Many people know that obesity and family history are pancreatic cancer risk factors, but fewer are aware that a sudden onset of diabetes in someone over 50 — particularly without the usual risk factors — can actually be an early symptom of pancreatic cancer rather than a separate disease. This happens because pancreatic tumors destroy the insulin-producing cells in the pancreas, triggering diabetes.
Patients with pancreatic adenocarcinoma (PAAD) who have new-onset diabetes (NOD) tend to have worse outcomes than those without it. However, the biological mechanisms linking NOD to more aggressive cancer behavior were not well understood before this study.
The researchers compared gene expression between pancreatic cancer patients with and without new-onset diabetes to identify differentially expressed genes. From 124 metabolism-related genes linked to NOD, three machine learning algorithms — Gradient Boosting Machine, Random Forest, and XGBoost — were used to select the most predictive genes.
The final set of genes was combined into a risk score called the NOD-related metabolism signature (NRMS). This score was then validated independently in four separate patient cohorts from different databases and geographic regions, ensuring the findings were not specific to one population.
In all four validation cohorts, patients with high NRMS scores had significantly shorter overall survival than those with low scores. Time-dependent analysis confirmed the score was predictive at 1, 2, and 3 years with AUC values ranging from 0.73 to 0.85 across cohorts. The NRMS score outperformed standard clinical factors alone in predicting survival.
High NRMS scores were associated with advanced tumor stage, higher histological grade, and more extensive lymph node involvement, linking this metabolic signature to the biological aggressiveness of the cancer. The NRMS score was also an independent prognostic factor after accounting for clinical variables in multivariate analyses.
One of the most important findings was that high NRMS scores were associated with an immunosuppressive tumor microenvironment. Patients in the high-risk group had fewer CD8+ T cells (the main cancer-killing immune cells), fewer CD4+ helper T cells, and reduced natural killer cell activity. Several immune checkpoint genes were also upregulated, further suppressing immune attack on the tumor.
The low-risk group showed a more active immune environment with greater infiltration of anti-tumor immune cells. This immune landscape difference likely explains why high-risk patients tend to respond poorly to immunotherapy — their tumors have built strong defenses against immune attack through mechanisms directly linked to the diabetes-related metabolic changes.
Among the genes in the signature, ALDH3A1 stood out as particularly important. It was significantly elevated in pancreatic cancer patients with new-onset diabetes compared to those without. Higher ALDH3A1 expression correlated with faster tumor growth, more invasion, and greater immune suppression in laboratory models.
ALDH3A1 is involved in detoxifying reactive oxygen species in cells, and its high activity in the NOD-cancer setting appears to help cancer cells survive the metabolic stress of their environment while simultaneously suppressing immune recognition. This makes it a potentially attractive drug target — inhibiting it could weaken the cancer's defenses on multiple fronts simultaneously.
The NRMS signature offers two distinct clinical values: it can serve as a prognostic tool at diagnosis to identify which patients face the highest risk of rapid progression, and it can help predict which patients are most likely to benefit from immunotherapy. Both capabilities could significantly improve the personalization of pancreatic cancer treatment.
Future clinical studies should focus on whether targeting ALDH3A1 or the broader metabolic pathways captured by the NRMS signature can reverse immune suppression and improve immunotherapy responses. The findings also raise the question of whether early intervention in NOD patients to normalize metabolic activity could change the trajectory of the cancer.