A 30% fall in liver stiffness is a weak proxy for fibrosis regression

Design

prospective analysis of paired elastography and liver biopsy at two time points within a randomised phase 2b trial, with a separate external validation cohort

Population

160 adults with biopsy-proven MASH and stage 2-3 fibrosis (64% female, median age 56, median BMI 36.5 kg/m2); validation cohort of 48 from the USA and Singapore

Primary outcome

fibrosis regression without worsening MASH on biopsy

Effect

AUC 0.68 (95% CI 0.58-0.77) for a 30% or greater fall in liver stiffness; adjusted OR 4.23 (95% CI 1.79-10.38); validation AUC 0.62 (95% CI 0.48-0.76)

The AASLD position that a decline in liver stiffness on vibration-controlled transient elastography can serve as a non-invasive endpoint for treatment response in MASH has been widely adopted and thinly evidenced. This study tested it against the thing it stands in for. One hundred and sixty adults with biopsy-proven MASH and stage 2-3 fibrosis, enrolled in a phase 2b placebo-controlled trial of the FGF21 analogue pegozafermin, had contemporaneous elastography and liver biopsy at two time points. Median age was 56 and median body mass index 36.5 kg/m2.

A relative fall of 30% or more in liver stiffness detected fibrosis regression without worsening MASH with an area under the curve of 0.68 (95% CI 0.58 to 0.77). Adjusted for age, sex, type 2 diabetes, body mass index and ethnicity, the same threshold was independently associated with regression, odds ratio 4.23 (95% CI 1.79 to 10.38), P=0.001. In a separate validation cohort of 48 patients from the United States and Singapore the area under the curve was 0.62 (95% CI 0.48 to 0.76).

The two results say different things and both belong in the reading. The odds ratio confirms the direction: patients whose stiffness falls by 30% are genuinely more likely to have regressed. The area under the curve says that as a test for any individual patient it is not much better than a coin weighted slightly in the right direction, and in validation the interval reached down to 0.48, which is chance.

In practice that means using the number as one input rather than as the answer. A 30% fall in a patient on treatment is encouraging and reasonable to act on when it agrees with weight, glycaemic control and aminotransferases. It is not sufficient on its own to declare fibrosis regression, and a patient whose stiffness has not fallen by 30% has not necessarily failed to regress. The authors say so plainly: better response biomarkers are needed.

• Use a 30% fall in liver stiffness as supporting evidence, not as proof of fibrosis regression
• Read it alongside weight change, glycaemic control and aminotransferases rather than in isolation
• Repeat elastography under the same conditions, fasted and with the same probe, since technique and body habitus move the number
• Do not tell a patient their fibrosis has improved on elastography alone
• Remember the cohort was trial patients with stage 2-3 fibrosis and a median BMI of 36.5; performance in milder disease is untested here

The statistics, in plain English

An area under the curve of 0.5 is a coin toss and 1.0 is perfect; 0.68 means that given one patient who regressed and one who did not, the test would rank them correctly about two-thirds of the time. The validation figure of 0.62 with an interval from 0.48 to 0.76 includes chance entirely, which is why the paper is cautious. The odds ratio of 4.23 looks much more impressive than the AUC, and both are correct: a marker can be strongly associated with an outcome at group level while still classifying individuals poorly. That gap is exactly the reason a surrogate endpoint needs testing rather than assuming.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.