Web-based app predicted mutation probability in gliomas

A novel interactive Web-based application appears to have utility in accurately identifying IDH1/2 mutation probability in gliomas, according to recent findings.

The program, developed by Craig Horbinski, MD, and colleagues at the University of Kentucky’s Markey Cancer Center, was formulated through research on a cohort of 89 patients with untreated WHO grades II to IV gliomas. Researchers also assessed an external validation cohort of 100 patients with untreated WHO grades II to IV gliomas.

Craig Horbinski

Originally, the variables considered for the prediction of the IDH1/2 mutations were patient age, location of tumor, glioblastoma multiforme diagnosis, previous history of grade II or III glioma, and immunohistochemistry results for R132H IDH1. The final prediction models were formulated using logistic regression and backward model selection through Akaike’s information criterion. Based on this model selection, tumor location was eliminated from the final model.

The researchers found that this model yielded an area under the curve (AUC) of 0.0934 (95% CI, 0.878-0.978) in the validation cohort and 0.941 (95% CI, 0.918-0.962) in the development cohort. AUC increased to 0.986 (95% CI, 0.967-0.998) upon the addition of R123H IDH1 immunostain information.

According to study results, the model exhibited an AUC of 0.947 (95% CI, 0.891-0.995) in predicting the presence of a less common IDH1 or IDH2 mutation in an R132H IDH1 immunonegative case.

Moreover, in terms of predicting IDH1/2 mutation likelihood in gliomas from The Cancer Genome Atlas, the models yielded a 94% rate of accuracy. The models have now been incorporated into the Web-based application, which is currently available.

“Currently, there are no universally accepted guidelines for when gliomas should be tested for this mutation,” Horbinski said in a press release. “Obtaining insurance pre-approval for additional molecular testing is becoming more commonplace, and this program will assist health care providers with an evidence-based rationale for when IDH1 screening is necessary.”

Disclosure: The researchers report no relevant financial disclosures.

Published by:

Sources/Disclosures

Collapse

Source: Chen L. Neuro Oncol. 2014;doi:10.1093/neuonc/nou097.