ABSTRACT
Background The pancreas is essential for metabolic homeostasis. Alterations in morphology and parenchymal integrity may impact proper function but are not routinely used for risk stratification. Here, we propose an AI-pipeline to quantify pancreas volume and fat content from MRI to identify individuals at high-risk for cardiometabolic disease in the general population.
Methods We quantified pancreas volume (milliliters, mL) and intrapancreatic fat content (defined as fat fraction; FF, %) from MRI of UK Biobank (UKB) and German National Cohort (NAKO) participants using deep learning. We 1) analyzed differences in volume and FF across age and sex, 2) computed percentile-curves and z-scores adjusted for age and sex to identify high-risk volumes/FF, and 3) conducted Cox regression to assess associations between z-score categories (volume: reference, z=−1 to 1; low, z=<−1; high, z>1; FF: low, z<1; moderate, z=0-1; high, z>1) and incident outcomes (diabetes, major adverse cardiovascular events (MACE), all-cause mortality) after adjustment for risk factors.
Results Among 63,548 UKB and NAKO-participants (57.7±12.8 years; BMI: 26.3±4.4 kg/m2, 46.9% female), automated pancreas analysis revealed a positive association between both volume and FF and age. In 33,099 UKB-participants (median 4.8 years follow-up), z-score categories were associated with incident diabetes (low volume, aHR:1.59, 95%CI[1.20-2.11]; high FF, aHR:1.70, 95%CI[1.31-2.19]), MACE (high volume, aHR: 0.79, 95%CI[0.61-1.01]; high FF, aHR: 1.32, 95%CI[1.01-1.73]), and all-cause mortality (low volume, aHR: 1.48, 95%CI[1.16-1.90]) beyond risk factors. Adding z-score categories to a baseline model including risk factors improved discrimination of future diabetes (volume:0.781 to 0.784, p=0.004; FF:0.781 to 0.787, p<0.001) and mortality (volume:0.781 to 0.787, p<0.001)
Conclusions Deviations from normalized pancreas volume and FF predicted cardiometabolic outcomes beyond known risk factors and alcohol intake. This automated approach identifies high-risk individuals who may benefit from cardiometabolic/endocrinology referral.
Competing Interest Statement
Matthias Jung: Research Funding: Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) 518480401. National Academy of Medicine, Healthy Longevity Grand Challenge Catalyst Award Christopher L. Schlett: Consulting or Advisory Role: Bayer; Speakers Bureau: Siemens Healthineers, Bayer Health; Research Funding: Siemens Healthineers (Inst), Bayer Health (Inst) Michael T. Lu: Research funding to his institution: American Heart Association, Astrazeneca, Ionis, Johnson & Johnson Innovation, Kowa, NIH/NHLBI, Risk Management Foundation of the Harvard Medical Institutions Incorporated. Fabian Bamberg: Consulting or Advisory Role: Bayer; Speakers Bureau: Siemens Healthineers, Bracco Diagnostics, Bayer Health; Research Funding: Siemens Healthineers (Inst), Bayer Health (Inst) Vineet K. Raghu: Research Funding: Norm Group Longevity Impetus Grant, NHLBI K01HL168231, and AHA Career Development Award 935176. National Academy of Medicine, Healthy Longevity Grand Challenge Catalyst Award Jakob Weiss: Consulting or Advisory Role: Onc.AI; Research Funding: Siemens Healthineers (Inst), Bayer (Inst), Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) 525002713, National Academy of Medicine, Healthy Longevity Grand Challenge Catalyst Award
Funding Statement
This study did not receive any funding
Author Declarations
I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained.
Yes
The details of the IRB/oversight body that provided approval or exemption for the research described are given below:
Informed consent was obtained from all participants in the UK Biobank and the German National Cohort study. In addition, we received local IRB approval (IRB of the University of Freiburg: 23-1316-S1-retro and 24-1099-S1-retro).
I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals.
Yes
I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance).
Yes
I have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable.
Yes
Footnotes
Matthias Jung, MD: matthias.jung{at}uniklinik-freiburg.de, Zeynep Berkarda, MD: zeynep.berkarda{at}uniklinik-freiburg.de, Marco Reisert, PhD: marco.reisert{at}uniklinik-freiburg.de, Susanne Rospleszcz, PhD: susanne.rospleszcz{at}uniklinik-freiburg.de, Tobias Pischon, PhD: tobias.pischon{at}mdc-berlin.de, Thoralf Niendorf, PhD: thoralf.niendorf{at}mdc-berlin.de, Hans-Ulrich Kauczor, MD: Hans-Ulrich.Kauczor{at}med.uni-heidelberg.de, Henry Völzke, MD: voelzke{at}uni-greifswald.de, Katharina Laubner, MD: katharina.laubner{at}uniklinik-freiburg.de, Christopher L. Schlett, MD: christopher.schlett{at}uniklinik-freiburg.de, Michael T. Lu, MD: mlu{at}mgh.harvard.edu, Jochen Seufert, MD: jochen.seufert{at}uniklinik-freiburg.de, Fabian Bamberg, MD: fabian.bamberg{at}uniklinik-freiburg.de, Vineet K. Raghu, PhD: vraghu{at}mgh.harvard.edu
DATA AVAILABILITY
Due to the restrictions imposed by the NAKO ethics committee, the data are not publicly available but can be requested from https://transfer.nako.de/transfer/index. The data from the UKB can be downloaded upon request from https://www.ukbiobank.ac.uk. The open-source online tool for z-score and percentile calculation will be available from https://circ-ml.github.io/.
