A result that shows up only in the group with the worst starting numbers is the one to slow down on, because two very different things produce exactly that picture — and this study has no arm that could tell them apart, which is what a missing comparator costs.
What was done
A secondary analysis of a multicenter Japanese cohort, looking at the 182 patients at high risk of metabolic dysfunction-associated steatotic liver disease who switched from a GLP-1 receptor agonist to tirzepatide. [1] Two calculated scores were tracked over six months: the Hepatic Steatosis Index and the Fibrosis-4 index, both assembled from routine measurements rather than from imaging or a biopsy.
Patients were split into low- and high-risk groups by baseline fibrosis status. Everyone switched. There is no group that stayed on the original drug, which is the design difference between this and a study that checked whether its own method could detect anything.
What improved
Glycated hemoglobin, body weight and liver enzymes all fell significantly. The steatosis index fell in both risk groups.
The fibrosis index fell significantly only in the high-risk group, P < 0.001. And baseline steatosis was positively correlated with the change in the fibrosis index, rho = 0.234, P = 0.005 — the worse a person started, the more their score moved.
The claim worth noting
The authors report that improvements in the hepatic indices were not correlated with changes in body mass index or glycated hemoglobin, and describe the benefit as not clearly explained by weight change.
That is their phrasing and it is the interesting part, because it points at something the weight loss does not account for. It also sits in a study with no control arm, which is a weak place to establish a mechanism — the same limit as the surgery comparison where more weight bought no more stiffness improvement.
What a reader takes from it
Not a reason to switch. Nobody in this study chose tirzepatide from a price page, and a before-and-after in 182 people who all did the same thing is the weakest form of comparison there is.
What is worth keeping is the check: when an effect appears only in the subgroup that started worst, ask what the people who started worst would have done anyway. It applies here, and it applies to every index calculated from routine bloods that somebody has sorted patients by.