We commend Chari et al1 for their rigorously designed prospective study establishing glycemically defined new-onset diabetes (GNOD) as a signal associated with a heightened 3-year incidence of pancreatic ductal adenocarcinoma (PDAC). Their real-time electronic phenotyping of GNOD is technically elegant and operationally scalable. However, we respectfully question the underlying assumption that GNOD represents a “high-risk state” predictive of future PDAC. The timing and distribution of cancer diagnoses in their cohort suggest an alternate—and arguably more clinically relevant—interpretation: GNOD may not precede PDAC; it may already be caused by it.
We commend Chari et al1 for their rigorously designed prospective study establishing glycemically defined new-onset diabetes (GNOD) as a signal associated with a heightened 3-year incidence of pancreatic ductal adenocarcinoma (PDAC). Their real-time electronic phenotyping of GNOD is technically elegant and operationally scalable. However, we respectfully question the underlying assumption that GNOD represents a “high-risk state” predictive of future PDAC. The timing and distribution of cancer diagnoses in their cohort suggest an alternate—and arguably more clinically relevant—interpretation: GNOD may not precede PDAC; it may already be caused by it.