The blood test scores doctors use to size up a fatty liver can pick out who is heading toward liver scarring, but cannot usefully tell anyone who should be screened for liver cancer. That is the result of a study of 853,131 US Veterans Affairs patients published September 21, 2026 in JAMA Internal Medicine, which tracked what happened to them over the following decade.
If you have been told you have a fatty liver, you can ask your doctor what your FIB-4 or SAFE score is and whether it means you need repeat testing for liver scarring. What that number cannot do is settle whether you need liver cancer surveillance. A reassuring score is not a clean bill of liver health, and that screening question stays with your doctor.
Can a blood test catch liver damage before symptoms appear? A newer kind of test was built to try.
Key takeaways
- Among 853,131 veterans with fatty liver, 33,794 (3.96%) developed cirrhosis within 10 years and 2,978 (0.35%) developed liver cancer, according to the study in JAMA Internal Medicine.
- The SAFE score was the most useful of the nine tested for predicting cirrhosis, while the scores showed what the authors call minimal utility for guiding liver cancer screening.
- The group studied was 92.7% male with a median BMI of 31.3, so how these scores behave in women and in ordinary primary care is still unproven.
What the study found
Researchers used national Veterans Affairs records to find adults whose fatty liver was confirmed on imaging between 2008 and 2020, excluding anyone with viral hepatitis or another primary liver disease. The median age was 61, median BMI was 31.3, and 275,898 people (32.3%) had diabetes. Nine common risk scores were calculated at that first scan, and people were followed until cirrhosis, liver cancer, death, or the end of follow up.
Over 10 years, 33,794 people (3.96%) developed cirrhosis, roughly 4 in every 100. Liver cancer was far rarer: 2,978 people (0.35%), or about 3 to 4 in every 1,000.
FIB-4, APRI and SAFE sorted cirrhosis risk best, and SAFE was the most clinically useful. At a score of 29.5, matching a 2.5% chance of cirrhosis over 10 years, the score found about 1.9 extra people who went on to develop cirrhosis for every 100 flagged as at risk, counting the cost of the false alarms it also produced.
For liver cancer, the same approach nearly flatlined. At a 0.25% ten year risk cutoff, the best score added about 1.6 correct catches per 1,000 people, a margin the authors describe as minimal utility for deciding who gets screened.
Dr. Kumar’s take
The most important word in this paper is “utility.” SAFE, Tate and FIB-4 were reported as the best of the nine at separating higher from lower liver cancer risk, and a wire summary could turn that into a headline saying blood scores predict liver cancer. Ranking first in a weak field is not the same as being good enough to act on. Once the authors asked whether using the score beats not using it after counting the false alarms, the liver cancer answer collapsed to almost nothing.
This is the thing patients most often get wrong about FIB-4. It is a triage number with one job, sorting who needs a closer look at liver scarring. It was never built to answer the cancer question, and this cohort of nearly a million people confirms it does not. A low FIB-4 tells you something real and narrow. It does not tell you your liver is fine.
The caveat a short summary will skip is who these people were: a veterans cohort, 92.7% male, median BMI 31.3, diabetes in about a third. Current guidelines favor FIB-4 as the first screening test in primary care, and the commentary published alongside the study notes primary care is under growing pressure to identify high risk patients. Whether SAFE really outperforms FIB-4 in women, or in a primary care population that looks nothing like this one, is not settled here.
Cases here were also counted from diagnosis codes and registry records rather than by a study team examining every patient, which is normal at this scale and means some were likely missed or mislabeled.
What it means for you
If you have a fatty liver on an ultrasound or CT report, ask what your FIB-4 score is and what the plan is for rechecking it. This study supports using these scores to decide who needs repeat testing for scarring over time.
Do not read a reassuring score as clearance from liver cancer risk. That decision rests on other factors, most importantly whether you already have cirrhosis, and it belongs with your doctor.
Fatty liver is also a condition where daily habits carry real weight, and one ordinary drink has been linked to lower rates of liver disease, liver cancer and death in large population data.