- Email: [email protected]
ALPlat criterion for the resection of hepatocellular carcinoma based on a predictive model of posthepatectomy liver failure
Surgery xxx (2019) 1e7
Contents lists available at ScienceDirect
Surgery journal homepage: www.elsevier.com/locate/surg
ALPlat criterion for the resection of hepatocellular carcinoma based on a predictive model of posthepatectomy liver failure Gen Yamamoto, MD, PhDa, Kojiro Taura, MD, PhDa,*, Iwao Ikai, MD, PhDb, Takahisa Fujikawa, MD, PhDc, Ryuta Nishitai, MD, PhDd, Satoshi Kaihara, MD, PhDe, Masazumi Zaima, MD, PhDf, Hiroaki Terajima, MD, PhDg, Tsunehiro Yoshimura, MD, PhDh, Yukinori Koyama, MD, PhDi, Kazutaka Tanabe, MD, PhDa, Takahiro Nishio, MD, PhDa, Yukihiro Okuda, MD, PhDa, Yoshinobu Ikeno, MD, PhDa, Kenji Yoshino, MD, PhDa, Keita Fukuyama, MDa, Satoru Seo, MD, PhDa, Etsuro Hatano, MD, PhDj, Shinji Uemoto, MD, PhDa a
Department of Surgery, Kyoto University Graduate School of Medicine, Kyoto, Japan Department of Surgery, National Hospital Organization Kyoto Medical Centre, Kyoto, Japan Department of Surgery, Kokura Memorial Hospital, Kitakyushu, Japan d Department of Surgery, Kyoto-Katsura Hospital, Kyoto, Japan e Department of Surgery, Kobe City Medical Centre General Hospital, Kobe, Japan f Department of Surgery, Shiga General Hospital, Moriyama, Japan g Department of Surgery, Tazuke Kofukai Medical Research Institute Kitano Hospital, Osaka, Japan h Department of Surgery, Tenriyorodusoudanjyo Hospital, Tenri, Japan i Department of Surgery, Shimane Prefectural Central Hospital, Izumo, Japan j Department of Surgery, Hyogo College of Medicine, Nishinomiya, Japan b c
a r t i c l e i n f o
a b s t r a c t
Article history: Accepted 26 September 2019 Available online xxx
Background: The indocyanine green test is used widely to evaluate the risk of posthepatectomy liver failure for hepatocellular carcinoma. A more convenient and reliable scoring system is desired owing to limited accuracy and availability of the indocyanine green test. This study aimed to establish a new selection criterion for liver resection in HCC. Methods: We reviewed retrospectively 876 patients undergoing a partial hepatectomy for hepatocellular carcinoma between 2007 and 2015 in 8 affiliated hospitals. Posthepatectomy liver failure grades B and C were regarded as posthepatectomy liver failure. We identified the risk factors for posthepatectomy liver failure and established a predictive model based on a formula for the probability of posthepatectomy liver failure. External validation was performed in an additional cohort of 250 patients. Results: Posthepatectomy liver failure occurred in 92 patients (11%). The area under the receiver operating characteristic curve for the prediction of posthepatectomy liver failure was 0.646 for the platelet count, 0.641 for albumin, 0.623 for the percentage of liver remnant, and 0.607 for the plasma disappearance rate of indocyanine green. Logistic regression analysis provided a formula for the probability of posthepatectomy liver failure consisting of platelet count, albumin, and liver remnant. We defined platelet count þ 90 albumin as the ALPlat index and established an ALPlat-based criterion for operative resection that secured the same risk assumed by the indocyanine greenebased criterion (Makuuchi’s criterion). This criterion exhibited a greater sensitivity and specificity than the indocyanine greenebased criterion in the validation cohort. Conclusion: The ALPlat criterion is a simple and useful method to assess liver function and to make therapeutic decisions in patients with hepatocellular carcinoma. © 2019 Elsevier Inc. All rights reserved.
* Reprint requests: Kojiro Taura, MD, PhD, Department of Surgery, Kyoto University Graduate School of Medicine, 54 Kawahara-cho, Shogoin, Sakyo-ku, Kyoto, 606-8507, Japan. E-mail address: [email protected] (K. Taura). https://doi.org/10.1016/j.surg.2019.09.021 0039-6060/© 2019 Elsevier Inc. All rights reserved.
2
G. Yamamoto et al. / Surgery xxx (2019) 1e7
Introduction Hepatocellular carcinoma (HCC) is the second leading cause of cancer mortality worldwide in men and the sixth in women.1 There are a variety of treatment options for HCC, such as liver resection, liver transplantation, radiofrequency ablation or other ablative techniques, transcatheter arterial chemoembolization, percutaneous ethanol injection therapy, molecular-targeted therapy, and radiation therapy.2 Liver resection is an effective therapeutic option. The outcomes after liver resection have substantially improved due to the refined surgical technique, perioperative care, and improvements in patient selection.3 Nevertheless, posthepatectomy liver failure (PHLF) occurs in a non-negligible number of patients and remains a leading cause of postoperative mortality in HCC patients.4e6 The incidence of PHLF varies anywhere from 9% to 40% and depends on the severity of the underlying liver disease.6e14 To decrease the incidence of PHLF, careful patient selection based on a close assessment of liver function is of paramount importance.15 Accurate prediction of PHLF before liver resection, however, is sometimes difficult because multiple factors are associated with PHLF, including the preoperative liver function, volume of the liver remnant, blood loss, operative time, and postoperative events, such as infection and unstable hemodynamic status. The results of laboratory tests14,16e23 and quantitative liver function tests, such as the indocyanine green (ICG) retention test,11,24e26 volume of the liver remnant,9,27e29 and non-invasive liver stiffness measurement,30e32 are useful parameters for predicting PHLF. The ICG retention test is used widely as an algorithm for a safe operative intervention for HCC.15,33 Makuuchi’s criterion, which determines permissible liver resection procedures based on the ICG retention test, has been the gold standard for safe hepatectomies in many Asian countries.33 Our previous studies21,32 and others,14,23 however, demonstrated some problems with the efficacy of the ICG test as a predictor for PHLF in several situations. These findings were based on data from a single institution, and further research involving larger patient numbers is needed for verification. Moreover, the ICG test is not performed routinely in daily clinical practice. For these reasons, physicians feel that this method can be inconvenient for making treatment decisions. The establishment of resection criterion solely based on routine laboratory tests would be highly desirable. This study aimed to reappraise the usefulness of common laboratory tests in predicting PHLF in a larger cohort and to establish an easily calculated criterion for patient selection for liver resection by generating a predictive formula for PHLF that incorporates only routine laboratory tests. Methods Patients We reviewed retrospectively 876 patients undergoing a partial hepatectomy for HCC in our department and 7 affiliated hospitals of Kyoto University between 2007 and 2015 (National Hospital Organization Kyoto Medical Centre: 2009e2015, Kokura Memorial Hospital: 2007e2013, Kyoto-Katsura Hospital: 2012e2015, Kobe City Medical Centre General Hospital: 2011e2015, Shiga General Hospital: 2011e2015, Tazuke Kofukai Medical Research Institute Kitano Hospital: 2007e2015, Tenriyorodusoudanjyo Hospital: 2011e2015, and Kyoto University Hospital: 2007e2011). The exclusion criteria included patients undergoing combined resection of an extrahepatic bile duct and patients undergoing preoperative portal vein embolization. This study was performed in accordance with the ethical guidelines for epidemiologic research in Japan and was approved by the Ethics Committee of the Kyoto University
Graduate School of Medicine (approval code: E2056) and separately by each hospital. For external validation of the criterion established with the aforementioned 876 cases, 250 additional HCC patients undergoing a partial hepatectomy in our department between 2011 and 2016 were reviewed subsequently. Data collection Demographic data, including sex, age, and status of viral infection, were collected for all patients. We ran routine preoperative laboratory tests, including the platelet count (Plt), international normalized ratio of prothrombin time (PT-INR), and serum levels of total bilirubin (T-bil), aspartate transaminase (AST), alanine transaminase (ALT), and albumin (Alb). A preoperative ICG test was performed in all patients. After intravenous injection of indocyanine green 0.5 mg/kg (Diagnogreen; Daiichi Pharmaceutical Co, Inc, Tokyo, Japan), the indocyanine green disappearance rate (KICG) was calculated by linear regression from the plasma concentrations of indocyanine green at 5, 10, and 15 minutes. The operative factors, including the operation time, estimated blood loss, requirement of transfusion, operative procedure (resected segments and the use of laparoscopy), and concomitant resection of other organs, were recorded. We estimated the liver remnant (Rem,%) according to the operative procedures, with reference to a previous study34 on the volumes and proportions of each segment for all patients. For nonanatomic resections, the Rem was estimated using the tumor size and weight of the specimen. PHLF was defined using the International Study Group of Liver Surgery criteria.5 In brief, when both T-bil and PT-INR were abnormal on or after postoperative day 5 according to the normal cutoff levels defined by the local hospital, patients were diagnosed with PHLF. PHLF patients were then categorized into grades A, B, or C based on the required treatment. The primary endpoint of this study was the incidence of PHLF grades B and C as defined by the International Study Group of Liver Surgery. Hepatectomy procedures Hepatectomy is the standard treatment for HCC whenever tumors are considered to be resectable, and the patient’s condition permits the operation. The indications for the resection of HCC include a Child-Pugh status of A or B and a future Rem of greater than 30% of the entire liver, but modifications by each institution were allowed. Intraoperative ultrasonography was performed to identify any occult tumors missed by preoperative imaging and to confirm the relative positions between tumors and vascular structures. Transection of the liver parenchyma was performed using a Cavitron Ultrasonic Surgical Aspirator (Valleylab, Inc, Boulder, CO) and bipolar cautery with a saline irrigation system. For laparoscopic liver resections, a combination of the Cavitron Ultrasonic Surgical Aspirator, THUNDERBEAT (Olympus Corporation, Tokyo, Japan), and BiClamp (ERBE Elektromedizin GmbH, Tubeingen, Germany) devices was used for liver transection. The Pringle maneuver was applied as indicated, with 15 minutes of occlusion alternating with 5 minutes of reperfusion. Statistical analysis Categorical variables are described as frequencies and percentages and were compared using the c2 or Fisher exact test as appropriate. Continuous variables are described as the mean ± standard deviation or the median (range) and were compared using the Mann-Whitney U test. Univariate analyses of risk factors for PHLF were performed using Fisher exact test. Multivariate analyses
G. Yamamoto et al. / Surgery xxx (2019) 1e7
3
Table I Patient characteristics and the comparison between patients with and without PHLF
Sex (female) Age (y) Underlying liver disease HBs Ag positive Anti-HCV Ab positive Non-B non-C positive Plt (109/L) PT-INR T-bil (mg/dL) AST (IU/L) ALT (IU/L) Alb (g/dL) KICG Rem (%) Operation time (min) Blood loss (g)
Total (N ¼ 876)
PHLF (þ) (n ¼ 92)
PHLF (-) (n ¼ 784)
P value
197 (23%) 69 ± 9
17 (19%) 69 ± 10
180 (23%) 69 ± 9
.320 .612
154 (18%) 367 (42%) 360 (41%) 158 ± 70 1.10 ± 0.15 0.81 ± 0.45 44 ± 30 40 ± 34 3.92 ± 0.46 0.144 ± 0.07 72 ± 21 342 (66e849) 502 (0e26,800)
12 (13%) 49 (53%) 32 (35%) 131 ± 59 1.17 ± 0.23 0.86 ± 0.33 53 ± 40 43 ± 30 3.73 ± 0.45 0.128 ± 0.05 63 ± 24 415 (124e822) 1,255 (38e26,800)
142 (18%) 318 (41%) 328 (42%) 161 ± 71 1.09 ± 0.14 0.81 ± 0.46 43 ± 28 40 ± 34 3.94 ± 0.46 0.146 ± 0.07 73 ± 20 337 (66e849) 460 (0e7,000)
.173 .025 .193 < .0001 .0002 .344 .005 .408 < .0001 .002 < .0001 < .0001 < .0001
Data represent the mean ± standard deviation or the median (range) or the number of patients. HB, hepatitis B; HC, hepatitis C; Non-B non-C, non-hepatitis B and non-hepatitis C.
of risk factors for PHLF were performed using the logistic regression model. Variables associated with the incidence of PHLF at P < .05 in the univariate analysis were included in a multivariate logistic regression analysis, and the odds ratios (OR) and 95% confidence intervals (CI) were calculated. A receiver operating characteristic (ROC) analysis was performed to evaluate the power of the predictors, and the area under the receiver operating characteristic curve (AUROC) and the 95% confidence intervals (CIs) were calculated. We established the criterion for liver resection based on a prediction formula developed by multivariate analysis of data from the 876 patients, and we compared this method with the ICG-based criterion (Makuuchi’s criterion). The criterion was then validated in an additional cohort of 250 patients. The calculation of sensitivity, specificity, positive predictive value, negative predictive value, positive likelihood ratio, negative likelihood ratio, and accuracy was performed based on the exact binary classification. All statistical evaluations were performed using JMP 13 software (SAS institute Inc, Cary, NC). All tests were 2-sided, and differences with P values less than .05 were considered statistically significant. Results Patient characteristics The characteristics of the study population are shown in Table I. There were 679 men and 197 women with a mean age of 69 years. The underlying liver diseases included chronic hepatitis B (positive for hepatitis B surface antigen; n ¼ 154, 18%), chronic hepatitis C (positive for anti-hepatitis C antibody; n ¼ 367, 42%), chronic hepatitis B and C (positive for both; n ¼ 5, 1%), and non-hepatitis B and non-hepatitis C (negative for both; n ¼ 360, 41%). PHLF developed more frequently in patients with chronic hepatitis C (P ¼ .025). Patient characteristics in the PHLF and non-PHLF groups PHLF occurred in 189 patients. Of these, 97 (11%) were classified as grade A, 77 (9%) as grade B, and 15 (2%) as grade C. We regarded PHLF grades B and C as PHLF because Grade A does not require intervention and is not clinically relevant. Table I shows the comparison between patients with and without PHLF. Statistically significant differences between PHLF and non-PHLF groups were observed for Plt, PT-INR, AST, Alb, KICG, and Rem.
Table II ROC analysis of preoperative factors for PHLF
Plt Alb PT-INR Rem AST KICG T-bil ALT
AUROC
95% CI
0.646 0.641 0.624 0.623 0.613 0.607 0.569 0.558
0.59e0.70 0.58e0.70 0.56e0.69 0.55e0.67 0.55e0.67 0.54e0.67 0.51e0.63 0.50e0.62
Table III ROC analysis of preoperative factors in combination with Rem for PHLF
Rem Plt Rem KICG Rem PT-INR Rem Alb Rem AST Rem T-bil Rem ALT
OR
95% CI
P value
AUROC
0.54 0.43 0.59 0.44 0.64 1.43 0.63 0.61 0.68 1.25 0.65 1.14 0.66 1.05
0.44e0.67 0.31e0.59 0.48e0.72 0.29e0.65 0.52e0.78 1.19e1.72 0.52e0.76 0.50e0.76 0.55e0.82 1.04e1.49 0.53e0.79 0.95e1.35 0.54e0.80 0.87e1.28
< .0001 < .0001 < .0001 < .0001 < .0001 .0001 < .0001 < .0001 < .0001 .0154 < .0001 .16 < .0001 .60
0.724 0.698 0.696 0.695 0.651 0.642 0.631
The adjusted ORs per standard deviation change are shown.
ROC analysis of preoperative factors for PHLF Table II shows the ROC analysis, where PHLF was predicted by a single factor. The AUROC of KICG was 0.607, which was less than the corresponding values of Plt (0.646), Alb (0.641), PT-INR (0.624), Rem (0.623), and AST (0.613). The best predictor for PHLF was Plt. We also analyzed the predictive capacity of laboratory factors in combination with Rem for PHLF (Table III). ROC analysis showed that the AUROC of Plt combined with Rem was 0.724, which was better than that of KICG with Rem (0.698) or any other single factor combined with Rem (PT-INR with Rem; 0.696, Alb with Rem; 0.695, AST with Rem; 0.651, T-bil with Rem; 0.642, and ALT with Rem; 0.631).
4
G. Yamamoto et al. / Surgery xxx (2019) 1e7
Multivariate analysis of preoperative factors for PHLF Multivariate logistic regression analysis was performed to identify the risk factors for PHLF. Plt (OR; 0.51, 95% CI; 0.37e0.70), PT-INR (OR; 1.27, 95% CI; 1.05e1.49), Alb (OR; 0.68, 95% CI; 0.53e0.85), and Rem (OR; 0.50, 95% CI; 0.39e0.62) were statistically significant independent preoperative predictors for PHLF (Table IV). ALPlat criterion for operative decision-making Logistic regression analysis yielded a formula for the probability of PHLF consisting of Rem, Plt, and Alb, which showed a lesser P value and served our purpose of establishing a simple index. The probability of PHLF was then calculated as follows:
resection) were given as 0.1475 and 0.1312, respectively, based on Makuuchi’s criterion. These results led to the generation of a criterion for a safe resection based on the ALPlat index (Table V), which we termed the ALPlat criterion. The minimum values of the ALPlat index that permitted a safe right hemihepatectomy, left hemihepatectomy, and segmentectomy were determined such that the risks for PHLF were equivalent to the permissible risks based on Makuuchi’s criterion. These values were 531 for right hemihepatectomy, 461 for left hemihepatectomy, and 423 for segmentectomy. ROC analysis showed that the AUROC of the RemALPlat index (0.753) was significantly greater than that of the RemKICG index (0.698) (Supplementary Table I). We also made a comparison between the RemALPlat index and the ALBI score (¼ [log10 bilirubin 0.66] þ [albumin e0.085], where bilirubin is in mmol/L and albumin in g/L) combined with
. h i Probability ðfor PHLFÞ ¼ 1 1 þ exp 0:0334 Rem ð%Þ þ 0:0113 Plt 109 =L þ 1:0154 Alb ðg=dLÞ 5:6797 h i i . h ¼ 1 1 þ exp 0:0334 Rem ð%Þ þ 0:0113 Plt 109 =L þ 90 Alb ½g=dL 5:6797 ¼1 1 þ exp 0:0334 Rem ð%Þ þ 0:0113 ALPlat index 5:6797 ¼ 1=ð1 þ exp ½RemALPlat index 5:6797Þ;
where Plt ð109 =LÞ þ 90 Alb ðg=dLÞ was defined as the ALPlat index and 0:0334 Rem ð%Þ þ 0:0113 ALPlat index was defined as the RemALPlat index. This formula allows the estimation of the probability of PHLF based on the ALPlat index and Rem, which should be useful for determining the safety and risk of PHLF for individual patients. We tried to establish a criterion that guaranteed equivalent safety to that of the ICG-based Makuuchi’s criterion. The probabilities of PHLF permitted by Makuuchi’s criterion were calculated. By logistic regression analysis,
Rem (defined as the RemALBI index) in the same manner. The AUROC of the RemALPlat index (0.753) was significantly greater than that of the RemALBI index (0.712) (Supplementary Table I). Figure 1, A shows the nomogram that determines the ALPlat index based on the Plt and Alb values. Figure 1, A also provides the probability of PHLF based on the ALPlat index and Rem, as well as the minimum requirement of Rem based on the ALPlat index and the acceptable probability of PHLF, which helps us estimate the surgical risk and determine an appropriate safe operative procedure. Figure 1, B demonstrates how the nomogram determines the probability of PHLF from the Plt, Alb, and Rem values. Figure 1, C
Probability ðfor PHLFÞ ¼ 1=ð1 þ exp ½0:0263 Rem ð%Þ þ 11:8358 KICG 1:2457Þ ¼ 1=ð1 þ exp ½RemKICG index 1:2457Þ;
where
0:0263 Rem ð%Þ þ 11:8358 KICG
was
defined
as
RemKICG index. In the case of right hemihepatectomy for example, substituting 66.7 (%) for Rem and 0.154 (because 10% of the ICG retention rate at 15 minutes corresponds to 0.154 of KICG) for KICG; this formula gave 0.1909 as the permissible risk for a safe right hemihepatectomy. In the same manner, the permissible risks for a safe left hemihepatectomy (33.3% resection) and segmentectomy (16.7%
Table IV Multivariate analysis of preoperative factors for PHLF
Plt PT-INR Alb Rem
shows how to estimate Rem from the values of Plt and Alb when the acceptable probability of PHLF is given. We have created an Excel file (Microsoft Corporation, Redmond, WA) that allows us to easily calculate the PHLF risk by simply entering the Plt, Alb, and Rem values (Supplementary File 1). It also provides a graph showing the relationship between Rem and PHLF when given the Plt and Alb values. A patient with Plt 150 (109/L) and Alb 3.5 (g/dL) is shown as an example.
Table V ALPlat criterion: Operative indication by the ALPlat index
OR
95% CI
P value
ALPlat index
0.51 1.27 0.68 0.50
0.37e0.70 1.05e1.49 0.53e0.85 0.39e0.62
< .0001 .011 .001 < .0001
>531 461e531 423e461 <423
Surgical procedures 2/3 resection 1/3 resection 1/6 resection <1/6 resection
Right hemihepatectomy, trisectionectomy Left hemihepatectomy, sectionectomy Segmentectomy Limited resection
G. Yamamoto et al. / Surgery xxx (2019) 1e7
5
Figure 1. (A) ALPlat nomogram. The nomogram can help to rapidly assess the ALPlat index. When the ALPlat index and Rem are connected by a straight line, the probability is obtained. (B) Example 1. Plt 200, Alb 4.0, Rem 80. (1) Connect Plt and Alb with a straight line (dotted). (2) Read the intersection of the dotted line and vertical axis of ALPlat, which yields the ALPlat index. (3) Connect the ALPlat index and Rem with a straight line (solid). (4) Read the intersection of the solid line to yield the probability of PHLF (2.5%45%).
Table VI Patient characteristics of the validation cohort consisting of 250 HCC patients with or without PHLF
Sex(female) Age (y) Underlying liver disease HBs Ag positive Anti-HCV Ab positive Non-B non-C Plt ( 109/L) PT-INR T-bil (mg/dL) AST (IU/L) ALT (IU/L) Alb (g/dL) KICG Rem (%) Operation time (min) Blood loss (g)
Total (N ¼ 250)
PHLF (þ) (n ¼ 27)
PHLF (-) (n ¼ 223)
P value
47 (19%) 68 ± 9
5 (19%) 68 ± 9
42 (19%) 68 ± 10
.772 .974
46 (18%) 92 (37%) 115 (46%) 155 ± 62 1.08 ± 0.09 0.81 ± 0.45 46 ± 49 41 ± 44 3.81 ± 0.45 0.135 ± 0.04 77 ± 19 400 (120e1044) 577 (0e8,650)
2 (7%) 13 (48%) 12 (44%) 138 ± 87 1.10 ± 0.09 0.97 ± 0.38 56 ± 28 51 ± 34 3.56 ± 0.49 0.122 ± 0.03 69 ± 18 479 (313e961) 1,450 (120e8,650)
44 (20%) 76 (35%) 103 (46%) 157 ± 56 1.08 ± 0.09 0.89 ± 0.34 45 ± 51 39 ± 45 3.84 ± 0.41 0.137 ± 0.04 77 ± 18 389 (120e1,044) 495 (0e7,610)
.087 .202 .864 .135 .152 .270 .286 .189 .001 .077 .023 .0004 < .0001
Data represent the mean ± standard deviation or the median (range) or the number of patients. HB, hepatitis B; HC, hepatitis C; Non-B non-C, non-hepatitis B and non-hepatitis C. Table VII The comparison between the ALPlat criterion and Makuuchi’s criterion in the external validation cohort (N ¼ 250) The number of patients
Within ALPlat Beyond ALPlat Within Makuuchi Beyond Makuuchi
PHLF (þ)
PHLF (-)
13 14 17 10
185 38 159 64
Sensitivity
Specificity
PPV
NPV
PLR
NLR
Accuracy
0.519
0.830
0.269
0.934
3.04
0.580
0.796
0.370
0.713
0.135
0.903
1.29
0.883
0.676
PPV, positive predictive value; NPV, negative predictive value; PLR, positive likelihood ratio; NLR, negative likelihood ratio.
External validation of the ALPlat criterion Our new criterion was validated in an additional cohort of 250 patients from Kyoto University Hospital between 2011 and 2016. The characteristics of the external validation cohort are shown in Table VI. There were 203 men and 47 women, with a mean age of 68 years. The underlying liver diseases were chronic hepatitis B (n ¼ 46, 18%), chronic hepatitis C (n ¼ 92, 37%), chronic hepatitis B and C
(n ¼ 3, 1%), and non-hepatitis B and non-hepatitis C (n ¼ 115, 46%). PHLF occurred in 48 patients. Of these, 19 (8%) were grade B and 8 (3%) were grade C. The ALPlat criterion had a greater sensitivity (0.519) and specificity (0.830) than those of Makuuchi’s criterion in this external cohort (Table VII). ROC analysis showed that the AUROC of the RemALPlat index was 0.749 (95% CI; 0.63e0.84), which did not yield to that of the RemKICG index (0.739, 95% CI; 0.63e0.83) and the RemALBI index (0.731, 95% CI; 0.63e0.81),
6
G. Yamamoto et al. / Surgery xxx (2019) 1e7
indicating that the ALPlat index can be a substitution for the ICG test as a preoperative liver function test. Discussion PHLF can be a fatal complication. Once PHLF occurs, there are few treatment options other than supportive therapy to allow time for the regeneration of the residual liver. The incidence of PHLF varies from 9% to 40% across studies.6e14 Operative mortality after major hepatectomy (resection of 2 or more segments, except left lateral sectionectomy) is 3.3% according to a nationwide survey in Japan,35 and PHLF is thought to be the leading cause of this mortality. Appropriate patient selection based on an accurate preoperative risk assessment is essential to avoid PHLF, but it is occasionally difficult to predict PHLF preoperatively. Various factors are associated with PHLF, including the preoperative liver function, volume of the Rem, blood loss, operative time, and postoperative events, such as infection and unstable hemodynamic status. It is reasonable to develop a resection criterion based on the probability of PHLF that consists of a combination of quantitative liver function parameters and the operative procedures (the resection volume or resection rate). Some resection criteria have been proposed. Particularly in Japan, criteria based on the ICG test are accepted widely, and Makuuchi’s criterion is the gold standard for operative decisionmaking.36 There are, however, some drawbacks to the ICG test. First, it cannot be used for patients with an excretory defect of ICG37 or in those who are allergic to ICG. Second, the ICG test is not performed routinely, except during the preoperative evaluation, which inconveniences physicians who need to make treatment decisions in daily clinical practice. Moreover, although the safety of liver resection satisfying Makuuchi’s criterion has been validated,38 the rationale behind the criterion is unclear. Actually, some reports showed that liver resection beyond the criterion can still be performed safely.39,40 Others have proposed resection criteria based on quantitative parameters rather than on the ICG test. We reported previously that the non-invasive measurement of liver stiffness and Mac-2 binding protein glycosylated isomers were useful for predicting PHLF.21,32 Ueno et al proposed a new prediction model for PHLF that incorporated the serum hyaluronic acid levels and percentage of the Rem.23 Although these studies suggested that liver stiffness, Mac-2 binding protein glycosylated isomers, or hyaluronic acid were more useful than the ICG test for predicting PHLF, they involved a limited number of patients at a single institution, and external validation is necessary to verify these findings. Our study aimed to reappraise the usefulness of general laboratory tests for predicting PHLF in a larger cohort and to establish a new criterion for patient selection. Univariate ROC analysis revealed that the Plt and serum Alb concentration were statistically significant and useful predictors for PHLF compared to the ICG test and other preoperative factors. This study demonstrated the limited efficacy of the ICG test as a predictor of PHLF development, concordant with previous reports.14,21,32 Multivariate analysis provided a model enabling a more accurate quantitative riskassessment for PHLF that incorporated the Plt, serum Alb concentration, and the future Rem. To establish a practical criterion, cutoff values were determined so that the risks for PHLF were comparable to permissible risk levels in Makuuchi’s criterion. This criterion was validated in an external cohort and was shown to be superior to Makuuchi’s criterion in terms of its sensitivity and specificity for PHLF. Although it should be acknowledged that the acceptable risk is adjustable in individual patients, our study provides a simple and useful criterion for evaluating the safety of a liver resection in terms of avoiding PHLF.
Our study does have limitations. First, because the present study was a retrospective study involving multiple institutions, CT volumetry data were not available for all participants, and therefore, we were forced to estimate Rem with reference to volumetric data obtained from healthy livers for anatomic resections or with the tumor size and weight of the specimens for non-anatomic resections.34 Although the possible inaccuracy of Rem might have impaired the reliability of our result to some extent, the result was consistent, even when we experimentally substituted available computed tomography volumetric data (n ¼ 200) for Rem, indicating that Rem was estimated with reasonable accuracy. Second, operative indications varied among hospitals. Some institutions adhered strictly to the ICG-based criteria, which may have resulted in a patient selection bias against the ICG test. Recently, the usefulness of the albumin-bilirubin (ALBI) score was reported for short-term outcomes after liver resection in HCC patients.41,42 Our ALPlat index may be criticized for its similarity to the ALBI score. However, the ALBI score was originally established to stratify patients based on long-term survival.43 In addition, the analyzed cohort included patients undergoing various treatments, including not only operative resection but also local ablation therapy, transcatheter arterial chemoembolization, and even best supportive care. The ALPlat index is a different concept from the ALBI score in that it was developed specifically to predict PHLF and the short-term outcome after liver resection. Indeed, poor liver function in HCC patients negatively affects not only the short-term outcomes but also long-term prognosis after hepatectomy by promoting multicentric carcinogenesis and restricting the therapeutic options for recurrence. Hence, it is natural that some factors will be shared by both short-term and long-term prognostic indices, and it is not surprising that ALBI predicts PHLF as well. In conclusion, we established the ALPlat index which quantitatively evaluated the risk of PHLF after liver resection using a combination of Plt, serum Alb concentration, and the future Rem. The ALPlat criterion is simple and useful to assess liver functions and make therapeutic decisions in patients with HCC. Funding/Support None. Conflict of interest/Disclosure The authors declare they have no conflicts of interest. Supplementary materials Supplementary material associated with this article can be found, in the online version, at https://doi.org/10.1016/j.surg.2019. 09.021. References 1. Torre LA, Bray F, Siegel RL, Ferlay J, Lortet-Tieulent J, Jemal A. Global cancer statistics, 2012. CA Cancer J Clin. 2015;65:87e108. 2. Song P, Tobe RG, Inagaki Y, et al. The management of hepatocellular carcinoma around the world: A comparison of guidelines from 2001 to 2011. Liver Int. 2012;32:1053e1063. 3. Grazi GL, Ercolani G, Pierangeli F, et al. Improved results of liver resection for hepatocellular carcinoma on cirrhosis give the procedure added value. Ann Surg. 2001;234:71e78. 4. Schroeder RA, Marroquin CE, Bute BP, Khuri S, Henderson WG, Kuo PC. Predictive indices of morbidity and mortality after liver resection. Ann Surg. 2006;243:373e379. 5. Rahbari NN, Garden OJ, Padbury R, et al. Posthepatectomy liver failure: a definition and grading by the International Study Group of Liver Surgery (ISGLS). Surgery. 2011;149:713e724.
G. Yamamoto et al. / Surgery xxx (2019) 1e7 6. Fukushima K, Fukumoto T, Kuramitsu K, et al. Assessment of ISGLS definition of posthepatectomy liver failure and its effect on outcome in patients with hepatocellular carcinoma. J Gastrointest Surg. 2014;18:729e736. 7. Chen X, Zhai J, Cai X, et al. Severity of portal hypertension and prediction of postoperative liver failure after liver resection in patients with Child-Pugh grade A cirrhosis. Br J Surg. 2012;99:1701e1710. 8. Huang Y, Huang B, Kan T, Yang B, Yuan M, Wang J. Liver-to-spleen ratio as an index of chronic liver diseases and safety of hepatectomy: A pilot study. World J Surg. 2014;38:3186e3192. 9. Lin XJ, Yang J, Chen XB, Zhang M, Xu MQ. The critical value of remnant liver volume-to-body weight ratio to estimate posthepatectomy liver failure in cirrhotic patients. J Surg Res. 2014;188:489e495. 10. Motoyama H, Kobayashi A, Yokoyama T, et al. Liver failure after hepatocellular carcinoma surgery. Langenbecks Arch Surg. 2014;399:1047e1055. 11. Kim HJ, Kim CY, Park EK, et al. Volumetric analysis and indocyanine green retention rate at 15 min as predictors of post-hepatectomy liver failure. HPB (Oxford). 2015;17:159e167. 12. Iguchi K, Hatano E, Yamanaka K, Tanaka S, Taura K, Uemoto S. Validation of the conventional resection criteria in patients with hepatocellular carcinoma in terms of the incidence of posthepatectomy liver failure and long-term prognosis. Dig Surg. 2015;32:344e351. 13. Nakamura N, Hatano E, Iguchi K, Seo S, Taura K, Uemoto S. Posthepatectomy liver failure affects long-term function after resection for hepatocellular carcinoma. World J Surg. 2016;40:929e936. 14. Tomimaru Y, Eguchi H, Gotoh K, et al. Platelet count is more useful for predicting posthepatectomy liver failure at surgery for hepatocellular carcinoma than indocyanine green clearance test. J Surg Oncol. 2016;113:565e569. 15. Clavien PA, Petrowsky H, DeOliveira ML, Graf R. Strategies for safer liver surgery and partial liver transplantation. N Engl J Med. 2007;356:1545e1559. 16. Kaneko K, Shirai Y, Wakai T, Yokoyama N, Akazawa K, Hatakeyama K. Low preoperative platelet counts predict a high mortality after partial hepatectomy in patients with hepatocellular carcinoma. World J Gastroenterol. 2005;11: 5888e5892. 17. Nanashima A, Yamaguchi H, Tanaka K, et al. Preoperative serum hyaluronic acid level as a good predictor of posthepatectomy complications. Surg Today. 2004;34:913e919. 18. Maithel SK, Kneuertz PJ, Kooby DA, et al. Importance of low preoperative platelet count in selecting patients for resection of hepatocellular carcinoma: A multi-institutional analysis. J Am Coll Surg. 2011;212:638e648. 19. Kim SH, Kang DR, Lee JG, et al. Early predictor of mortality due to irreversible posthepatectomy liver failure in patients with hepatocellular carcinoma. World J Surg. 2013;37:1028e1033. 20. Wang YY, Zhong JH, Su ZY, et al. Albumin-bilirubin versus Child-Pugh score as a predictor of outcome after liver resection for hepatocellular carcinoma. Br J Surg. 2016;103:725e734. 21. Okuda Y, Taura K, Yoshino K, et al. Usefulness of Mac-2 Binding Protein Glycosylation Isomer for Prediction of Posthepatectomy Liver Failure in Patients With Hepatocellular Carcinoma. Ann Surg. 2017;265:1201e1208. 22. Dong J, Zhang XF, Zhu Y, et al. The value of the combination of fibrosis index based on the four factors and future liver remnant volume ratios as a predictor on posthepatectomy outcomes. J Gastrointest Surg. 2015;19:682e691. 23. Ueno M, Uchiyama K, Ozawa S, Hayami S, Kiriyama S, Yamaue H. A new prediction model of postoperative complications after major hepatectomy for hepatocellular carcinoma. Dig Surg. 2009;26:392e399. 24. De Gasperi A, Mazza E, Prosperi M. Indocyanine green kinetics to assess liver function: Ready for a clinical dynamic assessment in major liver surgery? World J Hepatol. 2016;8:355e367.
7
25. Thomas MN, Weninger E, Angele M, et al. Intraoperative simulation of remnant liver function during anatomic liver resection with indocyanine green clearance (LiMON) measurements. HPB (Oxford). 2015;17:471e476. 26. Kokudo T, Hasegawa K, Amikura K, et al. Assessment of preoperative liver function in patients with hepatocellular carcinoma - The Albumin-Indocyanine Green Evaluation (ALICE) grade. PLoS One. 2016;11:1e15. 27. Zou H, Tao Y, Wang ZM. Integration of Cild-Pugh score with future liver remnant yields improved prediction of liver dysfunction risk for HBV-related hepatocellular carcinoma following hepatic resection. Oncol Lett. 2017;13: 3631e3637. 28. Chuang YH, Ou HY, Lazo MZ, et al. Predicting post-hepatectomy liver failure by combined volumetric, functional MR image and laboratory analysis. Liver Int. 2018;38:868e874. 29. Chapelle T, Op De Beeck B, Huyghe I, et al. Future remnant liver function estimated by combining liver volumetry on magnetic resonance imaging with total liver function on 99mTc-mebrofenin hepatobiliary scintigraphy: Can this tool predict post-hepatectomy liver failure? HPB (Oxford). 2016;18: 494e503. 30. Wong JS, Wong GL, Chan AW, et al. Liver stiffness measurement by transient elastography as a predictor on posthepatectomy outcomes. Ann Surg. 2013;257:922e928. 31. Cescon M, Colecchia A, Cucchetti A, et al. Value of transient elastography measured with fibroscan in predicting the outcome of hepatic resection for hepatocellular carcinoma. Ann Surg. 2012;256:706e712; discussion 712e713. 32. Nishio T, Taura K, Koyama Y, et al. Prediction of posthepatectomy liver failure based on liver stiffness measurement in patients with hepatocellular carcinoma. Surgery. 2016;159:399e408. 33. Miyagawa S, Makuuchi M, Kawasaki S, Kakazu T. Criteria for Safe Hepatic Resection. Am J Surg. 1995;169:589e594. 34. Mise Y, Satou S, Shindoh J, et al. Three-dimensional volumetry in 107 normal livers reveals clinically relevant inter-segment variation in size. HPB (Oxford). 2014;16:439e447. 35. Takahara T, Wakabayashi G, Konno H, et al. Comparison of laparoscopic major hepatectomy with propensity score matched open cases from the National Clinical Database in Japan. J Hepatobiliary Pancreat Sci. 2016;23: 721e734. 36. Makuuchi M, Kosuge T, Takayama T, et al. Surgery for small liver cancers. Semin Surg Oncol. 1993;9:298e304. 37. Takase S, Takada A, Matsuda Y. Studies on the pathogenesis of the constitutional excretory defect of indocyanine green. Gastroenterol Jpn. 1982;17: 301e309. 38. Imamura H, Seyama Y, Kokudo N, et al. One thousand fifty-six hepatectomies without mortality in 8 years. Arch Surg. 2003;138:1198e1206. 39. Wang XQ, Liu Z, Lv WP, et al. Safety validation of decision trees for hepatocellular carcinoma. World J Gastroenterol. 2015;21:9394e9402. 40. Lam CM, Fan ST, Lo CM, Wong J. Major hepatectomy for hepatocellular carcinoma in patients with an unsatisfactory indocyanine green clearance test. Br J Surg. 1999;86:1012e1017. 41. Andreatos N, Amini N, Gani F, et al. Albumin-bilirubin score: Predicting shortterm outcomes including bile leak and post-hepatectomy liver failure following hepatic resection. J Gastrointest Surg. 2017;21:238e248. 42. Zou H, Wen Y, Yuan K, Miao XY, Xiong L, Liu KJ. Combining albumin-bilirubin score with future liver remnant predicts post-hepatectomy liver failure in HBV-associated HCC patients. Liver Int. 2018;38:494e502. 43. Johnson PJ, Berhane S, Kagebayashi C, et al. Assessment of liver function in patients with hepatocellular carcinoma: A new evidence-based approach-the ALBI grade. J Clin Oncol. 2015;33:550e558.
Contents lists available at ScienceDirect
Surgery journal homepage: www.elsevier.com/locate/surg
ALPlat criterion for the resection of hepatocellular carcinoma based on a predictive model of posthepatectomy liver failure Gen Yamamoto, MD, PhDa, Kojiro Taura, MD, PhDa,*, Iwao Ikai, MD, PhDb, Takahisa Fujikawa, MD, PhDc, Ryuta Nishitai, MD, PhDd, Satoshi Kaihara, MD, PhDe, Masazumi Zaima, MD, PhDf, Hiroaki Terajima, MD, PhDg, Tsunehiro Yoshimura, MD, PhDh, Yukinori Koyama, MD, PhDi, Kazutaka Tanabe, MD, PhDa, Takahiro Nishio, MD, PhDa, Yukihiro Okuda, MD, PhDa, Yoshinobu Ikeno, MD, PhDa, Kenji Yoshino, MD, PhDa, Keita Fukuyama, MDa, Satoru Seo, MD, PhDa, Etsuro Hatano, MD, PhDj, Shinji Uemoto, MD, PhDa a
Department of Surgery, Kyoto University Graduate School of Medicine, Kyoto, Japan Department of Surgery, National Hospital Organization Kyoto Medical Centre, Kyoto, Japan Department of Surgery, Kokura Memorial Hospital, Kitakyushu, Japan d Department of Surgery, Kyoto-Katsura Hospital, Kyoto, Japan e Department of Surgery, Kobe City Medical Centre General Hospital, Kobe, Japan f Department of Surgery, Shiga General Hospital, Moriyama, Japan g Department of Surgery, Tazuke Kofukai Medical Research Institute Kitano Hospital, Osaka, Japan h Department of Surgery, Tenriyorodusoudanjyo Hospital, Tenri, Japan i Department of Surgery, Shimane Prefectural Central Hospital, Izumo, Japan j Department of Surgery, Hyogo College of Medicine, Nishinomiya, Japan b c
a r t i c l e i n f o
a b s t r a c t
Article history: Accepted 26 September 2019 Available online xxx
Background: The indocyanine green test is used widely to evaluate the risk of posthepatectomy liver failure for hepatocellular carcinoma. A more convenient and reliable scoring system is desired owing to limited accuracy and availability of the indocyanine green test. This study aimed to establish a new selection criterion for liver resection in HCC. Methods: We reviewed retrospectively 876 patients undergoing a partial hepatectomy for hepatocellular carcinoma between 2007 and 2015 in 8 affiliated hospitals. Posthepatectomy liver failure grades B and C were regarded as posthepatectomy liver failure. We identified the risk factors for posthepatectomy liver failure and established a predictive model based on a formula for the probability of posthepatectomy liver failure. External validation was performed in an additional cohort of 250 patients. Results: Posthepatectomy liver failure occurred in 92 patients (11%). The area under the receiver operating characteristic curve for the prediction of posthepatectomy liver failure was 0.646 for the platelet count, 0.641 for albumin, 0.623 for the percentage of liver remnant, and 0.607 for the plasma disappearance rate of indocyanine green. Logistic regression analysis provided a formula for the probability of posthepatectomy liver failure consisting of platelet count, albumin, and liver remnant. We defined platelet count þ 90 albumin as the ALPlat index and established an ALPlat-based criterion for operative resection that secured the same risk assumed by the indocyanine greenebased criterion (Makuuchi’s criterion). This criterion exhibited a greater sensitivity and specificity than the indocyanine greenebased criterion in the validation cohort. Conclusion: The ALPlat criterion is a simple and useful method to assess liver function and to make therapeutic decisions in patients with hepatocellular carcinoma. © 2019 Elsevier Inc. All rights reserved.
* Reprint requests: Kojiro Taura, MD, PhD, Department of Surgery, Kyoto University Graduate School of Medicine, 54 Kawahara-cho, Shogoin, Sakyo-ku, Kyoto, 606-8507, Japan. E-mail address: [email protected] (K. Taura). https://doi.org/10.1016/j.surg.2019.09.021 0039-6060/© 2019 Elsevier Inc. All rights reserved.
2
G. Yamamoto et al. / Surgery xxx (2019) 1e7
Introduction Hepatocellular carcinoma (HCC) is the second leading cause of cancer mortality worldwide in men and the sixth in women.1 There are a variety of treatment options for HCC, such as liver resection, liver transplantation, radiofrequency ablation or other ablative techniques, transcatheter arterial chemoembolization, percutaneous ethanol injection therapy, molecular-targeted therapy, and radiation therapy.2 Liver resection is an effective therapeutic option. The outcomes after liver resection have substantially improved due to the refined surgical technique, perioperative care, and improvements in patient selection.3 Nevertheless, posthepatectomy liver failure (PHLF) occurs in a non-negligible number of patients and remains a leading cause of postoperative mortality in HCC patients.4e6 The incidence of PHLF varies anywhere from 9% to 40% and depends on the severity of the underlying liver disease.6e14 To decrease the incidence of PHLF, careful patient selection based on a close assessment of liver function is of paramount importance.15 Accurate prediction of PHLF before liver resection, however, is sometimes difficult because multiple factors are associated with PHLF, including the preoperative liver function, volume of the liver remnant, blood loss, operative time, and postoperative events, such as infection and unstable hemodynamic status. The results of laboratory tests14,16e23 and quantitative liver function tests, such as the indocyanine green (ICG) retention test,11,24e26 volume of the liver remnant,9,27e29 and non-invasive liver stiffness measurement,30e32 are useful parameters for predicting PHLF. The ICG retention test is used widely as an algorithm for a safe operative intervention for HCC.15,33 Makuuchi’s criterion, which determines permissible liver resection procedures based on the ICG retention test, has been the gold standard for safe hepatectomies in many Asian countries.33 Our previous studies21,32 and others,14,23 however, demonstrated some problems with the efficacy of the ICG test as a predictor for PHLF in several situations. These findings were based on data from a single institution, and further research involving larger patient numbers is needed for verification. Moreover, the ICG test is not performed routinely in daily clinical practice. For these reasons, physicians feel that this method can be inconvenient for making treatment decisions. The establishment of resection criterion solely based on routine laboratory tests would be highly desirable. This study aimed to reappraise the usefulness of common laboratory tests in predicting PHLF in a larger cohort and to establish an easily calculated criterion for patient selection for liver resection by generating a predictive formula for PHLF that incorporates only routine laboratory tests. Methods Patients We reviewed retrospectively 876 patients undergoing a partial hepatectomy for HCC in our department and 7 affiliated hospitals of Kyoto University between 2007 and 2015 (National Hospital Organization Kyoto Medical Centre: 2009e2015, Kokura Memorial Hospital: 2007e2013, Kyoto-Katsura Hospital: 2012e2015, Kobe City Medical Centre General Hospital: 2011e2015, Shiga General Hospital: 2011e2015, Tazuke Kofukai Medical Research Institute Kitano Hospital: 2007e2015, Tenriyorodusoudanjyo Hospital: 2011e2015, and Kyoto University Hospital: 2007e2011). The exclusion criteria included patients undergoing combined resection of an extrahepatic bile duct and patients undergoing preoperative portal vein embolization. This study was performed in accordance with the ethical guidelines for epidemiologic research in Japan and was approved by the Ethics Committee of the Kyoto University
Graduate School of Medicine (approval code: E2056) and separately by each hospital. For external validation of the criterion established with the aforementioned 876 cases, 250 additional HCC patients undergoing a partial hepatectomy in our department between 2011 and 2016 were reviewed subsequently. Data collection Demographic data, including sex, age, and status of viral infection, were collected for all patients. We ran routine preoperative laboratory tests, including the platelet count (Plt), international normalized ratio of prothrombin time (PT-INR), and serum levels of total bilirubin (T-bil), aspartate transaminase (AST), alanine transaminase (ALT), and albumin (Alb). A preoperative ICG test was performed in all patients. After intravenous injection of indocyanine green 0.5 mg/kg (Diagnogreen; Daiichi Pharmaceutical Co, Inc, Tokyo, Japan), the indocyanine green disappearance rate (KICG) was calculated by linear regression from the plasma concentrations of indocyanine green at 5, 10, and 15 minutes. The operative factors, including the operation time, estimated blood loss, requirement of transfusion, operative procedure (resected segments and the use of laparoscopy), and concomitant resection of other organs, were recorded. We estimated the liver remnant (Rem,%) according to the operative procedures, with reference to a previous study34 on the volumes and proportions of each segment for all patients. For nonanatomic resections, the Rem was estimated using the tumor size and weight of the specimen. PHLF was defined using the International Study Group of Liver Surgery criteria.5 In brief, when both T-bil and PT-INR were abnormal on or after postoperative day 5 according to the normal cutoff levels defined by the local hospital, patients were diagnosed with PHLF. PHLF patients were then categorized into grades A, B, or C based on the required treatment. The primary endpoint of this study was the incidence of PHLF grades B and C as defined by the International Study Group of Liver Surgery. Hepatectomy procedures Hepatectomy is the standard treatment for HCC whenever tumors are considered to be resectable, and the patient’s condition permits the operation. The indications for the resection of HCC include a Child-Pugh status of A or B and a future Rem of greater than 30% of the entire liver, but modifications by each institution were allowed. Intraoperative ultrasonography was performed to identify any occult tumors missed by preoperative imaging and to confirm the relative positions between tumors and vascular structures. Transection of the liver parenchyma was performed using a Cavitron Ultrasonic Surgical Aspirator (Valleylab, Inc, Boulder, CO) and bipolar cautery with a saline irrigation system. For laparoscopic liver resections, a combination of the Cavitron Ultrasonic Surgical Aspirator, THUNDERBEAT (Olympus Corporation, Tokyo, Japan), and BiClamp (ERBE Elektromedizin GmbH, Tubeingen, Germany) devices was used for liver transection. The Pringle maneuver was applied as indicated, with 15 minutes of occlusion alternating with 5 minutes of reperfusion. Statistical analysis Categorical variables are described as frequencies and percentages and were compared using the c2 or Fisher exact test as appropriate. Continuous variables are described as the mean ± standard deviation or the median (range) and were compared using the Mann-Whitney U test. Univariate analyses of risk factors for PHLF were performed using Fisher exact test. Multivariate analyses
G. Yamamoto et al. / Surgery xxx (2019) 1e7
3
Table I Patient characteristics and the comparison between patients with and without PHLF
Sex (female) Age (y) Underlying liver disease HBs Ag positive Anti-HCV Ab positive Non-B non-C positive Plt (109/L) PT-INR T-bil (mg/dL) AST (IU/L) ALT (IU/L) Alb (g/dL) KICG Rem (%) Operation time (min) Blood loss (g)
Total (N ¼ 876)
PHLF (þ) (n ¼ 92)
PHLF (-) (n ¼ 784)
P value
197 (23%) 69 ± 9
17 (19%) 69 ± 10
180 (23%) 69 ± 9
.320 .612
154 (18%) 367 (42%) 360 (41%) 158 ± 70 1.10 ± 0.15 0.81 ± 0.45 44 ± 30 40 ± 34 3.92 ± 0.46 0.144 ± 0.07 72 ± 21 342 (66e849) 502 (0e26,800)
12 (13%) 49 (53%) 32 (35%) 131 ± 59 1.17 ± 0.23 0.86 ± 0.33 53 ± 40 43 ± 30 3.73 ± 0.45 0.128 ± 0.05 63 ± 24 415 (124e822) 1,255 (38e26,800)
142 (18%) 318 (41%) 328 (42%) 161 ± 71 1.09 ± 0.14 0.81 ± 0.46 43 ± 28 40 ± 34 3.94 ± 0.46 0.146 ± 0.07 73 ± 20 337 (66e849) 460 (0e7,000)
.173 .025 .193 < .0001 .0002 .344 .005 .408 < .0001 .002 < .0001 < .0001 < .0001
Data represent the mean ± standard deviation or the median (range) or the number of patients. HB, hepatitis B; HC, hepatitis C; Non-B non-C, non-hepatitis B and non-hepatitis C.
of risk factors for PHLF were performed using the logistic regression model. Variables associated with the incidence of PHLF at P < .05 in the univariate analysis were included in a multivariate logistic regression analysis, and the odds ratios (OR) and 95% confidence intervals (CI) were calculated. A receiver operating characteristic (ROC) analysis was performed to evaluate the power of the predictors, and the area under the receiver operating characteristic curve (AUROC) and the 95% confidence intervals (CIs) were calculated. We established the criterion for liver resection based on a prediction formula developed by multivariate analysis of data from the 876 patients, and we compared this method with the ICG-based criterion (Makuuchi’s criterion). The criterion was then validated in an additional cohort of 250 patients. The calculation of sensitivity, specificity, positive predictive value, negative predictive value, positive likelihood ratio, negative likelihood ratio, and accuracy was performed based on the exact binary classification. All statistical evaluations were performed using JMP 13 software (SAS institute Inc, Cary, NC). All tests were 2-sided, and differences with P values less than .05 were considered statistically significant. Results Patient characteristics The characteristics of the study population are shown in Table I. There were 679 men and 197 women with a mean age of 69 years. The underlying liver diseases included chronic hepatitis B (positive for hepatitis B surface antigen; n ¼ 154, 18%), chronic hepatitis C (positive for anti-hepatitis C antibody; n ¼ 367, 42%), chronic hepatitis B and C (positive for both; n ¼ 5, 1%), and non-hepatitis B and non-hepatitis C (negative for both; n ¼ 360, 41%). PHLF developed more frequently in patients with chronic hepatitis C (P ¼ .025). Patient characteristics in the PHLF and non-PHLF groups PHLF occurred in 189 patients. Of these, 97 (11%) were classified as grade A, 77 (9%) as grade B, and 15 (2%) as grade C. We regarded PHLF grades B and C as PHLF because Grade A does not require intervention and is not clinically relevant. Table I shows the comparison between patients with and without PHLF. Statistically significant differences between PHLF and non-PHLF groups were observed for Plt, PT-INR, AST, Alb, KICG, and Rem.
Table II ROC analysis of preoperative factors for PHLF
Plt Alb PT-INR Rem AST KICG T-bil ALT
AUROC
95% CI
0.646 0.641 0.624 0.623 0.613 0.607 0.569 0.558
0.59e0.70 0.58e0.70 0.56e0.69 0.55e0.67 0.55e0.67 0.54e0.67 0.51e0.63 0.50e0.62
Table III ROC analysis of preoperative factors in combination with Rem for PHLF
Rem Plt Rem KICG Rem PT-INR Rem Alb Rem AST Rem T-bil Rem ALT
OR
95% CI
P value
AUROC
0.54 0.43 0.59 0.44 0.64 1.43 0.63 0.61 0.68 1.25 0.65 1.14 0.66 1.05
0.44e0.67 0.31e0.59 0.48e0.72 0.29e0.65 0.52e0.78 1.19e1.72 0.52e0.76 0.50e0.76 0.55e0.82 1.04e1.49 0.53e0.79 0.95e1.35 0.54e0.80 0.87e1.28
< .0001 < .0001 < .0001 < .0001 < .0001 .0001 < .0001 < .0001 < .0001 .0154 < .0001 .16 < .0001 .60
0.724 0.698 0.696 0.695 0.651 0.642 0.631
The adjusted ORs per standard deviation change are shown.
ROC analysis of preoperative factors for PHLF Table II shows the ROC analysis, where PHLF was predicted by a single factor. The AUROC of KICG was 0.607, which was less than the corresponding values of Plt (0.646), Alb (0.641), PT-INR (0.624), Rem (0.623), and AST (0.613). The best predictor for PHLF was Plt. We also analyzed the predictive capacity of laboratory factors in combination with Rem for PHLF (Table III). ROC analysis showed that the AUROC of Plt combined with Rem was 0.724, which was better than that of KICG with Rem (0.698) or any other single factor combined with Rem (PT-INR with Rem; 0.696, Alb with Rem; 0.695, AST with Rem; 0.651, T-bil with Rem; 0.642, and ALT with Rem; 0.631).
4
G. Yamamoto et al. / Surgery xxx (2019) 1e7
Multivariate analysis of preoperative factors for PHLF Multivariate logistic regression analysis was performed to identify the risk factors for PHLF. Plt (OR; 0.51, 95% CI; 0.37e0.70), PT-INR (OR; 1.27, 95% CI; 1.05e1.49), Alb (OR; 0.68, 95% CI; 0.53e0.85), and Rem (OR; 0.50, 95% CI; 0.39e0.62) were statistically significant independent preoperative predictors for PHLF (Table IV). ALPlat criterion for operative decision-making Logistic regression analysis yielded a formula for the probability of PHLF consisting of Rem, Plt, and Alb, which showed a lesser P value and served our purpose of establishing a simple index. The probability of PHLF was then calculated as follows:
resection) were given as 0.1475 and 0.1312, respectively, based on Makuuchi’s criterion. These results led to the generation of a criterion for a safe resection based on the ALPlat index (Table V), which we termed the ALPlat criterion. The minimum values of the ALPlat index that permitted a safe right hemihepatectomy, left hemihepatectomy, and segmentectomy were determined such that the risks for PHLF were equivalent to the permissible risks based on Makuuchi’s criterion. These values were 531 for right hemihepatectomy, 461 for left hemihepatectomy, and 423 for segmentectomy. ROC analysis showed that the AUROC of the RemALPlat index (0.753) was significantly greater than that of the RemKICG index (0.698) (Supplementary Table I). We also made a comparison between the RemALPlat index and the ALBI score (¼ [log10 bilirubin 0.66] þ [albumin e0.085], where bilirubin is in mmol/L and albumin in g/L) combined with
. h i Probability ðfor PHLFÞ ¼ 1 1 þ exp 0:0334 Rem ð%Þ þ 0:0113 Plt 109 =L þ 1:0154 Alb ðg=dLÞ 5:6797 h i i . h ¼ 1 1 þ exp 0:0334 Rem ð%Þ þ 0:0113 Plt 109 =L þ 90 Alb ½g=dL 5:6797 ¼1 1 þ exp 0:0334 Rem ð%Þ þ 0:0113 ALPlat index 5:6797 ¼ 1=ð1 þ exp ½RemALPlat index 5:6797Þ;
where Plt ð109 =LÞ þ 90 Alb ðg=dLÞ was defined as the ALPlat index and 0:0334 Rem ð%Þ þ 0:0113 ALPlat index was defined as the RemALPlat index. This formula allows the estimation of the probability of PHLF based on the ALPlat index and Rem, which should be useful for determining the safety and risk of PHLF for individual patients. We tried to establish a criterion that guaranteed equivalent safety to that of the ICG-based Makuuchi’s criterion. The probabilities of PHLF permitted by Makuuchi’s criterion were calculated. By logistic regression analysis,
Rem (defined as the RemALBI index) in the same manner. The AUROC of the RemALPlat index (0.753) was significantly greater than that of the RemALBI index (0.712) (Supplementary Table I). Figure 1, A shows the nomogram that determines the ALPlat index based on the Plt and Alb values. Figure 1, A also provides the probability of PHLF based on the ALPlat index and Rem, as well as the minimum requirement of Rem based on the ALPlat index and the acceptable probability of PHLF, which helps us estimate the surgical risk and determine an appropriate safe operative procedure. Figure 1, B demonstrates how the nomogram determines the probability of PHLF from the Plt, Alb, and Rem values. Figure 1, C
Probability ðfor PHLFÞ ¼ 1=ð1 þ exp ½0:0263 Rem ð%Þ þ 11:8358 KICG 1:2457Þ ¼ 1=ð1 þ exp ½RemKICG index 1:2457Þ;
where
0:0263 Rem ð%Þ þ 11:8358 KICG
was
defined
as
RemKICG index. In the case of right hemihepatectomy for example, substituting 66.7 (%) for Rem and 0.154 (because 10% of the ICG retention rate at 15 minutes corresponds to 0.154 of KICG) for KICG; this formula gave 0.1909 as the permissible risk for a safe right hemihepatectomy. In the same manner, the permissible risks for a safe left hemihepatectomy (33.3% resection) and segmentectomy (16.7%
Table IV Multivariate analysis of preoperative factors for PHLF
Plt PT-INR Alb Rem
shows how to estimate Rem from the values of Plt and Alb when the acceptable probability of PHLF is given. We have created an Excel file (Microsoft Corporation, Redmond, WA) that allows us to easily calculate the PHLF risk by simply entering the Plt, Alb, and Rem values (Supplementary File 1). It also provides a graph showing the relationship between Rem and PHLF when given the Plt and Alb values. A patient with Plt 150 (109/L) and Alb 3.5 (g/dL) is shown as an example.
Table V ALPlat criterion: Operative indication by the ALPlat index
OR
95% CI
P value
ALPlat index
0.51 1.27 0.68 0.50
0.37e0.70 1.05e1.49 0.53e0.85 0.39e0.62
< .0001 .011 .001 < .0001
>531 461e531 423e461 <423
Surgical procedures 2/3 resection 1/3 resection 1/6 resection <1/6 resection
Right hemihepatectomy, trisectionectomy Left hemihepatectomy, sectionectomy Segmentectomy Limited resection
G. Yamamoto et al. / Surgery xxx (2019) 1e7
5
Figure 1. (A) ALPlat nomogram. The nomogram can help to rapidly assess the ALPlat index. When the ALPlat index and Rem are connected by a straight line, the probability is obtained. (B) Example 1. Plt 200, Alb 4.0, Rem 80. (1) Connect Plt and Alb with a straight line (dotted). (2) Read the intersection of the dotted line and vertical axis of ALPlat, which yields the ALPlat index. (3) Connect the ALPlat index and Rem with a straight line (solid). (4) Read the intersection of the solid line to yield the probability of PHLF (2.5%
Table VI Patient characteristics of the validation cohort consisting of 250 HCC patients with or without PHLF
Sex(female) Age (y) Underlying liver disease HBs Ag positive Anti-HCV Ab positive Non-B non-C Plt ( 109/L) PT-INR T-bil (mg/dL) AST (IU/L) ALT (IU/L) Alb (g/dL) KICG Rem (%) Operation time (min) Blood loss (g)
Total (N ¼ 250)
PHLF (þ) (n ¼ 27)
PHLF (-) (n ¼ 223)
P value
47 (19%) 68 ± 9
5 (19%) 68 ± 9
42 (19%) 68 ± 10
.772 .974
46 (18%) 92 (37%) 115 (46%) 155 ± 62 1.08 ± 0.09 0.81 ± 0.45 46 ± 49 41 ± 44 3.81 ± 0.45 0.135 ± 0.04 77 ± 19 400 (120e1044) 577 (0e8,650)
2 (7%) 13 (48%) 12 (44%) 138 ± 87 1.10 ± 0.09 0.97 ± 0.38 56 ± 28 51 ± 34 3.56 ± 0.49 0.122 ± 0.03 69 ± 18 479 (313e961) 1,450 (120e8,650)
44 (20%) 76 (35%) 103 (46%) 157 ± 56 1.08 ± 0.09 0.89 ± 0.34 45 ± 51 39 ± 45 3.84 ± 0.41 0.137 ± 0.04 77 ± 18 389 (120e1,044) 495 (0e7,610)
.087 .202 .864 .135 .152 .270 .286 .189 .001 .077 .023 .0004 < .0001
Data represent the mean ± standard deviation or the median (range) or the number of patients. HB, hepatitis B; HC, hepatitis C; Non-B non-C, non-hepatitis B and non-hepatitis C. Table VII The comparison between the ALPlat criterion and Makuuchi’s criterion in the external validation cohort (N ¼ 250) The number of patients
Within ALPlat Beyond ALPlat Within Makuuchi Beyond Makuuchi
PHLF (þ)
PHLF (-)
13 14 17 10
185 38 159 64
Sensitivity
Specificity
PPV
NPV
PLR
NLR
Accuracy
0.519
0.830
0.269
0.934
3.04
0.580
0.796
0.370
0.713
0.135
0.903
1.29
0.883
0.676
PPV, positive predictive value; NPV, negative predictive value; PLR, positive likelihood ratio; NLR, negative likelihood ratio.
External validation of the ALPlat criterion Our new criterion was validated in an additional cohort of 250 patients from Kyoto University Hospital between 2011 and 2016. The characteristics of the external validation cohort are shown in Table VI. There were 203 men and 47 women, with a mean age of 68 years. The underlying liver diseases were chronic hepatitis B (n ¼ 46, 18%), chronic hepatitis C (n ¼ 92, 37%), chronic hepatitis B and C
(n ¼ 3, 1%), and non-hepatitis B and non-hepatitis C (n ¼ 115, 46%). PHLF occurred in 48 patients. Of these, 19 (8%) were grade B and 8 (3%) were grade C. The ALPlat criterion had a greater sensitivity (0.519) and specificity (0.830) than those of Makuuchi’s criterion in this external cohort (Table VII). ROC analysis showed that the AUROC of the RemALPlat index was 0.749 (95% CI; 0.63e0.84), which did not yield to that of the RemKICG index (0.739, 95% CI; 0.63e0.83) and the RemALBI index (0.731, 95% CI; 0.63e0.81),
6
G. Yamamoto et al. / Surgery xxx (2019) 1e7
indicating that the ALPlat index can be a substitution for the ICG test as a preoperative liver function test. Discussion PHLF can be a fatal complication. Once PHLF occurs, there are few treatment options other than supportive therapy to allow time for the regeneration of the residual liver. The incidence of PHLF varies from 9% to 40% across studies.6e14 Operative mortality after major hepatectomy (resection of 2 or more segments, except left lateral sectionectomy) is 3.3% according to a nationwide survey in Japan,35 and PHLF is thought to be the leading cause of this mortality. Appropriate patient selection based on an accurate preoperative risk assessment is essential to avoid PHLF, but it is occasionally difficult to predict PHLF preoperatively. Various factors are associated with PHLF, including the preoperative liver function, volume of the Rem, blood loss, operative time, and postoperative events, such as infection and unstable hemodynamic status. It is reasonable to develop a resection criterion based on the probability of PHLF that consists of a combination of quantitative liver function parameters and the operative procedures (the resection volume or resection rate). Some resection criteria have been proposed. Particularly in Japan, criteria based on the ICG test are accepted widely, and Makuuchi’s criterion is the gold standard for operative decisionmaking.36 There are, however, some drawbacks to the ICG test. First, it cannot be used for patients with an excretory defect of ICG37 or in those who are allergic to ICG. Second, the ICG test is not performed routinely, except during the preoperative evaluation, which inconveniences physicians who need to make treatment decisions in daily clinical practice. Moreover, although the safety of liver resection satisfying Makuuchi’s criterion has been validated,38 the rationale behind the criterion is unclear. Actually, some reports showed that liver resection beyond the criterion can still be performed safely.39,40 Others have proposed resection criteria based on quantitative parameters rather than on the ICG test. We reported previously that the non-invasive measurement of liver stiffness and Mac-2 binding protein glycosylated isomers were useful for predicting PHLF.21,32 Ueno et al proposed a new prediction model for PHLF that incorporated the serum hyaluronic acid levels and percentage of the Rem.23 Although these studies suggested that liver stiffness, Mac-2 binding protein glycosylated isomers, or hyaluronic acid were more useful than the ICG test for predicting PHLF, they involved a limited number of patients at a single institution, and external validation is necessary to verify these findings. Our study aimed to reappraise the usefulness of general laboratory tests for predicting PHLF in a larger cohort and to establish a new criterion for patient selection. Univariate ROC analysis revealed that the Plt and serum Alb concentration were statistically significant and useful predictors for PHLF compared to the ICG test and other preoperative factors. This study demonstrated the limited efficacy of the ICG test as a predictor of PHLF development, concordant with previous reports.14,21,32 Multivariate analysis provided a model enabling a more accurate quantitative riskassessment for PHLF that incorporated the Plt, serum Alb concentration, and the future Rem. To establish a practical criterion, cutoff values were determined so that the risks for PHLF were comparable to permissible risk levels in Makuuchi’s criterion. This criterion was validated in an external cohort and was shown to be superior to Makuuchi’s criterion in terms of its sensitivity and specificity for PHLF. Although it should be acknowledged that the acceptable risk is adjustable in individual patients, our study provides a simple and useful criterion for evaluating the safety of a liver resection in terms of avoiding PHLF.
Our study does have limitations. First, because the present study was a retrospective study involving multiple institutions, CT volumetry data were not available for all participants, and therefore, we were forced to estimate Rem with reference to volumetric data obtained from healthy livers for anatomic resections or with the tumor size and weight of the specimens for non-anatomic resections.34 Although the possible inaccuracy of Rem might have impaired the reliability of our result to some extent, the result was consistent, even when we experimentally substituted available computed tomography volumetric data (n ¼ 200) for Rem, indicating that Rem was estimated with reasonable accuracy. Second, operative indications varied among hospitals. Some institutions adhered strictly to the ICG-based criteria, which may have resulted in a patient selection bias against the ICG test. Recently, the usefulness of the albumin-bilirubin (ALBI) score was reported for short-term outcomes after liver resection in HCC patients.41,42 Our ALPlat index may be criticized for its similarity to the ALBI score. However, the ALBI score was originally established to stratify patients based on long-term survival.43 In addition, the analyzed cohort included patients undergoing various treatments, including not only operative resection but also local ablation therapy, transcatheter arterial chemoembolization, and even best supportive care. The ALPlat index is a different concept from the ALBI score in that it was developed specifically to predict PHLF and the short-term outcome after liver resection. Indeed, poor liver function in HCC patients negatively affects not only the short-term outcomes but also long-term prognosis after hepatectomy by promoting multicentric carcinogenesis and restricting the therapeutic options for recurrence. Hence, it is natural that some factors will be shared by both short-term and long-term prognostic indices, and it is not surprising that ALBI predicts PHLF as well. In conclusion, we established the ALPlat index which quantitatively evaluated the risk of PHLF after liver resection using a combination of Plt, serum Alb concentration, and the future Rem. The ALPlat criterion is simple and useful to assess liver functions and make therapeutic decisions in patients with HCC. Funding/Support None. Conflict of interest/Disclosure The authors declare they have no conflicts of interest. Supplementary materials Supplementary material associated with this article can be found, in the online version, at https://doi.org/10.1016/j.surg.2019. 09.021. References 1. Torre LA, Bray F, Siegel RL, Ferlay J, Lortet-Tieulent J, Jemal A. Global cancer statistics, 2012. CA Cancer J Clin. 2015;65:87e108. 2. Song P, Tobe RG, Inagaki Y, et al. The management of hepatocellular carcinoma around the world: A comparison of guidelines from 2001 to 2011. Liver Int. 2012;32:1053e1063. 3. Grazi GL, Ercolani G, Pierangeli F, et al. Improved results of liver resection for hepatocellular carcinoma on cirrhosis give the procedure added value. Ann Surg. 2001;234:71e78. 4. Schroeder RA, Marroquin CE, Bute BP, Khuri S, Henderson WG, Kuo PC. Predictive indices of morbidity and mortality after liver resection. Ann Surg. 2006;243:373e379. 5. Rahbari NN, Garden OJ, Padbury R, et al. Posthepatectomy liver failure: a definition and grading by the International Study Group of Liver Surgery (ISGLS). Surgery. 2011;149:713e724.
G. Yamamoto et al. / Surgery xxx (2019) 1e7 6. Fukushima K, Fukumoto T, Kuramitsu K, et al. Assessment of ISGLS definition of posthepatectomy liver failure and its effect on outcome in patients with hepatocellular carcinoma. J Gastrointest Surg. 2014;18:729e736. 7. Chen X, Zhai J, Cai X, et al. Severity of portal hypertension and prediction of postoperative liver failure after liver resection in patients with Child-Pugh grade A cirrhosis. Br J Surg. 2012;99:1701e1710. 8. Huang Y, Huang B, Kan T, Yang B, Yuan M, Wang J. Liver-to-spleen ratio as an index of chronic liver diseases and safety of hepatectomy: A pilot study. World J Surg. 2014;38:3186e3192. 9. Lin XJ, Yang J, Chen XB, Zhang M, Xu MQ. The critical value of remnant liver volume-to-body weight ratio to estimate posthepatectomy liver failure in cirrhotic patients. J Surg Res. 2014;188:489e495. 10. Motoyama H, Kobayashi A, Yokoyama T, et al. Liver failure after hepatocellular carcinoma surgery. Langenbecks Arch Surg. 2014;399:1047e1055. 11. Kim HJ, Kim CY, Park EK, et al. Volumetric analysis and indocyanine green retention rate at 15 min as predictors of post-hepatectomy liver failure. HPB (Oxford). 2015;17:159e167. 12. Iguchi K, Hatano E, Yamanaka K, Tanaka S, Taura K, Uemoto S. Validation of the conventional resection criteria in patients with hepatocellular carcinoma in terms of the incidence of posthepatectomy liver failure and long-term prognosis. Dig Surg. 2015;32:344e351. 13. Nakamura N, Hatano E, Iguchi K, Seo S, Taura K, Uemoto S. Posthepatectomy liver failure affects long-term function after resection for hepatocellular carcinoma. World J Surg. 2016;40:929e936. 14. Tomimaru Y, Eguchi H, Gotoh K, et al. Platelet count is more useful for predicting posthepatectomy liver failure at surgery for hepatocellular carcinoma than indocyanine green clearance test. J Surg Oncol. 2016;113:565e569. 15. Clavien PA, Petrowsky H, DeOliveira ML, Graf R. Strategies for safer liver surgery and partial liver transplantation. N Engl J Med. 2007;356:1545e1559. 16. Kaneko K, Shirai Y, Wakai T, Yokoyama N, Akazawa K, Hatakeyama K. Low preoperative platelet counts predict a high mortality after partial hepatectomy in patients with hepatocellular carcinoma. World J Gastroenterol. 2005;11: 5888e5892. 17. Nanashima A, Yamaguchi H, Tanaka K, et al. Preoperative serum hyaluronic acid level as a good predictor of posthepatectomy complications. Surg Today. 2004;34:913e919. 18. Maithel SK, Kneuertz PJ, Kooby DA, et al. Importance of low preoperative platelet count in selecting patients for resection of hepatocellular carcinoma: A multi-institutional analysis. J Am Coll Surg. 2011;212:638e648. 19. Kim SH, Kang DR, Lee JG, et al. Early predictor of mortality due to irreversible posthepatectomy liver failure in patients with hepatocellular carcinoma. World J Surg. 2013;37:1028e1033. 20. Wang YY, Zhong JH, Su ZY, et al. Albumin-bilirubin versus Child-Pugh score as a predictor of outcome after liver resection for hepatocellular carcinoma. Br J Surg. 2016;103:725e734. 21. Okuda Y, Taura K, Yoshino K, et al. Usefulness of Mac-2 Binding Protein Glycosylation Isomer for Prediction of Posthepatectomy Liver Failure in Patients With Hepatocellular Carcinoma. Ann Surg. 2017;265:1201e1208. 22. Dong J, Zhang XF, Zhu Y, et al. The value of the combination of fibrosis index based on the four factors and future liver remnant volume ratios as a predictor on posthepatectomy outcomes. J Gastrointest Surg. 2015;19:682e691. 23. Ueno M, Uchiyama K, Ozawa S, Hayami S, Kiriyama S, Yamaue H. A new prediction model of postoperative complications after major hepatectomy for hepatocellular carcinoma. Dig Surg. 2009;26:392e399. 24. De Gasperi A, Mazza E, Prosperi M. Indocyanine green kinetics to assess liver function: Ready for a clinical dynamic assessment in major liver surgery? World J Hepatol. 2016;8:355e367.
7
25. Thomas MN, Weninger E, Angele M, et al. Intraoperative simulation of remnant liver function during anatomic liver resection with indocyanine green clearance (LiMON) measurements. HPB (Oxford). 2015;17:471e476. 26. Kokudo T, Hasegawa K, Amikura K, et al. Assessment of preoperative liver function in patients with hepatocellular carcinoma - The Albumin-Indocyanine Green Evaluation (ALICE) grade. PLoS One. 2016;11:1e15. 27. Zou H, Tao Y, Wang ZM. Integration of Cild-Pugh score with future liver remnant yields improved prediction of liver dysfunction risk for HBV-related hepatocellular carcinoma following hepatic resection. Oncol Lett. 2017;13: 3631e3637. 28. Chuang YH, Ou HY, Lazo MZ, et al. Predicting post-hepatectomy liver failure by combined volumetric, functional MR image and laboratory analysis. Liver Int. 2018;38:868e874. 29. Chapelle T, Op De Beeck B, Huyghe I, et al. Future remnant liver function estimated by combining liver volumetry on magnetic resonance imaging with total liver function on 99mTc-mebrofenin hepatobiliary scintigraphy: Can this tool predict post-hepatectomy liver failure? HPB (Oxford). 2016;18: 494e503. 30. Wong JS, Wong GL, Chan AW, et al. Liver stiffness measurement by transient elastography as a predictor on posthepatectomy outcomes. Ann Surg. 2013;257:922e928. 31. Cescon M, Colecchia A, Cucchetti A, et al. Value of transient elastography measured with fibroscan in predicting the outcome of hepatic resection for hepatocellular carcinoma. Ann Surg. 2012;256:706e712; discussion 712e713. 32. Nishio T, Taura K, Koyama Y, et al. Prediction of posthepatectomy liver failure based on liver stiffness measurement in patients with hepatocellular carcinoma. Surgery. 2016;159:399e408. 33. Miyagawa S, Makuuchi M, Kawasaki S, Kakazu T. Criteria for Safe Hepatic Resection. Am J Surg. 1995;169:589e594. 34. Mise Y, Satou S, Shindoh J, et al. Three-dimensional volumetry in 107 normal livers reveals clinically relevant inter-segment variation in size. HPB (Oxford). 2014;16:439e447. 35. Takahara T, Wakabayashi G, Konno H, et al. Comparison of laparoscopic major hepatectomy with propensity score matched open cases from the National Clinical Database in Japan. J Hepatobiliary Pancreat Sci. 2016;23: 721e734. 36. Makuuchi M, Kosuge T, Takayama T, et al. Surgery for small liver cancers. Semin Surg Oncol. 1993;9:298e304. 37. Takase S, Takada A, Matsuda Y. Studies on the pathogenesis of the constitutional excretory defect of indocyanine green. Gastroenterol Jpn. 1982;17: 301e309. 38. Imamura H, Seyama Y, Kokudo N, et al. One thousand fifty-six hepatectomies without mortality in 8 years. Arch Surg. 2003;138:1198e1206. 39. Wang XQ, Liu Z, Lv WP, et al. Safety validation of decision trees for hepatocellular carcinoma. World J Gastroenterol. 2015;21:9394e9402. 40. Lam CM, Fan ST, Lo CM, Wong J. Major hepatectomy for hepatocellular carcinoma in patients with an unsatisfactory indocyanine green clearance test. Br J Surg. 1999;86:1012e1017. 41. Andreatos N, Amini N, Gani F, et al. Albumin-bilirubin score: Predicting shortterm outcomes including bile leak and post-hepatectomy liver failure following hepatic resection. J Gastrointest Surg. 2017;21:238e248. 42. Zou H, Wen Y, Yuan K, Miao XY, Xiong L, Liu KJ. Combining albumin-bilirubin score with future liver remnant predicts post-hepatectomy liver failure in HBV-associated HCC patients. Liver Int. 2018;38:494e502. 43. Johnson PJ, Berhane S, Kagebayashi C, et al. Assessment of liver function in patients with hepatocellular carcinoma: A new evidence-based approach-the ALBI grade. J Clin Oncol. 2015;33:550e558.