Body composition, nutritional status, and inflammatory status as host phenotypes to predict outcome in pancreatic cancer
This project enables us to study the role of novel prognostic host phenotypes in patients with pancreatic cancer to predict treatment outcome and/or survival. Results of this project might lead to implementation of host phenotyping to standard clinical oncologic care alongside tumour scoring systems such as the TNM-classificationSteven Olde Damink
Patients with cancer often suffer from cachexia, a syndrome characterized by loss of body weight as a result of breakdown of skeletal muscle and fat tissue.
In cachexia, the tumour interacts with the body, causing loss of muscle, fat mass, and poor nutritional status. Cachectic patients often experience reduced physical fitness and low quality of life. Unfortunately, cachexia cannot be effectively treated and has a big negative impact on survival.
Our research group has shown that cachexia-related body composition alterations can be assessed using the CT-scan of the abdomen that is made as part of routine-care for cancer diagnosis and tumour staging. We found that low fat mass and high proportions of fat in muscles are associated with shorter survival of patients with cancer.
The aim of the currently proposed study is to define a host phenotype, meaning a combination of traits of a patient, and use it as a predictor of survival. For the host phenotype we will include data on body composition, nutritional status, and inflammation.
Aim and Objectives
We aim to define a set of patient traits within their body composition, nutritional status, and inflammatory state that is predictive of postoperative course and survival. We will study patients undergoing pancreatic resection for cancer treatment.
Our objective is to demonstrate that these non-tumour patient characteristics are predictive of overall survival, disease-free survival, and the occurrence of severe postoperative complications.
How it Will Be Done
We will study a large Dutch cohort of 1424 patients who underwent pancreatic resection for cancer. CT-scans made for diagnostic imaging will be used to determine body composition, focusing on the distribution and volume of fat and muscle mass.
The analysis will be performed using a newly developed but validated and accurate deep learning network, a type of artificial intelligence, that has been trained on over 5000 patients.
The nutritional status of patients and indicators of inflammation will be integrated with these body composition data to relate them to postoperative survival and relevant treatment outcomes such as complications after surgery.
This project will identify body compositions that are predictive of treatment outcome and survival of pancreatic cancer patients, providing valuable information for a patient’s individual prognosis.
Our approach will improve preoperative risk analyses by surgeons and support the patient’s and doctor’s shared decision-making process. In the future, comparable analyses can easily be done for more types of cancer. As such, the project has great potential for optimizing surgical anti-cancer treatment strategies and outcomes in general.
We are currently working towards implementing this tool in clinical practice, since it seems cost-effective and is more accessible than currently used methods of body composition screening.