Fatty pancreas was found in 8 of 22 patients with pancreatic neoplastic cystic lesions, or 36.4%, in the study. But the comparison with matched controls did not show a statistically significant overall link between fatty pancreas and these lesions. The reported 95% confidence interval for the prevalence ran from 17.2% to 59.3%, a wide range that reflects uncertainty in the estimate.
How the comparison was made
The researchers set out to see whether fatty pancreas was more prevalent among patients with pancreatic neoplastic cystic lesions, or PNCL, and whether it was more common in lesions with significant risk of malignancy. They reviewed records in a retrospective, single-center matched case-control study covering 2019 to 2024. CT and endoscopic ultrasound, or EUS, were used in the assessment. The methods defined fatty pancreas as a pancreas-to-spleen attenuation ratio below 0.7, a CT-based comparison of scan density in the two organs. There is a reporting wrinkle: the abstract's results sentence describes the prevalence criterion as above 0.7.
Among 42 patients evaluated by EUS, 22 also had CT and formed the case group. They were compared with 44 controls matched two to one by age, gender and nutritional status. The controls had no major comorbidities or prior biliopancreatic diseases. Within the 22 cases, eight had serous cystadenomas with no or minimal malignant-transformation risk, while 14 had significant-risk lesions. The latter group comprised 10 branch-duct IPMNs, two cystic NETs and two mucinous cystadenomas.
The overall comparison
The study reported no significant difference in the pancreas-to-spleen ratio between cases and controls, and no significant difference in the reported measure of fatty-pancreas prevalence. The p-values were 0.496 for the ratio and 0.573 for prevalence. The investigators treated p-values below 0.05 as statistically significant and described the analysis as hypothesis-generating and exploratory. The results therefore do not support a statistically significant overall association.
After adjustment for glycemia and dyslipidemia, the matched odds ratio for fatty pancreas was 3.811. An odds ratio compares the odds of a finding in one group with those in another, but this estimate came with a very wide 95% confidence interval, from 0.438 to 33.133, and a p-value of 0.225. The result was not statistically significant and leaves the size, or even the presence, of an association uncertain.
Signals in smaller groups
In the IPMN subgroup, fatty pancreas was identified in five of 10 patients, or 50%. The 95% confidence interval for that prevalence ran from 18.7% to 81.3%. The corresponding odds ratio was 5.854, with a confidence interval from 0.643 to 53.337 and p = 0.117. The subgroup result was not statistically significant, so it does not establish a specific association between IPMN and fatty pancreas.
Among the 14 cases with significant malignant-transformation risk, six had fatty pancreas, a prevalence of 42%. Compared with 28 matched controls, the subgroup showed no notable overall difference, with p = 0.113. A separate comparison found no significant difference in fat ratios among the head, body and tail of the pancreas, with p = 0.399.
The study's more specific tail comparison offered a possible signal, but not a firm result. In the significant-risk group, the tail P/S ratio was lower than in controls, but the difference was non-significant at p = 0.059. That is above the study's 0.05 cutoff. The finding therefore needs replication and should not be treated as evidence that tail fat can identify lesions likely to become malignant.
Why the result is provisional
The study was better positioned to detect very large associations than modest ones. Its reported statistical power, meaning its ability to detect an association of a given size, exceeded 90% for odds ratios of at least 5.6, was about 70% for an odds ratio around 3.8, and fell below 50% for odds ratios of 2.5 or less. A non-significant result from this sample therefore cannot rule out a smaller association.
Several design limits also narrow the conclusion. The study was retrospective, non-randomized and single-center. Although 42 patients were evaluated by EUS, only 22 also had CT and entered the analytic case group. The lesion types were heterogeneous, which may obscure subtype-specific patterns, and residual confounding may remain despite matching and adjustment. The data do not show that fatty pancreas causes pancreatic cystic lesions, malignant transformation or disease progression.
The authors concluded that pancreatic fat content did not significantly differ between PNCL patients and the general population. They said the non-significant tail-fat signal warrants further investigation and that additional studies are needed. For now, the study provides an estimate of how often CT-defined fatty pancreas appeared in this group, not an established risk marker.
Paper data and sources
Original title: Emerging Issue - Fatty Pancreas in Pancreatic Cystic Neoplasms: Results of a Matched Case-Control Study.
Authors: Cătălina Vlăduț, Dana Bilous, Vasile Sandru et al.
Journal/Repository: Digestive diseases and sciences
Status: Peer-reviewed
First online: 2026-08-21
DOI: 10.1007/s10620-026-10174-7
Original paper · Full text