BACKGROUND: Pancreatic cancer is often accompanied by cachexia, a syndrome of severe weight loss and muscle wasting. A suboptimal response to nutritional support may further aggravate cachexia, yet the influence of nutrition on protein kinetics in cachectic patients is poorly understood. METHODS: Eight cachectic pancreatic cancer patients and seven control patients received a primed continuous intravenous infusion of l-[ring-(2)H5]phenylalanine and l-[3,3-(2)H2]tyrosine for 8 h and ingested sips of water with l-[1-(13)C]phenylalanine every 30 min. After 4 h, oral feeding was started. Whole body protein breakdown, protein synthesis, and net protein balance were calculated. Results are given as median with interquartile range. RESULTS: Baseline protein breakdown and protein synthesis were higher in cachectic patients compared with the controls (breakdown: 67.1 (48.1-79.6) vs. 45.8 (42.6-46.3) micromol/kg lean body mass/h, P = 0.049; and synthesis: 63.0 (44.3-75.6) vs. 41.8 (37.6-42.5) micromol/kg lean body mass/h, P = 0.021). During feeding, protein breakdown decreased significantly to 45.5 (26.9-51.1) micromol/kg lean body mass/h (P = 0.012) in the cachexia group and to 33.7 (17.4-37.1) micromol/kg lean body mass/h (P = 0.018) in the control group. Protein synthesis was not affected by feeding in cachectic patients: 58.4 (46.5-76.1) micromol/kg lean body mass/h, but was stimulated in controls: 47.9 (41.8-56.7) micromol/kg lean body mass/h (P = 0.018). Both groups showed a comparable positive net protein balance during feeding: cachexia: 19.7 (13.1-23.7) and control: 16.3 (13.6-25.4) micromol/kg lean body mass/h (P = 0.908). CONCLUSION: Cachectic pancreatic cancer patients have a higher basal protein turnover. Both cachectic patients and controls show a comparable protein anabolism during feeding, albeit through a different pattern of protein kinetics. In cachectic patients, this is primarily related to reduced protein breakdown, whereas in controls, both protein breakdown and protein synthesis alterations are involved.