Abstract A087: Distinct age-specific prediagnostic metabolic trajectories in early- and late-onset pancreatic cancer: a nationwide longitudinal cohort study
Yeo Gyeong Ko, Hee Seung LeeAbstract
Background:
Prediagnostic changes in body weight and glycemic status are well-recognized metabolic features preceding pancreatic cancer diagnosis. However, previous studies have largely described average metabolic changes, potentially masking distinct prediagnostic metabolic phenotypes. Whether these phenotypes differ according to age at disease onset remains unknown. We therefore investigated age-specific and within-age metabolic heterogeneity using longitudinal body mass index (BMI) and fasting blood sugar (FBS) trajectories.
Methods:
We conducted a retrospective nationwide cohort study using Korean National Health Insurance Service health screening data. Repeated BMI and FBS measurements obtained up to 10 years before pancreatic cancer diagnosis were analyzed. Patients were classified as early-onset (<50 years) or late-onset (>=50 years) pancreatic cancer (EOPC and LOPC), with age- and sex-matched controls. Group-based trajectory modeling (GBTM) was applied separately to BMI and FBS trajectories in the combined cohort and within each age group to evaluate age-specific and within-group metabolic heterogeneity. Case-control z-score differences (delta z) were tracked throughout the prediagnostic period to assess the timing and pattern of metabolic changes.
Results:
A total of 42,603 pancreatic cancer patients (3,199 EOPC; 39,404 LOPC) and 42,603 matched controls were included. In the combined cohort, GBTM identified four BMI and three FBS trajectory classes. A persistently increasing BMI class showed the weakest association with pancreatic cancer (OR 1.47) and was more frequent in EOPC than LOPC (6.8% vs 2.4%), whereas the gradually declining BMI class showed a stronger association (OR 2.86) and was more frequent in LOPC (22.3% vs 12.8%). Likewise, a rapidly increasing FBS trajectory was more frequent in LOPC than EOPC (9.5% vs 5.2%) and showed a strong association with pancreatic cancer (OR 3.61). Within the EOPC and LOPC groups, the numerically dominant phenotype was metabolically quiet, whereas overt metabolic derangement was confined to clinically distinct minority subgroups whose profiles differed by age. On delta z analysis, temporal patterns diverged: EOPC patients remained heavier than matched controls throughout follow-up, converging toward control levels only near diagnosis, whereas LOPC declined below controls with progressively worsening glycemia.
Conclusions:
Pancreatic cancer does not follow a single preclinical metabolic pathway but exhibits distinct age-specific metabolic phenotypes with within-group heterogeneity. Most patients were metabolically quiet, indicating that classical metabolic warning signs characterize only a minority of patients. Divergent delta z trajectories further support distinct prediagnostic metabolic evolution between early- and late-onset pancreatic cancer, providing a rationale for phenotype-based risk stratification and age-tailored early detection strategies.
Citation Format:
Yeo Gyeong Ko, Hee Seung Lee. Distinct age-specific prediagnostic metabolic trajectories in early- and late-onset pancreatic cancer: a nationwide longitudinal cohort study [abstract]. In: Proceedings of the AACR Conference on Pancreatic Cancer: New Frontiers in Biology and Therapeutic Development; 2026 Sep 25-28; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(18_Suppl_2):Abstract nr A087.