Article Text
PDF
Abstract
Background Liver stiffness measurements (LSM) is increasingly used to monitor disease progression in patients with chronic liver disease. A 20% change in LSM as the threshold of ‘clinically meaningful changes’ to predict decompensation in compensated advanced chronic liver disease (cACLD) patients was proposed, but this requires validation. Importantly, when the current LSM was known, it remained unclear how to weigh the changes from prior LSM. We aimed to compare whether serial LSM is superior to the current LSM to predict liver-related events (LRE) in cACLD patients.
Methods We included an intercontinental, multinational cohort of cACLD patients with reliable serial LSM performed. All patients were followed up until index LRE (defined as decompensation or hepatocellular carcinoma). We compared the performance of both the dynamic LSM changes and the current LSM (LSMc) in predicting LRE using Cox-regression analysis, allowing non-linear association using restricted cubic splines. Time zero of follow-up was set at the date of LSMc. Sensitivity analyses performed included: 1) subgroup analysis among cACLD patients with etiological controlled, 2) incorporating LSM slope (LSM change over time), and 3) secondary outcomes of liver decompensation.
Results 480 cACLD patients with serial LSM were included from Singapore, USA, Italy, Thailand and New Zealand. The commonest etiology of cACLD was viral (53%) and MASLD (34%). Over a median follow-up of 68 (45-92) months, 32% experienced LSM decrease to levels below 10kPa (resolved-cACLD) and 5.8% experienced LRE (IDDF2024-ABS-0119 Figure 1. At baseline, most patients had LSM1 between 10-25kPa. During follow-up, about one-third of patients achieved resolved-cACLD (defined as LSMc less than 10kPa). Resolved-cACLD were more likely to be non-diabetic and had better liver function. While a higher value of the current LSM (LSMc) was associated with higher LREs, LSM changes over time (LSM slope) were not associated with LRE (IDDF2024-ABS-0119 Figure 2. While the risk of LRE increases with the prior LSM (LSM1), this association begins to dissociate when the prior LSM goes beyond 25kpa. In contrast, the risk of LRE continues to increase with the increasing ‘current’ LSM (LSMc)). In multivariable Cox regression of various subgroups, both prior LSM and LSM slope did not add predictive value to LSMc. Resolved-cACLD is clinically meaningful and associated with lower LRE and death. In patients with controlled etiology, HCC is an important contributor to index LRE.
Conclusions Once the current LSM is known, previous LSM values add little to the prediction of index decompensation in cACLD patients.
