Personalized diagnostics for endometrial cancer to optimize local decision-making and patient triage.

The rising incidence of endometrial cancer places an increasing burden on public health. The current diagnostic process lacks proper risk stratification, leading to the overuse of invasive procedures. To solve this problem, a diagnostic algorithm is needed that (1) provides an outcome detailed enough for decisions on further testing and treatment, (2) offers reliable individualized predictions, and (3) is applicable in diverse centers.
I propose a two-step pre-operative algorithm consisting of risk models based on ultrasound: the first to distinguish between four classes of endometrial pathology in symptomatic patients and to indicate treatment or the need for invasive diagnostics; the second, in case of malignancy, to determine whether lymphadenectomy is needed.
Innovative aspects are the multi-category outcome; the use of ultrasound in a risk model to predict lymph node metastasis; the use of a molecular predictor; and the novel methodology for multicenter studies. I have unique access to the largest existing international database (18 diverse centers) on endometrial cancer. Ultrasound predictors were measured according to strict definitions agreed upon by leading experts. I aim to valorize the results through implementation in ultrasound machines.
The models are expected to facilitate the triage of patients to minimize costly invasive procedures in patients with low-risk disease, and encourage referral for specialized care for high-risk patients.