Recent strides in the field of genomics, exemplified by initiatives like The Cancer Genome Atlas (TCGA), have illuminated the genetic landscape of endometrial carcinoma (EC). The identification of four distinct EC subtypes – POLE ultramutated, MSI-H, CNV-L, and CNV-H – based on their genetic patterns, offers a novel framework for understanding the disease biology. Building on these advancements, a proteogenomic study published by Cancer Cell has scrutinized 138 EC tumors alongside 20 samples of normal endometrial tissue using a multifaceted approach.
Among the key findings, researchers identified potential immunotherapeutic targets by pinpointing specific peptides that can predict the activity of the antigen processing and presentation machinery.
“One important clinical question is related to tumors with deficient mismatch repair with high mutational burden which are approved for immune checkpoint inhibition therapy but only about half of the patients respond to therapy” said lead author Dr. David Fenyö, “we found that endometrial tumors with deficient mismatch repair have a large variation in the levels of the proteins comprising the antigen presenting machinery.” Dr. Fenyö and his team propose that this could be a useful biomarker for predicting success of immune checkpoint inhibition therapy.
Furthermore, they found correlations between the activity of MYC and metformin treatment. This connection implies that metformin, a medication often used to manage diabetes, might hold promise for non-diabetic EC patients exhibiting elevated MYC activity.
The study also sheds light on the impact of PIK3R1 and CTNNB1 mutations. PIK3R1 in-frame indels are associated with upregulated AKT1 phosphorylation and were found to heighten the sensitivity of tumors to AKT inhibitors. The presence of CTNNB1 hotspot mutations were found to inhibit DKK-induced degradation, in turn reducing the efficacy of Wnt-FZD antagonists. Notably, the team suggests protein-based assays like IHC could serve as an effective option (when sequencing is not available) for detecting mutation status in these patients.
The team also explored deep learning, assessing if features derived from traditional histopathology images could be used to predict the molecular features of a tumor. Their algorithm achieved a high degree of accuracy as a predictive model, and the move to integrate deep learning represents an innovation for the field and could accelerate EC diagnosis and treatment selection in the future.