A blood-based prognostic liver secretome signature predicts long-term risk of hepatic decompensation in cirrhosis

INTRODUCTION

Cirrhosis is the terminal stage of progressive liver fibrosis, affecting 1–2% of the global population and accounting for 1.3 million deaths annually.^{1, 2} Median survival for persons with compensated cirrhosis is approximately 12 years, compared to only two years for those with hepatic decompensation. Accurate prediction of hepatic decompensation is an unmet need to enable identification of patients with cirrhosis who could benefit from close monitoring and timely medical interventions. Besides, risk stratification of cirrhosis patients could help inform patient selection for trials evaluating therapies to prevent hepatic decompensation. Although various clinical scores such as the albumin-bilirubin (ALBI) and fibrosis-4 (FIB-4) indices (ALBI-FIB4 score) have been proposed to predict long-term risk of hepatic decompensation,³ external validation has often shown suboptimal prognostic capability and revealed room for improvement.⁴

We previously identified and validated a hepatic-transcriptome-based Prognostic Liver Signature (PLS) that predicts long-term risk of developing hepatic decompensation in patients with hepatitis C virus (HCV)-related cirrhosis.^{5, 6} However, widespread adoption of PLS in clinical practice was limited by its biopsy-tissue-based nature, so we recently identified a blood-based surrogate marker, Prognostic Liver Secretome signature (PLSec), to facilitate broader clinical application.⁷ Herein, we aimed to characterize the performance of PLSec for prediction of decompensation risk in patients with cirrhosis from mixed etiologies.

METHODS

As previously described, 122 patients with Child-Pugh class A cirrhosis from mixed etiologies were prospectively enrolled at the University of Michigan between January 2004 and September 2006.⁸ Patients were prospectively followed and assessed for development of hepatic decompensation or HCC. Blood serum was collected at baseline and stored at −80 °C until use.

PLSec was derived by computationally converting PLS via our integrative bioinformatic pipeline, TexSEC (www.texsec-app.org) and consists of 6 high-risk-associated proteins, including vascular cell adhesion molecule 1 (VCAM-1), insulin-like growth factor-binding protein 7 (IGFBP-7), gp130, matrilysin, interleukin 6 (IL-6), and C-C motif chemokine ligand 21 (CCL-21), and 2 low-risk-associated proteins, including angiogenin and protein S. It was calculated in a semi-quantitative manner with ranging from 0 to 8 with high-risk PLSec defined as ≥4 as previously reported.⁷ The validated antibodies for the PLSec were implemented in an FDA-approved multiplex clinical diagnostic technology, xMAP platform and run on the Bio-Plex 200 systems (Bio-Rad) at UT Southwestern BioCenter.

The primary outcome was time to incident hepatic decompensation, defined as first occurrence of new ascites, hepatic encephalopathy, bleeding from gastroesophageal varices, or liver transplantation. Transplantation for HCC was regarded as a censored observation. Association of PLSec with the primary outcome was assessed by multivariable Cox regression analysis, adjusted for age and sex, and compared with those of clinical scores using time-dependent area under receiver operating characteristics curve, integrated Brier score, and Harrell’s c-index. All statistical analyses were performed using the R statistical language ver. 4.0.3 (www.r-project.org).

RESULTS

Patient demographics are summarized in Supplementary table 1. During a median follow-up of 5.5 years (IQR, 1.8–12.1 years), 29 patients developed hepatic decompensation (ascites, 13; variceal bleeding, 3; hepatic encephalopathy, 16; and/or liver transplantation, 6). No deaths before hepatic decompensation were observed. PLSec identified 29 (24%) and 93 (76%) patients as being high- and low-risk for developing hepatic decompensation, respectively (Figure 1A). High-risk PLSec score was associated with hepatic injury (AST and ALT) and function reverse (albumin and bilirubin) and alpha-fetoprotein (Supplementary table 1). HCV-infected patients tended to be classified to the high-risk PLSec group. High-risk PLSec was significantly associated with incident hepatic decompensation (adjusted hazard ratio [aHR], 3.51; 95% confidence interval [CI], 1.61–7.63; p=0.002) (Figure 1B). In subgroup analyses, the prognostic association remained significant irrespective of sex and HCV infection (Figure 1C). The PLSec showed a better discrimination compared to the ALBI-FIB4 during the first 3 years and to MELD score over the follow-up time period (Figure 1D). In multivariable Cox regression, high-risk PLSec (aHR, 2.40; 95% CI, 1.00–5.76) and ALBI-FIB4 (aHR, 2.63; 95% CI, 1.09–6.36) were independently associated with decompensation, suggesting that these variables complementarily improve prediction of decompensation.

Figure 1.

(A) Patten of PLSec protein abundance and associated clinical variables. (B) Association of PLSec with long-term risk of hepatic decompensation. (C) Association of PLSec with incident hepatic decompensation in various subgroups. The size of box is proportional to the number of patients. (D) Time-dependent AUROC of the PLAF, PLSec, ALBI-FIB-4, and MELD score. (E) Calibration plot of the PLAF score at 5 years. The grey dash line indicates ideal calibration. (F) Association of the PLAF score with long-term risk of hepatic decompensation.

HCV, hepatitis C virus; PLSec, prognostic liver secretome signature; HR, hazard ratio; CI, confidence interval; ALBI, albumin-bilirubin; FIB-4, fibrosis-4; MELD, model for end-stage liver disease; PLAF, PLSec-ALBI-FIB4.

To explore the idea, we developed a composite score, named PLSec, ALBI, and FIB-4 (PLAF) score, as follows: PLAF score = (1 for PLSec ≥4, otherwise 0) + (1 for ALBI grade ≥2, otherwise 0) + (1 for FIB-4 >3.25, otherwise 0), using validated cut-offs for each score.^{3, 4, 7} As expected, the PLAF score yielded substantially improved prognostic association (C-statistic = 0.72) throughout the observation period (Figure 1D, Supplementary table 2) with good calibration (Figure 1E). An exploratory stratification of the patients into four risk groups enabled significant and distinct prognostication that warrants further validation (Figure 1F, Supplementary table 2).

DISCUSSION

In this independent validation, we successfully confirmed prognostic capability of PLSec as well as limited performance of the ALBI-FIB4 score for prediction of decompensation risk in early-stage cirrhosis from mixed etiologies. Our observation indicates premise of exploring prognostic biomarkers to be integrated with existing clinical scores to jointly achieve more accurate risk prediction for a transformative improvement of the care of cirrhosis patients. The PLAF score represents such integrative score to further pursue this concept to overcome the limitation of saturated performance of the existing clinical-variable-based scores. These findings should be further assessed in larger cohorts with diverse etiologies and demographics.

What you need to know.

Background

Accurate prediction of hepatic decompensation will enable to identify cirrhosis patients who benefit most from close monitoring and timely medical intervention, although it is still an unmet need.

Findings

A blood-based prognostic secretome signature (PLSec) predicts long-term risk of hepatic decompensation in early-stage cirrhosis from mixed etiologies. Its integration with clinical prognostic score may further improve prognostication.

Implication for patient care

The PLSec will refine the care of compensated cirrhosis patients by guiding allocation of medical resources to those with elevated long-term risk of developing hepatic decompensation and improve patient survival.