DOI:
Original papers
DOI: 10.11152/mu-1548Good performance of liver stiffness measurement in the prediction of postoperative hepatic decompensation in patients with cirrhosis complicated with hepatocellular carcinoma
Bogdan Procopet
1,2, Petra Fischer
1, Adelina Horhat
1, Emil Mois
3, Horia Stefanescu
2, Mihai Comsa
4, Florin Graur
3, Adrian Bartos
3, Monica Lupsor-Platon
4, Radu Badea
4, Mircea Grigorescu
1,2, Marcel Tantau
1,2, Zeno Sparchez
1,2, Nadim Al Hajjar
313rd Medical Clinic, Gastroenterology and Hepatology Department, University of Medicine and Pharmacy “Iu- liu Hatieganu”, 2Regional Institute of Gastroenterology and Hepatology “Octavian Fodor”, Gastroenterology and Hepatology Department, 33rd Surgery Department, University of Medicine and Pharmacy “Iuliu Hatieganu”, 4Imaging Department, University of Medicine and Pharmacy “Iuliu Hatieganu”, Cluj-Napoca, Romania
Received 21.04.2018 Accepted 29.06.2018 Med Ultrason
2018, Vol. 20, No 3, 272-277
Corresponding author: Bogdan Procopet
3rd Medical Clinic, Gastroenterology and Hepatology Department, “Iuliu Hatieganu”, University of Medicine and Pharmacy 19-21 Croitorilor Street, 400162, Cluj-Napoca, Romania
Phone: + 40 727 343 098, Fax: + 40 364 819 765 E-mail: [email protected]
Introduction
The evaluation and management of patients with early hepatocellular carcinoma (HCC) and portal hy-
pertension (PHT) referred for surgical resection is a matter of debate. The current recommendation is that patients with early HCC, Barcelona-Clinic Liver Can- cer (BCLC) stage 0 and A without PHT should undergo liver resection when possible [1-2]. The main concern in patients with PHT remains the occurrence of post- hepatectomy liver failure (PHLF) although there is no consensus regarding its definition. The gold standard method to assess PHT is hepatic venous pressure gradi- ent (HVPG) measurement [3] which is not widely avail- able and is considered invasive [4]. When unavailable, surrogate markers of PHT (esophageal varices, spleno- megaly and low platelet count) are recommended for Abstract:
The evaluation of patients with early hepatocellular carcinoma (HCC) referred for liver resection is still a matter of de- bate. Aims: 1) to compare liver stiffness measurement (LSM) by transient elastography with hepatic venous pressure gradient (HVPG) in the prediction of decompensation after liver resection in patients with cirrhosis and early HCC; 2) to identify which definition for posthepatectomy liver failure is better associated with survival. Material and methods: Fifty-one patients (MELD score of 10±3) were included. In this group, 34 patients underwent HVPG measurement, of which 13 (38%) had clinically sig- nificant portal hypertension (CSPH) and 35 patients underwent LSM (21.8±17.9 kPa). The study’s end-points were: posthepa- tectomy liver failure (PHLF) defined according to International Study Group of Liver Surgery criteria and 3-month decompensa- tion defined as de novo ascites, variceal bleeding, jaundice, hepatic encephalopathy and acute kidney injury. The performance of LSM compared to HVPG in predicting the end-points were assessed by AUROC curves and accuracy. Results: Twenty (39%) patients developed PHLF and 15 (29%) developed decompensationat 3 months. Three-month decompensation tended to be bet- ter correlated with survival. LSM performed well in predicting decompensation at 3 months (AUROC=0.78, 95%CI: 0.63-0.94;
p=0.01), comparable with HVPG (AUROC=0.89, 95%CI: 0.79-1.00; p<0.01) (DeLong test p=0.21). LSM was not sufficiently accurate to predict PHLF. Conclusion: LSM has a similar performance to HVPG in predicting decompensation at 3 months in patients with early HCC submitted to liver resection. Three-month decompensation is better associated with survival.
Keywords: clinical significant portal hypertension; hepatic venous pressure gradient; prognosis; liver resection; liver elastography
patient selection for surgery [1], but their accuracy is inadequate [5].
Presently there is no consensus regarding the assess- ment of post-hepatectomy complications but two main perspectives on this issue are prevalent. Some studies calculate a short-term in-hospital prognosis based mainly on the “50-50 criteria” on the 5th postoperative day (in- crease of bilirubin >50 mmol/L, equivalent of 3 mg/dl, and decrease of prothrombine index to less than 50%) [6-7], while others determine a mid-term prognosis (at 3 months), assessing complications related to PHT such as ascites or PHT-related bleeding [8-9].
In the last decade liver stiffness measurement (LSM) by transient elastography has proven to be very well cor- related with HVPG [4] and presently is recommended for the selection of patients at risk of clinical significant PHT (CSPH) [3], defined as HVPG ≥10 mmHg and which implies an increased risk of decompensation in patients with cirrhosis [10].
Although there are 8 studies that have validated the use of LSM in the prediction of PHLF [11-18], only one compared LSM with the direct intraoperative measure- ment of the portal pressure [18]. The direct comparison between LSM and HVPG in this clinical scenario was performed in 97 patients and confirmed the good ability of LSM to identify patients with CSPH, but in this report the patients with CSPH were excluded from hepatic re- section [19].
The primary aim of our observational study was to compare LSM with HVPG in the prediction of PHLF and 3 months decompensation in patients with cirrhosis and early HCC treated by hepatic resection. The secondary aim was to analyze which end-point (PHLF and 3 months decompensation) is better associated with survival.
Material and methods Patients
Between May 2016 and February 2018, 169 patients with HCC were diagnosed in our tertiary healthcare center. Among them 68 (40%) were diagnosed with early HCC (BCLC A) and were considered for percutaneous ablation or potential candidates for liver resection. As recommended by the current guidelines [1], a multidisci- plinary team establishes the allocation for different treat- ment options according to the BCLC stage, number and localization of the tumors and presence of PHT (HVPG or surrogate markers as presence of esophageal varices).
HVPG and LSM were done during the initial evaluation, before treatment allocation. In our center HVPG is not systematically used for selecting patients for surgery, but together with tumor size and localization, may contribute
to reject patients from resection. After this assessment, 51 patients were referred for liver resection and were in- cluded in this observational study.
The Ethics Committee of our institution approved the study protocol (No. 69/2016) and an informed consent was obtained in each case, according to the principles of the Declaration of Helsinki (revision of Edinburgh, 2000).
HVPG measurement
Under local anesthesia and ultrasonographic guid- ance, a 9F venous catheter introducer (St. Jude Medical, Minnesota, USA) was placed in the right internal jugular vein using the Seldinger technique. Thereafter, a 7F bal- loon-tipped catheter (Edwards Lifesciences, Irvine, CA, USA) was advanced into the right hepatic vein to meas- ure wedged and free hepatic venous pressures (WHVP and FHVP, respectively). HVPG was calculated as the difference between WHVP and FHVP and measured in triplicate. Clinically significant portal hypertension was defined as HVPG ≥10mmHg [10]. The operator was blinded to transient elastography results.
Liver stiffness measurement by Transient Elastography
During the same hospitalization, after an overnight fast, measurements of liver stiffness were performed on the right lobe of the liver through intercostal spaces on patients lying supine with the right arm in maximal abduction. The tip of the probe was placed on the skin on an intercostal space at the level of the right hepatic lobe under ultrasound guidance (to locate the liver and avoid large vessels). Ten successful measurements were performed on each patient. A success rate of 60% and an interquartile range lower than 30% were considered reliable. The investigator was blind to the HVPG result.
Surgical resection
Surgery was performed within a maximum 30 days after initial evaluation and the treatment allocation by the multidisciplinary team. Two types of resection were per- formed: minor hepatic resection (≤2 segments resected) and major hepatic resection (>2 segments resected) de- pending on the localization, number and size of tumors.
All the procedures were anatomical resections performed mainly by open surgery (only a small minority were lapa- roscopic). In selected cases intraoperative ultrasound was performed to assess the vascular relation with the tumor or the section plane.
End-points of the study
In the present study two end points were considered:
PHLF and 3-month decompensation.
Post-hepatectomy liver failure (PHLF) was defined and graded according to the International Study Group of Liver Surgery (ISGLS) criteria [7]. ISGLS criteria are
based on the ability of the liver to preserve its synthetic, excretory, and detoxifying functions, and rely mainly by an increased INR and hyperbilirubinemia after postoperative day 5. The PHLF is graded as follows: grade A (abnormal laboratory parameters but requiring no change in the clini- cal management of the patient); grade B (deviation from the regular clinical management but manageable without invasive treatment) and grade C (deviation from the regu- lar clinical management and requiring invasive treatment).
However, according to ISGLS recommendations, diuretic treatment or vasoactive drugs are also included in “devia- tion from regular clinical management” and, therefore, pa- tients who had decompensated ascites without “50-50 cri- teria” were still considered to have PHLF. The end-point of the study was grade B and C PHLF.
Three-month decompensation was defined as de novo ascites, PHT-related bleeding, jaundice (serum bilirubin
>3 mg/dl), hepatic encephalopathy and acute kidney inju- ry occurring during the first 3 months after surgery [8-9].
Statistical analysis
Quantitative variables were expressed as mean and standard deviation (SD), whereas qualitative variables as absolute and relative frequencies. For comparison of quantitative variables, Student’s t or Mann-Whitney test (when appropriate) were used, whereas for the qualita- tive variables Chi square or Fisher’s test were used. Cor- relations between variables were computed using the Spearman Rho coefficient. To assess the performance of LSM and HVPG sensitivity (Se), specificity (Sp), posi- tive and negative likelihood ratios (LR), accuracy and re- ceiver operating characteristic curve (AUROC) were cal- culated. The diagnostic accuracy is defined asthe number of well-classified patients. The AUROC graphics and comparison between ROC curves was done using Stata version 12 (StataCorp LP, Texas, United States). The Mc- Nemar test was used in the 2×2 contingency table for as- sessing differences in the proportion of misclassified pa- tients at risk for 3-month decompensation. The statistical analysis was performed using the SPSS software version 20 (SPSS Inc., USA).
Results
Patient characteristics
The baseline characteristics of the study group are presented in Table I. Eighteen patients (35%) had ma- jor liver surgery (resection of ≥2 segments) and only a minority (3 patients, 6%) had laparoscopic surgery. The mean duration of surgery was 110±42 minutes. HVPG measurement was performed in 34 patients of whom 13 (38%) had CSPH; 35 patients were screened for esopha- geal varices (EV), which were found in14 patients (40%).
LSM was performed in 35 patients and reliable re- sults were obtained in 32 patients, with a median value of 15.7 (70.3) kPa.
A good correlation between LSM and HVPG (Rho=0.610, p=0.001) was found and LSM performed well in identifying the patients with CSPH (AU- ROC=0.86, 95%CI: 0.70-1.00; p=0.004) (fig 1), with 19 kPa as the best cut-off according to the Youden in- dex (Se=87%; Sp=79%; positive LR=4.16; negative LR=0.16, accuracy=81.4%).
Fig 1. The performances of LSM in predicting CSPH Table I. Patient’s characteristics
Variable
Sex (m) 38 (74.5)
Etiology
alcohol 12 (23.5)
virus 33 (64.7)
others 6 (11.8)
AST (U/L) 56±61
ALT(U/L) 51±81
Bilirubin (mg/dl) 1.1±0.7
Platelets count *103/L 150±71
Albumin g/L 34±15
MELD 10±3
HVPG (mmHg) 10±6
CSPH 13 (38)
Liver stiffness kPa 21.8±17.9
Esophageal varices 14 (40)
Splenomegaly+platelets count <105 20 (50) Data are expressed as mean±SD or count (%); MELD = Model for End stage Liver Disease, AST = aspartate aminotransaminase, ALT = alanine aminotransaminase, HVPG = hepatic venous pres- sure gradient, CSPH = clinical significant portal hypertension (HVPG≥10 mmHg)
End-points
During hospitalization 20 (39%) patients developed PHLF (15 patients class B and 5 class C) as follow: PI be- low 50% (11 patients), ascites (12 patients), acute kidney injury (10 patients), variceal bleeding (3 patients) and in- creased bilirubin >3mg/dl (4 patients).
The factors associated in univariate analysis with PHLF were HVPG, AST, bilirubin, PI and the MELD score. In multivariate analysis the only independent variable associated with the development of PHLF was HVPG (OR=1.32, 95%CI: 1.08-1.62, p=0.007).
Among the included patients 15 (29%) patients pre- sented 3 months decompensation as follow: 12 patients had ascites, 3 patients variceal bleeding, 3 patients he- patic encephalopathy and 10 patients acute kidney injury of whom 3 met the hepato-renal syndrome diagnostic cri- teria. In univariate analysis only HVPG and LSM were associated with 3-month decompensation. In multivari- ate analysis HVPG was the only variable associated with decompensation (OR=1.44, 95%CI: 1.07-1.95).
During a mean follow-up of 14.5±10.2 months, 9 (17.6%) patients died and among them 5 (56%) had PHLF and 4 (44%) 3-month decompensation. None of the study end points was significantly associated with mortality.
However, 26% of patients with 3-month decompensation died in comparison with only 9% of patients without de- compensation (Fischer’s Exact test, p=0.11). PHLF was less associated with survival: 25% of patients with PHLF died in comparison with 16% of patients without PHLF (Fischer’s Exact test, p=0.48).
When Kapplan-Meier curves were analyzed, 3-month decompensation proved to have better discriminative power for survival (Log rank p=0.06) than PHLF (Log rank p=0.27) (fig 2).
The performance of LSM in the prediction of the end-points
LSM was not sufficiently accurate to predict PHLF (AUROC=0.60, 95%CI: 0.38-0.81; p=0.3) but was able to fairly predict 3-month decompensation (AU- ROC=0.78, 95%CI: 0.63-0.94; p=0.01) (fig 3).
Contrary to LSM, HVPG was able to accurately predict both end-points: PHLF (AUROC=0.85, 95%CI:
0.71-0.99; p=0.01) and 3-month decompensation (AU- ROC=0.89, 95%CI: 0.79-1.00; p<0.01). When the analy- sis was performed only in patients with both LSM and HVPG (n=27), there was no difference between LSM and HVPG for the prediction of 3-month decompensa- tion, DeLong test p=0.21 (fig 4).
Using the 21 kPa Baveno VI recommended cut-off to rule-in CSPH [3], the accuracy for the prediction of 3-month decompensation was good: 71% (p=0.08) and not significantly different from HVPG which had 79%
Fig 2. The survival of patients with and without 3-months de- compensation (A); the survival of patients with and without PHLF (B)- Kapplan-Meier analysis
Fig 3. The performances of LSM in predicting 3-months de- compensation.
Fig 4. The comparison of LSM and HVPG AUROC curves in predicting 3-months decompensation
accuracy (p=0.005) (the McNemar test for assessing dif- ferences in the proportion of misclassified patients at risk, p=0.68). None of the patients with LSM <13.6 kPa (the cut-off of rule-out CSPH) decompensated at 3 months,
while 3 of the 8 patients in the “grey zone” (between 13.6 kPa and 21 kPa) decompensated.
Surrogate markers for CSPH (the presence of EV or splenomegaly and low platelet count <100.000/L) have an accuracy in identifying patients at risk for PHLF and 3-month decompensation lower than that of HVPG and LSM (58% and 52%, respectively).
Discussions
In the present study we found that LSM had a simi- lar accuracy to HVPG measurement in selecting patients with cirrhosis complicated with HCC referred for liver resection who were at risk to develop posthepatectomy decompensation. The management of patients with early HCC is still a matter of debate. While the present guide- lines recommend HVPG measurement whenever pos- sible to identify patients at risk of complications after resection [1], some groups use a more permissive policy using the liver function assessment by MELD score rath- er than the PHT assessment by HVPG [20]. According to Cucchetti et al the HVPG-based selection of potential candidates for liver resection (using 10 mmHg as cut- off) excludes almost 25% of patients that would benefit from resection without PHLF [20]. This is somehow in contrast with the current recommendations [1] which are based mainly on the evidence generated by the BCLC group [8,9]. A possible explanation for these differenc- es may lay in different end-points. While in the BCLC group the end-point was 3-month persistent decompen- sation [8] in the Cucchetti’s et al study the end-point was PHLF according to the ISGLS criteria [20] which are mainly based on the “50-50 criteria” in the 5th postop- erative day and, thus, implies a more rapid censoring of patients. In our study we compared the two end-points from the survival perspective and we found no differ- ence. However, the 3-month decompensation seems to be better correlated with survival, as may be seen from the Kapplan Meier curves, and the explanation of non- statistical significance lies in the low number of patients.
When we analyzed strictly the “50-50 criteria” (increase of bilirubin >50 mmol/L, equivalent of 3mg/dl, and de- crease of prothrombine index to less than 50%) which are the basis of PHLF diagnosis, there was no discrimi- native power for survival (83% in patients without vs 80% in patients with “50-50 criteria”, Log rank p=0.66).
The increased performance of PHLF (according to the ISGLS recommendations) is explained by the inclusion of patients with ascites, leading to a very good (86%) concordance between both end points. Although the data were not statistically significant, due to the low number of patients included, we believe that PHT-related decom-
pensation is a more appropriate end-point for patients referred for liver resection.
Despite the current recommendations, in our center patients with CSPH may still benefit from surgery if the tumor size and localization is favorable. This is in con- cordance with the experience of other centers in which up to two thirds of resected patients do not meet the guideline resection criteria [21]. Moreover, resection in patients outside the resectability criteria has a better prognosis when compared to embolization or other treat- ments but a lower survival rate than transplantation or percutaneous ablation [21]. Interestingly, in this study the presence of portal hypertension was not significantly correlated with survival but it should be stressed that PHT was estimated only by the presence of its surrogate markers (EV or splenomegaly associated with low plate- let count). In a recent meta-analysis surrogate markers, although significantly associated with prognosis, had a lower prediction power than HVPG [5]. In our study, surrogate markers were not significantly associated with both end-points and, therefore, were unable to select pa- tients at risk for decompensation.
Because HVPG measurement is not widely avail- able and is an invasive technique, non-invasive methods that may overcome the disadvantages of surrogate mark- ers are needed. Among all non-invasive methods, LSM proved to be the most validated method able to assess the risk of CSPH in patients with advanced liver dis- ease. Recently its role in risk assessment was endorsed by the Baveno VI recommendation [3] and LS values of more than 21 kPa are able to rule-in CSPH. The perfor- mance of LSM in the prediction of PHLF were good in the previous studies [11-18], with AUROCs ranging from 0.79 [13] to 0.86 [12], but none of the studies compared LSM with HVPG and the great majority used end-points based on the “50-50 criteria”. The only study that used invasive measurement of portal pressure (intraopera- tory measurement by direct puncture of the portal vein) and 3-months decompensation as end-point, found that LSM is slightly better than portal pressure measure- ment in prognosis assessment, with an AUROC of 0.81 for LSM vs 0.71 for direct portal pressure measurement [18]. However, in this study only 50% of the patients had cirrhosis and probably only they would benefit from por- tal pressure measurement, while LSM is more useful for fibrosis staging which is also correlated with prognosis.
In a subgroup of patients with available LSM, Cucchetti et al found similar performances for LSM and HVPG in PHLF prediction, 0.81 and 0.79, respectively [20]. In our cohort, HVPG was the only variable associated with both end points while LSM was associated only with 3-month decompensation.
The small number of patients included represents the main drawback of the study. Despite this, we believe that the results are convincing because we used HVPG, which is the “gold standard” to assess portal hypertension and because we compared our end points with survival. An- other limitation is the lack of homogeneity in the man- agement of patients with early HCC which are outside the BCLC resection criteria, since this management is usually based on liver resection. However, as previously mentioned, this tendency is very frequent in many cent- ers [21] and in our center is partially explained by the limited access to liver transplantation.
Conclusions
LSM had a similar performance to HVPG in predict- ing 3-month posthepatectomy decompensation in pa- tients with early HCC suitable for liver resection. Three- month decompensation tended to be better correlated with the survival in this clinical scenario. More studies are required to confirm these findings.
Acknowledgements:
This work was supported by an internal grant of the University of Medicine and Pharmacy “Iuliu Hatie- ganu” (grant no. 4945/2016) and partially supported by the grant “Dezvoltarea infrastructurii publice de cerce- tare dezvoltare si crearea de noi infrastructuri” – SMIS 107124, Academy of Medical Sciences of Romania.
References
1. European Association For The Study Of The Liver1; Euro- pean Organisation For Research And Treatment Of Cancer.
EASL – EORTC Clinical Practice Guidelines: Management of hepatocellular carcinoma. J Hepatol 2012;56:908-943.
2. Heimbach JK, Kulik LM, Finn RS, et al. AASLD guide- lines for the treatment of hepatocellular carcinoma. Hepa- tology 2018;67:358-380.
3. de Franchis R. Expanding consensus in portal hyperten- sion: Report of the Baveno VI Consensus Workshop: Strati- fying risk and individualizing care for portal hypertension.
J Hepatol 2015;63:743-752.
4. Castera L, Pinzani M, Bosch J. Non invasive evaluation of portal hypertension using transient elastography. J Hepatol 2012;56:696-703.
5. Berzigotti A, Reig M, Abraldes JG, Bosch J, Bruix J. Portal hypertension and the outcome of surgery for hepatocellular carcinoma in compensated cirrhosis: A systematic review and meta-analysis. Hepatology 2015;61:526-536.
6. Balzan S, Belghiti J, Farges O, et al. The “50-50 criteria”
on postoperative day 5: an accurate predictor of liver failure and death after hepatectomy. Ann Surg 2005;242:824-828.
7. Rahbari NN, Garden OJ, Padbury R, et al. Posthepatec- tomy liver failure: A definition and grading by the Inter- national Study Group of Liver Surgery (ISGLS). Surgery 2011;149:713-724.
8. Bruix J, Castells A, Bosch J, et al. Surgical resection of hepatocellular carcinoma in cirrhotic patients: prognos- tic value of preoperative portal pressure. Gastroenterol- ogy1996;111:1018-1022.
9. Llovet JM, Fuster J, Bruix J. Intention-to-treat analysis of surgical treatment for early hepatocellular carcinoma: re- section versus transplantation. Hepatology 1999;30:1434- 1440.
10. Bosch J, Abraldes JG, Berzigotti A, Garcia-Pagan JC. The clinical use of HVPG measurements in chronic liver dis- ease. Nat Rev Gastroenterol Hepatol 2009;610:573-582.
11. Kim SU, Ahn SH, Park JY, et al. Prediction of postopera- tive hepatic insufficiency by liver stiffness measurement (FibroScan®) before curative resection of hepatocellular carcinoma: a pilot study. Hepatol Int 2008;2:471-477.
12. Cescon M, Colecchia A, Cucchetti A, et al. Value of Transient Elastography Measured With Fibroscan in Pre- dicting the Outcome of Hepatic Resection for. Ann Surg 2012;256:706-713.
13. Wong J, Wong G, Chan A, et al. Liver Stiffness Measure- ment by Transient Elastography as a Predictor on Posthepa- tectomy Outcomes. Ann Surg.2013;257:922-928.
14. Li C, Zhang JY, Zhang XY, Wen TF, Yan LN. FibroScan predicts ascites after liver resection for hepatitis B virus-re- lated hepatocellular carcinoma: A prospective cohort study.
Int J Surg 2015;20:21-25.
15. Cucchetti A, Cescon M, Colecchia A, et al. Adding Liver Stiffness Measurement to the Routine Evaluation of Hepa- tocellular Carcinoma Resectability Can Optimize Clinical Outcome Ultraschall Med 2017;38:515-22.
16. Fung J, Poon RTP, Yu WC, et al. Use of Liver Stiffness Measurement for Liver Resection Surgery: Correlation with Indocyanine Green Clearance Testing and Post-Oper- ative Outcome. PLoS One 2013;8:1-6.
17. Chong C, Wong G, Chan A, et al. Liver stiffness meas- urement predicts high-grade post-hepatectomy liver fail- ure: A prospective cohort study. J Gastroenterol Hepatol 2017;32:506-514.
18. Rajakannu M, Cherqui D, Ciacio O, et al. Liver stiffness measurement by transient elastography predicts late posth- epatectomy outcomes in patients undergoing resection for hepatocellular carcinoma. Surgery 2017;162:766-774.
19. Llop E, Berzigotti A, Reig M, et al. Assessment of portal hypertension by transient elastography in patients with compensated cirrhosis and potentially resectable liver tu- mors. J Hepatol 2012;56:103-108.
20. Cucchetti A, Cescon M, Golfieri R, et al. Hepatic venous pressure gradient in the preoperative assessment of pa- tients with resectable hepatocellular carcinoma. J Hepa- tolv2016;64:79-86.
21. Roayaie S, Jibara G, Tabrizian P, et al. The role of hepatic resection in the treatment of hepatocellular cancer.Hepatol- ogy 2015;62:440-451.
