Article
Impact of resurgery on the survival of recurrent glioblastoma. Introducing a semi-automatically optimized matched-pair analysis
Search Medline for
Authors
-
T. Behm
- Klinik für Neurochirurgie, Universitätsmedizin Göttingen
-
C.K. Krauleidies-Behm
- Fakultät Technik und Informatik, Hochschule für Angewandte Wissenschaften, Hamburg
-
F. Stockhammer
- Klinik für Neurochirurgie, Universitätsmedizin Göttingen; Klinik für Neurochirurgie, Campus Virchow-Klinikum, Charité - Universitätsmedizin Berlin
| Published: | June 4, 2012 |
|---|
Outline
Text
Objective: The role of resurgery in recurrent glioblastoma is still a matter of debate. There are no controlled randomized trials, therefore, only retrospective data are available. The highest level of evidence in retrospective analysis can be achieved by matched pair analysis. However, individual compilation of matched pairs is prone to selection bias. We introduce a tool for semi-automated matched-pairs, investigating the role of surgery for recurrent glioblastoma.
Methods: We performed a retrospective mono-institutional cohort analysis of recurrent glioblastoma treated from 1999 to 2005 at neurosurgical department of Virchow Klinikum, Berlin. For generating matched pairs a virtual basic add-in based and in Microsoft Excel (Microsoft corp., Office 2003) implemented tool was created. Different matchings regarding age, gender, resectability, time to progression, adjuvant local and systemic therapy, localization of tumor, Karnofski Performance Score (KPS) at time of diagnosis, first surgery, recurrence and resurgery were done and statistically evaluated, using Kaplan-Meier survival curves.
Results: 162 recurrent glioblastoma were diagnosed from 1999 to 2005 by primary resection. In 102 cases tumor resection was feasible, but only in 29 resurgery was performed. All 102 patients received chemotherapy after recurrence, thus 73 were treated by chemotherapy only. Semi-automatically optimized analysis matching pairs for age and KPS at tumor recurrence was applied. Median deviation in matched pairs is 0.02 years for age and 0% for KPS. Kaplan-Meier analysis revealed no significant difference in median OS of 45.9 weeks for resurgery group and 32.4 weeks for conservative treatment.
Conclusions: The developed tool for matching pairs is well-working, easy to use and free of actual selection bias. It allows exact matchings with a minimum of deviation between the groups. In our small cohort of 29 matched pairs no statistical benefit of resurgery as additional treatment in case of tumor recurrence could be shown. In order to constitute results of higher power, this method should be applied to a larger retrospective cohort.
