AI model accurately predicts endometrial cancer recurrence
Researchers from the Leiden University Medical Center (LUMC) have developed an AI model that accurately predicts the risk of endometrial cancer recurrence.
The model, HECTOR (Histopathology-based Endometrial Cancer Tailored Outcome Risk), offers better predictive power than current methods and could play an important role in the future of cancer treatment. The study’s findings have been published in the prestigious journal Nature Medicine.
Personalised treatment
Endometrial cancer is the most common cancer of the female sexual organs and mainly affects women between the ages of 55 and 80. Although the disease is often detected early and can then be successfully treated, the prognosis is much worse if the cancer recurs within five years.
Additional treatment such as radiotherapy or chemotherapy can reduce the risk of recurrence. But this can have a huge impact on patients, and not all types of endometrial cancer respond equally well to the therapy. The pathologist assesses the tumour tissue under a microscope and develops a personalised treatment plan. Lead researcher and pathologist Tjalling Bosse and colleagues investigated whether AI can predict this risk.
The role of HECTOR
To develop HECTOR, the researchers used microscopic images of tumours and data from previous clinical studies (PORTEC-1/2/3) of more than a thousand patients. Then they tested HECTOR on patient images not used in the training phase. The results were impressive: HECTOR proved highly accurate at predicting the risk of endometrial cancer recurrence, better than the current gold standard.
Cheaper
One of the biggest advantages of HECTOR is that it only needs a microscopic image and the stage of the tumour, making it cheaper than the current standard. It can also categorise patients into different risk groups (low, intermediate, high), thus offering a more personalised approach to cancer care. High-risk patients, as predicted by HECTOR, proved to benefit most from chemotherapy treatment in addition to postoperative radiotherapy. ‘And perhaps more important: patients with a low risk can be spared additional chemotherapy’, said Bosse.
Future perspective
Although additional research is needed, Bosse sees great potential in AI in oncology, ‘Our findings with HECTOR are a first indication that AI really can improve the care for endometrial cancer patients.’ He compares HECTOR’s development with self-driving cars, ‘Regulation is needed before this kind of AI can be used in healthcare but in the meantime, we can learn from HECTOR what to look out for as pathologists.’
Technical revolution in diagnostics
HECTOR is an important step in the technical revolution within oncology. Improved diagnostics will mean patients are offered the right treatment plans, which will ultimately improve their survival rate. ‘We are on the brink of a diagnostics revolution that will ultimately lead to better treatment, a higher survival rate and a better quality of life for our patients’, says Bosse.
How does HECTOR work?
Due to the selected cookie settings, we cannot show this video here.
Watch the video on the original website orThe study was made possible with funding from the Hanarth Fund and was conducted by the AIRMEC consortium (Sarah Volinsky and Jurriaan Barkey-Wolf), with contributions from the LUMC’s Department of Pathology and Radiotherapy (Nanda Horeweg), the PORTEC Trial consortium (Carien Creutzberg) and the University of Zürich’s Department of Pathology and Molecular Pathology (Viktor Koelzer).
Source: LUMC
Banner image: YouTube/Airmec_LUMC