The Community Resource in Targeted Therapies
Driving Knowledge. Empowering Change. Optimizing Outcomes.
ONCAlert | Upfront Therapy for mRCC

Gene Expression Signature Predicts Outcomes in Endometrial Cancer

Silas Inman
Published Online: 11:51 PM, Tue March 14, 2017
A 7-transcript classifiers (MS7) effectively predicted metastatic disease for patients with endometrioid endometrial cancer (EEC), with added promise shown when the gene expression signature was combined with pathologic features, according to findings presented at the 2017 SGO Meeting.
 
Expression of the MS7 signature was found to effectively predict progression-free survival (PFS). After nearly 60 months from diagnosis, approximately 80% of those with MS7-low expression were progression free. Those in the MS7-high group were significantly more likely to experience disease progression, with just 55% progression free at the same analysis (multivariate HR, 1.96; 95% CI, 1.09-3.59; P = .025). 
 
"We have developed and validated a gene expression signature for metastasis in endometrioid endometrial cancer," said G. Larry Maxwell, MD, from the Gynecologic Cancer Center of Excellence (GYN-COE) at Walter Reed National Military Medical Center. "Extreme expression of genes in the MS7 panel are correlated with progression-free survival."
 
The risk for metastasis is typically predicted by tumor grade using area under the curve (AUC) calculations based on receiver operator characteristics for patients with EEC. Additionally, the prediction of nodal metastasis has been the focus of prior gene expression tests; however, the predictive performance for these panels was not robust in independent validation, suggesting the need for a new approach.
 
For the new genetic classifier, researchers identified differentially expressed genes from The Cancer Genome Atlas (TCGA) and Gynecologic Oncology Group (GOG) databases. In the first training step, RNA sequencing was conducted on tumor samples with >70% purity from TCGA (n = 75) followed by a second training step on data from the GOG, wherein tumors were >95% pure and Affymetrix arrays were used in addition to RNA sequencing (n = 64).
 
These initial findings were validated by RNA sequencing in a larger set of 245 patients from both TCGA and GYN-COE. A second validation was conducted using Affymetrix on 81 samples from the GYN-COE. A final assessment of the relationship between the classifier and clinical outcomes was conducted on an outcomes cohort of 389 samples from TCGA. 
 
In the first training set, tumors were primarily grade 3 (G3; 64%) and stage IIIc/IV (61.3%). Myometrial invasion (MI) ≥50% was present in 40.3%. In the second training set, 57.8% of samples were G3 and 51.7% were stage IIIc/IV. MI of ≥50% was present for 30.3%. In the first validation set, 42% of samples were G3. In the second validation group, 27.2% were G3. In the validation cohorts, tumors were primarily stage I (92.6% and 85.2%) and approximately 30% were MI ≥50%.
 
When looking at single variables, the odds ratio (OR) for metastasis for MS7 was 2.72 in both training sets (P <.0001). For G3, the OR for metastasis was 3.92 in the first training set and 1.64 in the second, which was not statistically significant.
 
For two variable testing, the OR was 2.69 and 2.91 for MS7 in the two training sets and 1.12 and 0.56 for G3. For the MS7 versus pathologic features, the OR was 0.889 to 0.744 across the training and validation sets. For G3 versus pathologic features, the OR ranged from 0.560 to 0.732. For the two combined, MS7 and G3, the OR ranged from 0.756 to 0.889.
 
MS7 low expression by itself was 79% sensitive and 62% specific. To improve upon this, MS7 was combined with grade (low), for a sensitivity of 88% and a specificity of 45%. The most promise was shown for the combination of MS7 and MI along with a low intraoperative test. This combined test had a sensitivity of 100% and a negative predictive value of 100%. The specificity was 45% and the positive predictive value was 26%.
 
"Prediction of metastasis using MS7/MI may provide improve intraoperative predictive performance compared to histologic features alone," Maxwell noted. "The MS7 panel is comparable to grade 3 in prediction of metastatic disease in EEC, and the combination of MS7 and grade 3 shows some incremental improvement."
Maxwell GL, Cassablanca Y, Wang G, et al. A 7-Transcript Classifier Identifies Metastatic Disease and Worse Progression-Free Survival in Endometrioid Endometrial Cancer Patients. Presented at: The Society of Gynecologic Oncology Annual Meeting on Women’s Cancer; National Harbor, Maryland; March 12-15, 2017.

Copyright © TargetedOnc 2018 Intellisphere, LLC. All Rights Reserved.