixmeo SARL, Geneva, Switzerland). Resection volumes were measured by water displacement. To compensate for the lower volume of exsanguinated liver tissue, 13% was added as suggested by Niehues et al.¹⁴ The remnant liver volume was calculated by subtracting the resected volume from the total liver volume.

Patient characteristics, operative details, laboratory parameters and data on the postoperative course were extracted from electronic patient charts and documented.

Statistical analyses

Unless otherwise indicated, all values are presented as the median (range). For statistical analyses and graphics, GraphPad Prism Version 6.0 (GraphPad Software, Inc., La Jolla, CA, USA) was used. Differences in continuous parameters were determined by Student's t-tests. Linear regression analysis was used to determine correlations between liver volume and function. A P-value of <0.05 was considered to indicate statistical significance.

Results

Patient characteristics

Twenty patients with primary and secondary liver tumours were included. Patient characteristics are detailed in Table 1. The severity of pre-existing liver disease was extracted from the postoperative pathological report of the resected liver parenchyma and is summarized in Table S1 (online). Haemodynamic differences, as a relevant source of changes in intraoperative ICG measurements, could be excluded. There were no clinically relevant differences in mean arterial pressure or central venous pressure between t₁ and t₃.

Table 1.

Characteristics of patients submitted to liver resection (n = 20)

Characteristic	Value
Male/female, n	10/10
Age, years, median (range)	60 (23–84)
Disease, n
Hepatocellular carcinoma	5
Colorectal cancer metastasis	3
Cholangiocellular carcinoma	4
Other metastasis	4
Benign lesions	4
Hepatitis B, n	2
Hepatitis C, n	1
Non-hepatitis B/non-hepatitis C, n	17
Operation, n
Right hemi-hepatectomy	14
Extended right hemi-hepatectomy	3
Left hemi-hepatectomy	1
Left lateral segmentectomy	2
Total liver volume, ml, median (range)	1702 (1290–3473)
Resected liver volume, ml, median (range)	840 (213–2500)
Remnant liver volume, ml, median (range)	940 (610–1873)
Postoperative hospital stay, days, median (range)	14 (6–129)
Preoperative MELD score, median (range)	7 (6–15)
Postoperative MELD score at PoD 3, median (range)	10 (7–22)
Postoperative MELD score at PoD 5, median (range)	8 (6–23)

MELD, Model for End-stage Liver Disease; PoD, postoperative day.

Intraoperative ICG pulse spectrophotometry

The median PDR at t₀ was 16.5%/min; values ranged from 9.9%/min to 35.1%/min, reflecting the actual heterogeneity of baseline liver function. In most patients TC (t₁) resulted in a significant reduction in PDR to a median of 10.5%/min (range: 4.7–20.0%/min). The median value obtained after resection at t₂ was 10.5%/min (range: 4.7–20.0%/min) and thus was similar to that obtained under TC. Inversely, median R₁₅ was 8.2% (range: 0.5–22.7%) at baseline, increased to 20.7% (range: 5.0–49.5%) at t₁ and subsequently to 27.7% (range: 6.5–32.0%) at t₂. Results are summarized in Fig. 2. Individual measurements are shown in Table S1 (online).

Figure 2.

Intraoperative simulation of remnant liver function [mean ± standard error of the mean (SEM)] indocyanine green (ICG) plasma disappearance rate (PDR) and ICG clearance rate at 15 min (R₁₅)

Trial clamping accurately predicted liver function after resection as determined by linear regression modelling (P < 0.0001) of PDR values measured under TC and after resection (Fig. 3a). No correlation was found for preoperative measurements of total PDR and post-resection PDR (Fig. 3b).

Figure 3.

Correlations between (a) indocyanine green (ICG) plasma disappearance rate (PDR) measured preoperatively and post-resection (r = 0.3589, r² = 0.1288, P = 0.1202), (b) ICG PDR at clamping and at post-resection (r = 0.8045, r² = 0.6473, P < 0.0001), and (c) the percentage of ICG PDR at t₂/t₀ and the percentage of remnant/total liver volume (r = 0.4797, r² = 0.2301, P = 0.0514)

Further, measured liver function after resection did not correlate with calculated post-resection liver function as determined by the percentage of liver resected and total liver function prior to resection (Fig. 3c). Thus, the determination of post-resection liver function by combining data for preoperative liver function and volumetric assessment of resection planes is highly inaccurate.

Postoperative course

Patients who experienced PHLF (n = 6) according to the definition of the ISGLS¹³ showed significantly higher R₁₅ values under TC than did patients without PHLF (Fig. 4). Consequently, patients with PHLF also had a significantly longer hospital stay than patients without PHLF (Fig. 5).

Figure 4.

Intraoperative indocyanine green clearance rate at 15 min (R₁₅) (t₁) in patients with and without post-hepatectomy liver failure (PHLF)

Figure 5.

Postoperative hospital stay in patients with and without post-hepatectomy liver failure (PHLF)

Surgical complications of Clavien–Dindo Grade IIIb or greater occurred in three patients. Two patients died in the postoperative course, one (Patient 9) as a result of the rapid tumour progress of a fibrolammelar hepatocellular carcinoma on PoD 129 and another (Patient 5) secondary to PHLF on PoD 21.

One patient (Patient 3) required an operative thrombectomy and revision of the portal vein on PoD 3.

Discussion

Liver resection provides the chance for cure in patients with early-stage primary liver tumours and metastatic colorectal cancer confined to the liver. In order to achieve the goal of a radical resection with tumour-free margins, liver resection up to the limit of the functional liver reserve is justified. Unfortunately, regenerative capacity is not fixed, but depends on various factors including haemodynamics, parenchymal integrity and the functional reserve of the future liver remnant.¹⁵

Currently, decisions on whether or not to perform the required resection rely on the preoperative assessment of full liver function and on the intraoperative judgement of an experienced hepatobilary surgeon. However, postoperative liver failure still occurs in 5–8% of patients.

A major source of misjudgement derives from the fact that postoperative remnant liver function cannot be extrapolated by preoperative liver function tests. There are several reasons for this. Firstly, liver function may already be compromised below the detection limit of conventional markers. The half-life of liver function markers used in routine clinical practice is too long to support intraoperative real-time measurements of liver function. Secondly, liver function is influenced by haemodynamics in the remnant liver tissue. Liver resection may also compromise liver in- and outflow distant of the resection line; large tumours in particular can divert blood flow significantly. Thirdly, liver volume may not necessarily correlate with function.

The present study suggests that, in individual patients, the intraoperative measurement of hepatic function with ICG spectrophotometry adds an objective intraoperative real-time parameter to facilitate decision making.¹⁶ Under TC of arterial and portal venous inflow, measured liver function closely resembles liver function after resection. Trial clamping resulted in a significant decline in PDR and an increase in R₁₅ values. Values under TC and after resection were very similar. Liver function under TC was clinically relevant because changes corresponded well with changes between pre- and postoperative labMELD scores on PoD 3, the occurrence of PHLF and hospital length of stay.

Measured liver function after TC closely reflected liver function after resection. The linear regression analysis revealed a clear numerical correlation between liver function prior to resection under TC and liver function after resection. By contrast, the correlation between liver function extrapolated from CT volumetry and actual PDR measured in the remnant liver was poor. Refinement of volumetric measurement by various correction factors, such as by subtracting tumour volume in some tumour entities, may increase accuracy, but may also fail to resemble liver function. Therefore, the extrapolation of volumetric full liver data is inferior to direct measurement under TC.

This method is limited by its restriction to anatomic resections and the underestimation of liver function in patients with severe bilary obstruction.⁵ In addition, the present study is limited by the low number of patients included. In order to generate sufficient data to determine a clear cut-off value, an appropriately powered prospective study should be conducted.

To date, no clear cut-off values have been established. Cheung et al. suggest a preoperative cut-off R₁₅ value of 14% for safe resections in patients with Child–Pugh class A cirrhosis.¹⁷ de Liguori Carino et al. established a preoperative PDR cut-off of 17.6%/min.¹⁸ In their population, the median PDR value on PoD 1 in patients who subsequently developed liver dysfunction was 6.7%/min (interquartile range: 6.3–10.4%/min), suggesting that a PDR value of <10%/min may indicate a borderline liver reserve. After liver transplantation, a PDR value of <12.85%/min at any time between PoD 0 and PoD 5 was shown to be predictive of early severe complications, primary non-function of the graft or acute rejection.¹⁹ Schneider et al. suggest a PDR cut-off value of 9.6%/min on PoD 7 predicts death and graft loss in liver transplant patients.²⁰ Hori et al. established a cut-off ratio of 3.1175 for ICG to graft weight for an adequate donor cell mass to avoid small-for-size grafts in living donor liver transplantation.²¹

The measurement of liver function under TC offers additional information to the hepatobilary surgeon who must decide whether to adapt the surgical strategy. In addition to the option to abort the procedure, the surgeon may decide to use a parenchyma-sparing non-anatomic resection, convert to a two-stage procedure or use an in situ split technique.

Thus, this simple method may allow the intraoperative reassessment of liver function under realistic conditions and may facilitate decision making in patients with critically small liver remnants and impaired preoperative liver function.

Conflicts of interest

None declared.

Supporting Information

Table S1 Liver function test before surgery, intraoperatively at t₀ (preclamping, total liver), t₁ (trial clamping, remnant liver), t₂ (after resection, actual remnant liver) and on postoperative days 3 and 5.