Identification of effective treatments that can safely accompany and work in tandem with the SOC is essential for improving the prognosis of GBM. Thus, in this systematic review and meta-analysis, we assessed the published literature to determine whether LEV, an AED reported to improve GBM prognosis, prolongs survival for GBM patients in a safe and generalizable manner.
Pooling of published HRs demonstrated that LEV does not have a significant effect on survival. While there was a trend towards longer survival, the pooled treatment effect was not statistically significant. Our findings contrast those of a recent retrospective study and meta-analysis by Jabbarli et al. that reported a significant survival benefit but did not grade the quality of evidence, assess for potential sources of heterogeneity or bias, or evaluate the safety of LEV with the SOC [49]. All the studies included in the aforementioned study were captured by our search strategy and selection process. However, we were also able to identify four additional investigations published between 2012-2021, two that were graded as “good” quality and two that were graded as “fair” [36, 38-40]. Of these four studies, none reported significantly improved survival, and one found LEV to be associated with significantly worse survival [38]. Furthermore, our assessment of publication bias using funnel plots indicated that results suggestive of LEV having an insignificant or deleterious impact on survival may be underreported. This was reaffirmed by two articles that found no survival benefit with LEV and were reviewed during our selection process but had to ultimately be excluded from meta-analysis because the authors did not report statistics that could be pooled [46, 50]. Thus, the treatment effect reported by Jabbarli et al., and potentially this study, is likely overestimated and not as substantial as it is currently published for all patients with GBM.
While LEV may not have a survival impact on the entire GBM population, it is important to note that GBM is a highly heterogenous disease with several distinct molecular subtypes that vary in gene expression, tumor mutational burden, epigenetic modifications, and, consequently, treatment responses [51-53]. Our meta-analysis highlighted this by concluding that there was significant heterogeneity across our pooled studies. To evaluate potential sources of heterogeneity, we performed meta-regression to identify any correlations between demographics and treatment effect and found that MGMT methylation and sex had a significant correlation. Specifically, populations with a lower rate of MGMT methylation or larger proportion of female patients had more robust differences in survival following LEV treatment.
In the original study reporting on the cellular effects of LEV on GBM, Bobustuc et al. found that LEV enhances tumor cell apoptosis to TMZ by inhibiting MGMT expression. Given this mechanism of action, LEV’s therapeutic efficacy may be enhanced in patients who are not methylated at the MGMT promoter, since those who are methylated will already have lower MGMT expression and experience the survival benefit associated with it [12]. Furthermore, sex differences in GBM incidence and survival are well reported, with being female consistently found to be associated with lower incidence and better outcome [54, 55]. More importantly, however, investigations into the potential mechanisms behind these sex differences have highlighted that increased incidence of GBM, tumorigenesis, and poor treatment responses in males is attributed to sexual dimorphism in the p53 pathway, specifically greater TP53 mutational burden and increased susceptibility to deregulated X-linked genes that negatively regulate p53 in males [56, 57]. The second main finding in Bobustuc et al.’s original study was that LEV’s inhibition of MGMT expression is dependent on the expression of p53, mSin3A, and HDAC1, meaning patients with reduced p53 viability, such as males, may not be positioned to experience the benefits of LEV treatment, as seen in our findings.
These results may highlight the potential limitations of LEV as a therapeutic modality; however, they do not encourage the abandonment of LEV as a therapy worth investigating. Instead, this study highlights the importance of first identifying the GBM patient population who should be actively sought out and targeted for adjuvant LEV treatment, in addition to those with epileptic seizures, before conducting an RCT as has been previously encouraged so that adequate controls and study designs are implemented to account for LEV’s mechanism of action. Biomarker-driven patient selection is becoming increasingly important in the study design of clinical trials to ensure accurate conclusions, especially in a disease like GBM that is not only very heterogenous but has also experienced many clinical trial failures [58, 59]. With our current trajectory towards greater utilization of personalized oncology treatments and molecular diagnoses, LEV could still have great potential for treating GBM even if only certain populations benefit, especially given its safety profile that was confirmed to be acceptable alongside the SOC in this study. Further studies correlating molecular and genomic data with the clinical effect of LEV are needed to confirm the trends identified in this study and ensure that a robust RCT study design is undertaken to truly determine whether LEV improves survival.
This review had several limitations. Although an exhaustive search strategy was designed and utilized, it is possible that relevant studies were not identified due to screening errors or inappropriate indexing. While most studies were “fair” or “good” quality, scores were most commonly marked down due to an inability to confirm the adequacy of follow-up. Given this, treatment effects may have been improperly estimated because too many patients were lost to follow-up or not followed for the appropriate duration. Furthermore, the topic of follow-up underscores another significant limitation known as protopathic bias. In this situation, a patient treated with LEV and another without may have had similar biological survival from tumor initiation to time of death. However, the patient with LEV may be documented as having longer clinical survival because their need to control seizures with LEV may have caused earlier diagnosis. Finally, while our study determined that MGMT methylation correlated with reduced LEV treatment effect, there is likely heterogeneity across studies in how methylated and unmethylated patients were categorized given that there is no universally-accepted test or consensus cut-off for MGMT status [60]. Future prospective studies will need to ensure that there is adequate follow-up, consistent MGMT status determination criteria, and properly matched and randomized patients.