The substudy selected an imaging population
SURPASS-3 MRI included 296 participants with type 2 diabetes and an elevated fatty-liver index from the larger open-label randomized trial. It compared tirzepatide with insulin degludec. The primary imaging comparison pooled the 10 mg and 15 mg tirzepatide groups and measured liver fat with MRI at 52 weeks. [1]
A substudy can add detail that would be difficult to collect from an entire trial. Its selection criteria also matter. The people undergoing imaging are not automatically interchangeable with every participant in the parent study or every person considering a weight-management medicine.
The liver-fat measurement
From an overall baseline liver-fat content of 15.71%, the absolute reduction was 8.09 percentage points in the pooled tirzepatide groups and 3.38 with insulin degludec. The estimated between-group difference was 4.71 percentage points. The report also examined abdominal fat compartments. Eli Lilly funded the study. [1]
A percentage-point reduction in liver-fat content is not the same as the percentage of total body weight lost. It is also different from the relative percentage decrease in the starting liver-fat value. These measures can look similar in a headline while describing very different quantities.
Imaging cannot answer every liver question
The outcome was MRI-based fat content, not biopsy-confirmed resolution of steatohepatitis, fibrosis reversal or avoidance of liver failure. Calling this a liver cure would combine several claims that the measurement does not establish. The value of an imaging study is clearer when its scope is preserved.
The active comparator matters too. The trial did not compare tirzepatide with no care or with semaglutide. A result against insulin degludec should retain that comparator when quoted. Removing it makes it difficult for a reader to understand what the randomized contrast represents.
Pooling active doses is another detail to preserve: the main imaging estimate does not describe a single target-dose group. A reader should avoid assigning the combined result to each dose separately. The table or methods can clarify which comparisons were planned and which were supporting analyses.
Use the example to recognize evidence levels
When a liver claim appears in an offer, ask what supports it: a laboratory test, a scan, a biopsy or a clinical event. Then ask whether the source studied the same population and intervention. This is a practical way to separate different kinds of evidence without dismissing useful early or supporting research.
Our SYNERGY-NASH review examines biopsy outcomes in a separate MASH trial. The ESSENCE article explains another biopsy-based study. Neither should be merged with this MRI substudy to create a claim that all liver outcomes have been established for every product.