Effect of liver volume in morbidity and mortality after elective transjugular intrahepatic portosystemic shunt

Abstract

Introduction

Transjugular intrahepatic portosystemic shunt (TIPS) is a minimally invasive procedure that has mostly replaced surgical portosystemic shunts for the treatment of many of the complications of portal hypertension in patients with liver cirrhosis, including recurrent variceal bleeding, refractory ascites, and hepatic hydrothorax.1 Several models have been used to predict outcomes after elective TIPS. MELD (Model for End-Stage Liver Disease) score is the most commonly used model in clinical practice to predict outcomes after TIPS and it is also used for organ allocation for liver transplantation. Despite many advantages over the Child–Pugh score, MELD score has some limitations; most prominently, the lack of accurately predicting mortality in 15–20% of patients with cirrhosis.2 The search continues for scoring systems to improve prognostic accuracy. Recent research suggests incorporation of other parameters such as the serum sodium levels, hepatic vein pressure gradient, serum albumin level, and indocyanine green clearance (a measure of liver blood flow) to the MELD score to improve the accuracy of prediction of mortality in patients with well-compensated cirrhosis.2

In patients with cirrhosis, progressive liver failure eventually leads to liver volume (LV) depletion. A small liver is a well-known poor prognostic factor in cirrhosis.3 However, LV has not been studied as a potential predictor of outcomes after elective TIPS. The purpose of this retrospective series was to assess the potential role of LV in predicting major adverse events [hepatic encephalopathy (HE) requiring hospital admission, increase in >2 points in MELD score >18 points, need for emergent orthotopic liver transplant (OLT) and/or death] within 6 months after elective TIPS.

Methods

A retrospective review of electronic medical records was performed in patients admitted for elective TIPS in a single tertiary liver transplant center between 2003 and 2009. Inclusion criteria for this study were: (1) elective TIPS for management of complications of portal hypertension including refractory ascites, hepatic hydrothorax, hepatorenal syndrome, and/or history of recurrent variceal bleeding; (2) computed tomography (CT) and or magnetic resonance imaging (MRI) performed within 3 months of the procedure; and (3) follow-up available for at least 6 months after the procedure. Patients receiving TIPS as an emergency procedure were excluded. This retrospective study was conducted in compliance with Institutional Review Board and Health Insurance Portability and Accountability Act Requirements.

LV was measured using CT (n = 69) or MR (n = 11) images transferred to a dedicated workstation (Vitrea, Vital Images, Minnetonka, MN, USA), by manually outlining the perimeter of the liver (excluding the gallbladder and inferior vena cava). In four patients the digital images were not available and LV was obtained from the radiology report. LV was calculated in cubic centimeters (Fig. 1).

TIPS procedures were performed by three different interventional radiologists with 5 years, 10 years, and 12 years of clinical experience performing this procedure. Self-expandable 10-mm Wallstents (Boston Scientific, Natick, MA, USA) were used prior to 2003 in 17 patients. Thereafter, 10-mm Viatorr stent grafts (W.L. Gore and Associates, Flagstaff, AZ, USA) were routinely inserted in 63 patients. Portosystemic gradients (PSGs) were measured prior to and after the procedure (Fig. 2). The MELD score was calculated using the standard formula:

After the procedure, MELD score was calculated using the first available data, usually within 1 month after TIPS. In some cases, MELD score was calculated later as all the laboratory data to calculate the score became available. In patients with acute decompensation after TIPS, the last MELD score prior to OLT, and/or death was calculated. Major adverse events after TIPS included: HE requiring hospital readmission, increase in ≥2 points in MELD score >18 points, need for emergent OLT, and/or death within 6 months after TIPS.

Unadjusted contrasts of continuous factors on death status or transplant status were based on Wilcoxon tests. The significance of unadjusted association between categorical variables was assessed using Fisher’s exact test. Multivariate logistic regression models were performed to predict death status in terms of LV with adjustment for covariates. Those variables with P ≤ 0.1 in the crude analysis were included in the adjusted analysis. All statistical testing was two-sided with a significance level of 5%. SAS version 9.2 (SAS Institute, Cary, NC, USA) was used throughout.

Results

Eighty patients (22 female and 58 male, aged 77–31 years, median age 51.5 years) were included in this retrospective review study. Indications for elective TIPS were refractory ascites (n = 59), recurrent variceal bleeding (n = 15), and HE (n = 6). Eight patients had OLT prior to the procedure, and TIPS was performed for refractory ascites in all of these patients.

Patient characteristics by death status are shown in Table 1. The MELD score prior to the procedure ranged from 7 to 23 (median: 14). After the procedure the MELD score ranged from 7 to 43 (median: 17). Post-MELD score varied significantly with death status (living median = 16, dead median = 20; P = 0.004). Twenty-four patients (34.8%) developed severe HE requiring hospital admission, with three patients requiring TIPS reduction. The risk of severe HE was increased in dead relative to living patients [living 16 (28.6%), dead 8 (61.5%); P = 0.05].

A multivariable logistic regression of death (yes, no) in terms of MELD score prior to TIPS, age, severe HE, and LV yielded the following independent predictors: pre-MELD score (odds ratio = 1.29, 95% confidence interval: 1.03–1.61, P = 0.03), severe HE (odds ratio = 5.39, 95% confidence interval: 1.25–23.22, P = 0.02); age and LV did not contribute significantly to the model (Table 2).

OLT was performed in 18 patients (22.5%). Among them, eight patients had already planned liver transplantation before the TIPS and OLTs were performed when suitable donors became available (1 of these patients had a previous OLT and was retransplanted). Elective OLTs were performed 5–120 days (mean: 60 days) after TIPS. In this group of patients the median MELD score was 16.5 (n = 8) prior to and 19 (n = 7) after TIPS. In 10 patients (12.5%) that were not yet scheduled to have liver transplant, emergent OLT was required 17–140 days (mean: 60 days) after TIPS, as a rescue for liver failure secondary to the procedure. In this subgroup of patients (n = 10) the median MELD score was 17 prior to and 21 after TIPS. A positive correlation was found between LV and OLT after TIPS for all patients. The median LV for patients that did not undergo OLT after TIPS (n = 62) was 1479 cm³, and was 1298 cm³ for patients undergoing OLT after TIPS (n = 18; P = 0.04). LV was not significantly different between those patients that did not receive transplantation after TIPS and patients that had planned transplantation after TIPS (n = 8; P = 0.4). However, LV was significantly lower in those patients (n = 10) who required rescue OLT after TIPS (median: 1168 cm³), than those patients that did not require OLT after the procedure (P = 0.03; Table 3).

Discussion

TIPS is an established procedure for the treatment of severe complications of portal hypertension. The accepted indications of elective TIPS include treatment of refractory ascites, recurrent variceal bleeding, hepatic hydrothorax, and in some selected cases, hepatorenal syndrome.1,4,5 Refractory ascites, the most common indication for elective TIPS, is a disabling complication of portal hypertension with 1-year mortality rates of 50–80%.6 Alternative treatments are limited to repeat paracentesis, surgical shunts, and in eligible candidates, OLT.7 Several comparative studies have demonstrated that TIPS is more effective in controlling refractory ascites that repeat paracentesis but usually with a higher incidence of HE.6,8,9 Recent meta-analysis has shown that patients receiving TIPS had a significantly better transplant-free survival at 24 months (49%) than the paracentesis group (35%),10 with TIPS reducing the need for paracentesis by ~50% but increasing the risk of encephalopathy by about twofold.1

TIPS is a safe procedure in experienced hands, however, serious complications including death can occur; most of the time from liver failure. Factors that have been associated with increased mortality after TIPS include renal insufficiency with serum creatinine >1.5 mg/dL, hyperbilirubinemia >3 mg/dL, advanced age, and poor response to TIPS.7,11 Several scores have been developed to predict morbidity and mortality after TIPS. Of all the predictive scores, the MELD score is the most commonly used in clinical practice. This score has been validated in patients undergoing elective TIPS and in patients with end-stage liver disease awaiting a liver transplant.12,13 A score >18 has been associated with poor outcomes after elective TIPS with scores >25 considered as a contraindication for the procedure in elective patients.

Traditional methods for estimation of hepatic functional reserve, including liver function tests, Child–Pugh classification, and MELD score can have limitations, and some patients may still develop liver failure after TIPS despite relatively preserved liver function by laboratory data.2,14 In patients with cirrhosis, progressive liver failure eventually leads to LV depletion, since the hepatic cells account for 70–80% of the liver parenchyma. A small LV is a poor prognostic factor in cirrhosis, often gradually decreasing in the presence of progressive liver disease.3 Theoretically, the estimation of LV using CT or MRI scans that are commonly obtained in the work-up of elective TIPS procedures may improve prediction of liver failure after TIPS, if the liver reserve is further compromised in patients with advanced cirrhosis and small LV. One potential advantage of using LV as a predictor is that it is constant and tends not to vary as much as other parameters in cirrhosis, such those used in the Child or MELD classifications.15 Data derived from imaging techniques have demonstrated that hepatic functional reserve is significantly correlated with LV.14–17 In patients with advanced cirrhosis, there is a progressive loss of LV with radiographic signs of advanced cirrhosis including atrophy of the liver, enlarging of the hepatic fissures >2 cm, regeneration of liver nodules, indented contour, and signs of portal hypertension. In the initial stages of cirrhosis, the left lateral and caudate lobe hypertrophied to compensate for the loss of volume of the right and quadrate lobe, and as the cirrhosis progresses, the whole liver atrophies dramatically.14,15 Tu et al14 estimated the hepatic functional reserve by cirrhosis grading and LV measurement using CT in patients undergoing liver resection for HCC. They found a significant correlation between CT grades and Child–Pugh classification. Ito et al18 described a MR scoring system based on volume index of the spleen; volume index of right posterior + left medial + left lateral segments; presence of ascites; and presence of varices and collaterals. They found an accuracy of 89% in distinguishing between clinical Child–Pugh grade A cirrhosis and further grades, with a sensitivity of 93% and specificity of 82%.18 Other authors have described the use of splenic volume to LV ratio as of prognostic importance in patients with primary biliary cirrhosis and used it as an index in the diagnosis of liver cirrhosis.19

Accurate preoperative measurement of LV has become an important tool in predicting the extent of hepatectomy in patients with liver cancer, estimating hepatic metastatic tumor burden, and determining segmental LV of donors for living related liver transplantation.3 LV has also been recommended to evaluate liver reserve function in the setting of acute and chronic liver disease.15 In patients undergoing liver resection, the remnant LV is closely related to the incidence of complications after liver resection. The incidence of serious complications after resection is higher in patients with remaining LV <25%.20

The current use of multidetector CT scanners with faster speed of data collection and image processing, fewer artifacts from respiratory motion and thinner slices allow more accurate three-dimensional reconstruction images and finer anatomic images even in patients with significant ascites.15 Estimation of LV by CT scans has proven to be a reliable method to estimate LV with accuracy within 5%.20

Normal LV has been studied in Western populations with an estimated mean volume of 1531 cm³ (range: 649–3558 cm³).21 In a study by Shiano et al,3 the CT LV was compared with the recipient LV at the time of OLT. The median CT LV was 1308 cm³ (range: 338–3847 cm³) for patients with hepatocellular disease (e.g., viral hepatitis, or alcohol-related), 1651 cm³ (range: 641–3861 cm³) for patients with cholestatic disease (e.g., primary biliary cirrhosis), and 1210 cm³ (range: 348–2575 cm³), for patients with cryptogenic cirrhosis. In the study by Xiang et al,15 the mean total LV of all patients with viral hepatitis cirrhosis was significantly smaller than that of the healthy controls with mean LV in Child–Pugh class A, B, and C patients of 1100.92 ± 336.68 cm³, 1043.88 ± 364.75 cm³ and 798.01 ± 203.64 cm³, respectively.15

In the present study of 80 patients undergoing elective TIPS, the median LV was 1423.5 cm³ (range: 742–3671 cm³). The smallest liver in this study was 742 cm³. It is likely that patients with smaller livers would not have been referred for elective TIPS due to a bias that deterioration might occur, or the liver dysfunction was already advanced and contraindicated elective TIPS. Other limitations of this study included its retrospective nature and the limited number of patients included. The inclusion of eight patients with planned OLT may also be another limitation because the effect of elective TIPS in this group of patients is not well known.

In most retrospective and prospective studies, the mortality rates at 6 months after TIPS ranged from 25% to 50%.7 Fifteen patients (18.7%) died within 6 months in this series. In the present study 30% of the patients developed severe HE requiring hospital admission, with three patients requiring TIPS reduction. This is consistent with larger series showing a rate of new or worsening HE from 20% to 40%.7 In a logistic model analysis, pre-MELD score and presence of severe HE after TIPS were predictors of death within 6 months after the procedure. A crude analysis found that PSG <8 mm Hg, age, or type of stent had no correlation with mortality. Other studies also have demonstrated that the presence of HE affects survival of patients with cirrhosis and patients receiving TIPS.22 Although other studies have also suggested an association between PSG <8 mm Hg after TIPS creation with increased mortality,23 no association was found between PSG prior to or after TIPS.

In conclusion, a retrospective study of 80 patients undergoing elective TIPS demonstrated that developing severe HE and increased MELD score prior to the procedure were factors predicting early mortality. Although LV had no correlation with developing major adverse events after elective TIPS, patients requiring rescue OLT for liver decompensation after the procedure had lower LV that patients that did not require OLT after TIPS, suggesting that LV may be a useful prognostic factor for predicting the need of rescue OLT after elective TIPS. Further studies on the effect of LV in patients with advanced cirrhosis undergoing elective TIPS are warranted.

Article information

References