Prediction of treatment outcomes in patients with oesophagogastric cancer

Various treatments exist to remove tumours, alleviate symptoms and combat recurrence in oesophagogastric cancer patients. Survival often remains poor and treatments can come at the cost of health-related quality of life (HRQoL). Providing accurate information to patients and physicians about the outcomes of relevant treatments, such as survival and HRQoL, is crucial to determine which course of treatment is best and coincides with the patients’ preferences. However, these outcomes depend on many factors and accurately determining these outcomes beforehand is complex. The goal of this thesis is therefore to provide accurate and evidence-based information on treatment outcomes.
We developed and validated prediction models to predict overall survival in individual patients based on readily available patient, tumour and treatment characteristics. These models were aimed at patients with both metastatic and potentially curable oesophagogastric cancer. The models had a fair discriminatory ability and demonstrated a close agreement between predicted and observed survival.
Furthermore, meta-analyses were performed to determine the effect of treatment of HRQoL. In patients with metastatic disease, HRQoL was impaired prior to treatment and remained stable for most treatments. However, in potentially-curable patients short-term differences were found between treatments, favouring neoadjuvant treatment. A lasting impaired HRQoL was observed in patients that underwent oesophagectomy.
Finally, a patient-friendly web-interface and training were created to inform patients with oesophagogastric cancer about treatment outcomes in an evidence-based, precise, personalised, and tailored manner. The resulting prediction models and analyses can provide complementary information and visualisations to facilitate shared decision making in clinical practice.