We read with great interest a recent paper by Berendsen et al. on the prognostic relevance of epilepsy at presentation in patients diagnosed with glioblastoma (GBM).1 The authors found that epilepsy was an independent prognostic factor for longer survival in GBM patients.
We commend the authors for performing the largest study to date on the association between epilepsy at presentation and survival in GBM patients. However, we have concern regarding the statistical methods used to select the variables for inclusion into multivariate analysis. In this paper, the variables were selected based on statistically significant difference on chi-square analysis between epilepsy and non-epilepsy groups. For example, IDH1 mutational status was not included in the multivariate analysis because it did not correlate with epilepsy status (chi-square test; P = .98). We believe that this approach can be biased because variables that did not correlate with epilepsy on chi-square testing may still be a significant predictor of survival by either the log-rank test or univariate Cox regression. Our threshold for entering variables to be tested in multivariate analysis is not only demonstration of clinical significance in prior studies but also significance on the univariate survival analysis.
The reported incidence of epilepsy is higher in lower-grade gliomas (LGGs) than GBMs.2,3 Refractory epilepsy is also more common in LGGs, occurring in 30%–35%.2 Few studies have investigated the association between epilepsy and survival in LGGs.4 Consequently, we reviewed hospital records for all adult patients with histologically confirmed LGGs (grades II and III) between 2012 and 2016. This study was conducted following approval by the ethical committee and institutional review board of Xiangya Hospital. Statistical analyses were performed using SPSS 20.0 (IBM). P values <.05 were considered significant.
We identified 113 LGG patients. The median follow-up was 37 months. According to the presence of epileptic seizure at presentation, patients were divided into 2 groups: 42 patients with epilepsy and 71 patient without epilepsy. Epilepsy was not a significant predictor of progression-free survival (log-rank test; P = .094) or overall survival (OS) (log-rank test; P = .131). We also analyzed the significance of epilepsy in the 477 patients from the Cancer Genome Atlas. The median follow-up was 107 months. Epilepsy was significantly associated with longer OS (log-rank test; P = .048) on Kaplan-Meier analysis, with a median survival of 109 months (95% CI: 88–150) in the epilepsy group versus 79 months for patients without epilepsy (95% CI: 51–108). However, epilepsy was not a significant predictor of OS (P = .642) after correcting for age, KPS, tumor location, WHO grade, histological classification, radiotherapy, and chemotherapy (Table 1).
Table 1.
Multivariate analysis of prognostic factors affecting survival
| Overall Survival |
|||
|---|---|---|---|
| Factor | B | HR (95% CI) | P Value |
| Epilepsy | .642 | ||
| Age | .060 | 1.062 (1.040–1.084) | <.01 |
| KPS | −.038 | .963 (.941–.985) | .001 |
| Histologic subtype | .285 | ||
| Tumor grade | −.847 | .429 (.236–.778) | .005 |
| Radiotherapy | .530 | ||
| Chemotherapy | 1.038 | 2.823 (1.705–4.673) | <.01 |
| Tumor location | .968 | ||
Abbreviations: B, coefficient value; CI, confidence interval; HR, hazard ratio; KPS, Karnofsky performance status.
*P value < .05 for multivariate analysis.
In conclusion, the inclusion of variables into multivariate survival analysis should not be based on significance from the chi-square test. Epilepsy may not have a prognostic significance on survival in LGGs as it does in GBMs. Further studies are needed to verify this finding.
Conflict of interest statement. None declared.
References
- 1. Berendsen S, Varkila M, Kroonen J et al. . Prognostic relevance of epilepsy at presentation in glioblastoma patients. Neuro Oncol. 2015;18 (5):700–706. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Kerkhof M, Vecht CJ. Seizure characteristics and prognostic factors of gliomas. Epilepsia. 2013;54 (Suppl 9):12–17. [DOI] [PubMed] [Google Scholar]
- 3. Ruda R, Bello L, Duffau H, Soffietti R. Seizures in low-grade gliomas: natural history, pathogenesis, and outcome after treatments. Neuro Oncol. 2012;14 (Suppl 4):iv55–iv64. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Capelle L, Fontaine D, Mandonnet E et al. . Spontaneous and therapeutic prognostic factors in adult hemispheric World Health Organization Grade II gliomas: a series of 1097 cases: clinical article. J Neurosurg. 2013;118 (6):1157–1168. [DOI] [PubMed] [Google Scholar]