A Trial About the Measuring Instrument

SAMARA will be reported as a trial showing semaglutide improves fatty liver disease. That is not what it was built to ask. It was built to ask whether blood tests and scans can replace a liver biopsy for deciding who to treat and whether treatment is working — and semaglutide was the tool used to answer it.^[1] The distinction changes what the result is good for.

Why the question exists

Liver fibrosis has historically been graded by biopsy: a needle through the abdominal wall, a small piece of liver, a pathologist. It is the reference standard and it is invasive, expensive, occasionally harmful, and impossible to repeat often enough to track whether a treatment is working.

Non-invasive tests were developed to avoid it. This trial used two: a liver stiffness measurement by ultrasound elastography, and a composite score built from stiffness, an enzyme level and a fat measurement. Both are already used in practice. What was missing was randomized evidence that they could select patients for treatment and then register whether that treatment did anything.

What the trial did and found

Fifty-five people meeting non-invasive criteria for at-risk fatty liver disease — a liver stiffness of at least 8 kPa and a composite score of at least 0.5 — were randomized to weekly semaglutide 2.4 mg or placebo for 52 weeks. Average age was 48.8, average BMI 40.2, and one in five had diabetes.

Gastrointestinal side effects were common in both arms and did not differ between them, which in a 55-person trial mostly tells you the trial was too small to separate them.

What 'validated' can and cannot mean here

That is a genuinely useful thing to have shown, and it is narrower than it sounds. A test that responds to treatment is a candidate for monitoring treatment. Whether it responds in proportion to what is happening in the tissue, or responds to something adjacent — liver fat falling, weight falling, enzymes falling for reasons unconnected to fibrosis — is a separate question this design cannot reach.

The composite score used here includes a fat measurement as one of its inputs. A drug that removes liver fat will move that score partly by moving one of its own components. That does not make the result wrong; it means the score is not a pure fibrosis measure and should not be read as one.

Fifty-five people

The sample deserves its own paragraph, because everything above is built on it. Fifty-five randomized, split two ways, is roughly twenty-seven per arm. At that size, each participant is worth close to four percentage points of any proportion in the table.

The liver fat result at P = 0.047 sits directly on the conventional threshold, which in a trial this size means a small number of people determined which side of it the finding landed. The composite score result at P = 0.002 is more robust, and the weight difference — 64% against 8.3% — is large enough to be beyond argument.

A trial can be well designed, honestly reported and too small to settle what it was designed to ask.

What it means practically

• For patients: it supports using blood tests and scans rather than a biopsy to decide about treatment and follow it. That is a real improvement in what getting care feels like.
• For the drug question: it is consistent with semaglutide helping this condition, and it is far too small and too indirect to establish it. Larger trials with histology exist and are the place to look.
• For anyone quoting it: the endpoint was a test score, not liver tissue, and the trial's purpose was the test.

The broader problem of markers standing in for outcomes is in a biomarker is not an outcome.