Introduction Currently liver biopsies are used to assess extent of liver fibrosis and provide prognostic information in liver disease. Hyaluronic acid (HA) is a non-invasive serum marker that has been shown to correlate well with different degrees of hepatic fibrosis. Despite this the prognostic abilities of HA still remain to be established.
Aim In this study we aim to establish the relationship between levels of serum HA and survival, as well as determine the sensitivity and specificity of HA for predicting death in patients with chronic liver disease of varying aetiology.
Method Patients seen at the department of hepatology, Royal Infirmary of Edinburgh between 1995 and 2010 who had serum HA levels measured were followed up for death from the date of HA measurement until the 1 November 2010 by examination of clinical records on TRAK (NHS Lothian). The cumulative probability for survival from liver related death and overall deaths at 1-, 5- and 8-year follow-up was determined. Receiver operating characteristic (ROC) curves were generated to assess the sensitivity and specificity of HA levels at predicting liver-related death at 1, 3 and 5 years.
Results HA levels were available for 632 patients. The median follow-up time (from HA measurement to death or ‘last known alive’) was 2.7 years (range 0.0–8.0; IQR 0.9–4.4). Survival analysis showed that HA levels >400 μg/l are associated with a significantly lower probability of survival from liver related death compared to patients with values of <100 μg/l and 100–400 μg/l at 1 year (88%, p<0.01), 5 years (68%, p<0.001) and 8 years (49%, p<0.05). In addition, the probability of survival from liver related death at 8 years was also significantly less in patients with HA value of 100–400 μg/l compared to patients with HA <100 μg/l (83%, p<0.05). Similar results were observed when assessing probability of survival from overall deaths. The unadjusted area under the ROC curve for predicting liver related mortality for HA at 1-, 3- and 5 years was 0.89 (CI 95% 0.84 to 0.94), 0.85 (CI 95% 0.79 to 0.91) and 0.79 (CI 95% 0.71 to 0.87) respectively. Optimum cut-off value for 1-, 3-and 5 year for predicting liver death was 326 μg/l (88% sensitivity/79% specificity), 267 μg/l (74% sensitivity/79% specificity) and 203 μg/l (71% sensitivity/72% specificity) respectively.
Conclusion In this study we have shown that patients with high levels of serum HA have an increased liver and all cause mortality. In addition, HA levels are both sensitive as well as specific at determining liver related mortality. This highlights the potential use of HA as a non-invasive marker for prognosis in patients with chronic liver disease of varying aetiology.