MicroRNAs (miRNAs) are tiny non-coding RNA molecules - short sequences of genetic material that do not code for proteins but instead regulate the expression of other genes. They work by binding to messenger RNA molecules and either blocking their translation into protein or triggering their destruction, effectively turning down the activity of specific genes.
In cancer, miRNA expression patterns are often dramatically altered compared with normal tissue. Some miRNAs act as tumor suppressors - when their expression is reduced, the genes they normally suppress become overactive, driving tumor growth. Others act as oncomiRs - when overexpressed, they silence genes that normally constrain cell proliferation.
For endometrial cancer, accurate subtype classification is critical because different histological types have very different prognoses and treatment implications. Low-grade endometrioid carcinoma (EEC) has a good prognosis, while serous carcinoma (SEC), clear cell carcinoma (CCC), and carcinosarcoma (CaSa) are aggressive high-risk types. A reliable molecular marker that distinguishes these subtypes could complement traditional pathological diagnosis and guide treatment intensity.
The study enrolled 182 endometrial cancer patients representing five histological subtypes: 45 patients with low-grade endometrioid carcinoma (EEC G1), 47 with high-grade endometrioid carcinoma (EEC G3), 44 with serous carcinoma (SEC), 21 with carcinosarcoma (CaSa), 20 with clear cell carcinoma (CCC), and 5 with mucinous carcinoma (MC). This breadth of subtypes allows comparison of miRNA expression patterns across the full spectrum of endometrial cancer biology.
MiRNA expression was measured by quantitative PCR (a laboratory technique for accurately measuring RNA amounts) in tumor tissue samples. The study first performed an exploratory analysis to identify miRNAs that differed between EEC and SEC, and then validated the most promising candidates in an independent patient group to confirm the findings were not due to chance.
After identifying miR-497-5p as the most consistently differentially expressed miRNA, the study examined its association with several clinical and pathological features: tumor grade, FIGO staging (the international cancer staging system), hormone receptor status (estrogen and progesterone receptor expression), p53 mutation status, and the proliferation marker Ki-67. This comprehensive analysis mapped the biological contexts in which miR-497-5p expression is altered.
Among all microRNAs tested, miR-497-5p was the only one that consistently discriminated between low-risk endometrioid carcinoma (EEC) and the more aggressive serous carcinoma (SEC) in independent validation. This specificity - being reproducible in a separate patient group - is a key requirement for a biomarker to have clinical utility.
miR-497-5p expression was significantly lower in all three aggressive subtypes: serous carcinoma (SEC), carcinosarcoma (CaSa), and clear cell carcinoma (CCC) compared with endometrioid carcinoma (EEC). Within endometrioid tumors, expression was also lower in high-grade (G3) compared with low-grade (G1) cancers. These patterns are consistent with miR-497-5p acting as a tumor suppressor - when its expression is lost, the genes it normally restrains become active, driving more aggressive behavior.
miR-497-5p expression was also significantly lower in tumors with: advanced FIGO staging (more spread disease), negative hormone receptor status (loss of estrogen and progesterone receptors, associated with aggressive biology), positive p53 mutation status, and high Ki-67 expression (a marker of rapid cell proliferation). Each of these associations independently places miR-497-5p expression as an indicator of high-risk tumor biology.
The study evaluated whether clinical parameters alone, or combined with miR-497-5p, could accurately classify endometrial cancer subtypes using three machine learning algorithms: SVM (Support Vector Machine), Neural Network, and Random Forest. These algorithms were tested on their ability to correctly identify whether a patient had a low-risk (EEC) or high-risk (SEC, CaSa, CCC) tumor.
Using clinical parameters alone - such as patient age, tumor grade, stage, and hormone receptor status - the machine learning models achieved 60-80% correct diagnostic accuracy depending on the specific algorithm and the clinical scenario tested. This represents a meaningful level of prediction, but falls short of what would be needed for confident clinical decision-making in many situations.
When miR-497-5p expression was added to the clinical parameter models, predictive accuracy improved by 2-7 percentage points depending on the algorithm and scenario. While this improvement may appear modest, in the context of a diagnostic test that directly influences treatment planning for an aggressive cancer, even a few percentage points of additional accuracy can translate into better outcomes for a meaningful number of patients. The consistent direction of improvement across all three algorithms adds credibility to the finding.
The identification of miR-497-5p as a discriminating marker for high-risk endometrial cancer subtypes has practical diagnostic implications. Currently, subtype classification relies primarily on pathological examination of tissue samples - a process that can be challenging when tissue quality is poor or when tumors show mixed histological features. A molecular marker like miR-497-5p could provide additional objective data to support classification decisions.
The association of low miR-497-5p expression with multiple high-risk features - aggressive histological subtypes, advanced stage, hormone receptor negativity, p53 mutation, high proliferation - suggests it may capture a fundamental aspect of endometrial cancer biology. Understanding what genes miR-497-5p normally regulates, and how loss of its expression contributes to aggressive tumor behavior, could reveal new therapeutic targets.
Future directions should include larger validation studies across diverse patient populations, evaluation of miR-497-5p in blood or urine samples as a potentially non-invasive biomarker, and mechanistic studies to identify the specific genes regulated by miR-497-5p in endometrial cancer cells. If the tumor suppressor activity of miR-497-5p can be restored therapeutically, this could represent a novel treatment approach for high-risk endometrial cancer.