Researchers use AI to find subgroup of endometrial cancer linked to poor outcomes

Thanks to artificial intelligence (AI), Canadian researchers have identified a subgroup of endometrial cancer that is associated with a higher risk of recurrence or death. It means that physicians can better identify patients with endometrial cancer who could benefit from more comprehensive treatment, based on histopathological features.“Endometrial cancer is a diverse disease, with some patients much more likely to see their cancer return than others,” said Dr. Jessica McAlpine, professor and Dr. Chew Wei chair in gynaecologic oncology at the University of British Columbia and surgeon-scientist at BC Cancer and Vancouver General Hospital. “It’s so important that patients with high-risk disease are identified so we can intervene and hopefully prevent recurrence. This AI-based approach will help ensure no patient misses an opportunity for potentially lifesaving interventions.”Previous research by Dr. McAlpine and colleagues helped show that endometrial cancer can be classified into four molecular subtypes, each associated with a different risk profile. The World Health Organization recommends that pathologists include the molecular subtype when reporting on endometrial cancers. Dr. McAlpine and her team wanted to further differentiate within one of the subtypes, called NSMP, or no specific molecular profile. More than half of all endometrial cancer cases are in this subtype because they lack the features of the other three subtypes. Most NSMP tumours are early-stage, low-grade estrogen-driven tumours that can be treated with surgery alone, but other patients with NSMP tumours have an aggressive disease course.“There are patients in this very large category who have extremely good outcomes, and others whose cancer outcomes are highly unfavourable. But until now, we have lacked the tools to identify those at-risk so that we can offer them appropriate treatment,” said Dr. McAlpine in a UBC press release.Read: Restorative reproductive medicine: Meet the doctor and pharmacist collaborating on this approach to fertilityThe team worked with Ali Bashashati, an assistant professor of biomedical engineering and pathology and laboratory medicine at UBC, to use a deep learning AI model to analyse tissue samples of patients with NSMP tumours. The team fed 2,318 tissue images from 1,272 patients with NSMP subtype endometrial cancer into the model. It identified a new subgroup of patients with poor survival rates based on histopathology. The researchers called this group p53abn-like NSMP because of the p53abn-like histological features. Where 89.6% of patients in the other NSMP subgroup had a no death or recurrence at 10 years, only 55.7% of patients in the p53abn-like NSMP subgroup had no death or recurrence at 10 years. This association of poor outcome for patients in the p53abn-like NSMP subgroup compared to other patients in the NSMP subtype remained significant even when information on grade, stage and histology was added. Further genetic analysis suggested p53abn-like NSMP subgroup tumours were biologically distinct from other NSMP tumours. The findings demonstrate the power of deep learning models to detect differences in histopathology that would not otherwise be found. Dr. McAlpine and colleagues are now investigating how the AI model can be used in clinical practice. The model can analyse images of tissues that are already obtained by pathologists as a regular part of diagnosis.Once incorporated into clinical practice, the findings can help physicians further personalize endometrial cancer treatment based on a patient’s histopathology.“We envision that an AI algorithm, after appropriate validation, could be utilized on diagnostic biopsy specimens, along with molecular subtype markers,” the researchers wrote.Patients in the NSMP subtype could be further distinguide by whether they are in the p53abn-like subgroup or not. Those in the p53abn-like could be “directed to cancer centres for lymph node assessment, omental sampling and directed biopsies given a higher likelihood of upstaging,” and those not in the p53abn-like subgroup would be “candidates for de-escalation of treatment (e.g., simple hysterectomy in the community).” “It is possible that with further refinement and validation of the algorithm, which can be run in minutes on the diagnostic slide image, that it could take the place of molecular subtype markers, saving time and money,” the researchers wrote.The study was published in Nature Communications.Read: An imbalance in the vaginal microbiome linked to gynecological cancers

Researchers use AI to find subgroup of endometrial cancer linked to poor outcomes

Register for free access or login.

Related Topics