Oops, you're using an old version of your browser so some of the features on this page may not be displaying properly.

MINIMAL Requirements: Google Chrome 24+Mozilla Firefox 20+Internet Explorer 11Opera 15–18Apple Safari 7SeaMonkey 2.15-2.23

Poster display session

5578 - Which patients with recurrent Glioblastoma will require a second surgery during their treatment? A machine learning solution


10 Sep 2017


Poster display session


Alejandro Ruiz-Patiño


Annals of Oncology (2017) 28 (suppl_5): v109-v121. 10.1093/annonc/mdx366


A. Ruiz-Patiño1, A.F. Cardona2, L. Rojas3, O. Arrieta4, Z. Zatarain-Barrón4

Author affiliations

  • 1 School Of Medicine, Hospital Universitario San Ignacio-Pontificia Universidad Javeriana, 110231 - Bogotá/CO
  • 2 Clinical Oncology, Foundation for Clinical and Applied Cancer Research FICMAC, Bogotá/CO
  • 3 School Of Medicine, Pontificia Universidad Javeriana, 220246 - Bogotá/CO
  • 4 Thoracic Oncology Unit And Laboratory Of Personalized Medicine, Instituto Nacional de Cancerologia - Mexico, 14080 - Ciudad de México/MX


Abstract 5578


Deciding upon the therapeutic approach for patients with recurrent glioblastoma (Gb) is a challenge. Although a second surgery may provide effective palliation, it has yet to be established whether it prolongs survival and/or improves quality of life; previous reported data is scarce to demonstrate that reoperation is indicated for all patients with recurrence. The few studies investigating this issue are retrospective and have been conducted on small series with heterogeneous data sets. The aim of the present study was to analyze potential predictors of outcome in patients with recurrent Gb selected for second surgery.


A statistical learning model based on artificial neural networks was performed. 144 Hispanic patients with Gb were selected; included variables were age, performance status (PS), MGMT promoter methylation (MGMTmet), IDH1/2, and extent of primary surgical resection (ESR). The objective was to identify patients who were candidates for a second surgery considering multiple variable combination models (342). Based on the overall survival (OS) 17 comparisons were made to identify the best model for later validation.


41 patients (49.7%) were female, median age was 52-years old (SD+/-14.3), 63 cases (43.8%) were older than 60 years, 125 (86.8%) had a Karnofsky Performance Index (KPS)>80%, 73 (50.7%) had methylated MGMT and 124 (86.1%) underwent total or subtotal primary resection. The best predictive variables for requiring a second surgery were age, PS, extent of surgical resection and MGMT methylation status status. With an area under the curve of 0.984, combined age plus MGMTmet had a sensitivity of 78% and a specificity of 95%. Other models including MGMTmet+IDH, age+ KPS and age+KPS+ESR yielded an AUC of 0.563, 0.861, and 0.854, respectively. All differences were statistically significant with p value


The identification of patients who will require a second surgical intervention can be achieved, offering patients and clinicians an objective tool to plan and carry multiple therapeutic options.

Clinical trial identification

Legal entity responsible for the study





All authors have declared no conflicts of interest.

This site uses cookies. Some of these cookies are essential, while others help us improve your experience by providing insights into how the site is being used.

For more detailed information on the cookies we use, please check our Privacy Policy.

Customise settings
  • Necessary cookies enable core functionality. The website cannot function properly without these cookies, and you can only disable them by changing your browser preferences.