Previous studies have shown that depression (13-17%, measured with the HADS), anxiety (30-35%, measured with the HADS), and subjective cognitive impairment (80%) frequently occur in glioma patients.(8, 9) Numerous factors can be the causative or contributing factor of these impactful symptoms in glioma patients,(27, 28, 38) including AEDs.(11, 15, 21-23) The above mentioned neuropsychiatric symptoms are commonly reported as adverse effects of AEDs and glioma patients seem to be more vulnerable for adverse drug reactions of AEDs compared to patients with non-BTRE.(10, 14, 26) Therefore, we hypothesized that AED use is independently associated with self-reported depression, anxiety, and subjective cognitive impairment in glioma patients. In addition, we hypothesized patients on LEV would have an increased risk for depression and anxiety, while patients on VPA would have an increased risk for subjective cognitive impairment. The findings in this study, however, do not support any of the three hypotheses. Although we found that the prevalence of depression was significantly higher in patients using AEDs compared to patients not using AEDs, this effect disappeared after adjustment for potential confounders, suggesting that the risk of depression is caused by other factors than AED use. Thereby, a lack of sufficient statistical power might have played a role in the absence of a statistically significant difference between AED types.
LEV has generally become one of the preferred AEDs in glioma patients due to the lack of any known pharmacological interactions.(10) A perceived higher risk of psychiatric adverse effects in patients on LEV is a concern of physicians and sometimes a reason to choose another AED over LEV.(18, 39) Similar considerations apply to VPA with regard to a perceived higher risk of cognitive adverse effects.(22) Our data showed that the risk of having depression, anxiety, or subjective cognitive impairment does not significantly differ between patients on LEV, VPA, other AEDs and patients not using AEDs. Therefore, choosing certain AEDs over others or withholding AEDs in order to reduce the risk of depression, anxiety, or subjective cognitive impairment does in general not seem to be justified by our results. Nevertheless, on an individual basis different choices can be made.
Our results are in contrast with other studies in brain tumour patients, that demonstrated that LEV had an increased risk for psychiatric adverse effects, including anxiety.(16-18) This might be partly due to differences in patient populations,(17) the instrument used for measurement of anxiety,(16, 18) and/ or adjustment of different confounding variables.(16-18) This does not fully explain the differences and it remains unclear why certain confounding variables in other studies, such as a tumour in the frontal lobe,(16-18) were not related to depression and/or anxiety in our study. Different factors were associated with different outcomes in our study: prescription medications (excluding AEDs) with >1% risk of depression as adverse effect and poor performance status seemed to be the most important contributing factors for developing depression, while this was a history of mood disorder treatment in case of anxiety, and seizure severity in case of subjective cognitive impairment. Particularly the use of prescription medications (excluding AEDs) with >1% risk of developing depression is of interest, as this could be managed by a physician. Replacing medication with a relevant risk of depression as adverse effect, for medication with a low or no risk, should be considered at a low threshold in glioma patients with depressive mood symptoms. For instance, a dopamine-antagonist such as metoclopramide, which has a >1% risk of depression as adverse effect, can be exchanged for a 5HT3-antagonist like ondansetron as anti-emetic prophylaxis for chemotherapy induced nausea and vomiting. The limited use of older AEDs, such as phenytoin and phenobarbital, which are known for their cognitive adverse effects,(23) might explain the absence of an association in our study between AED use and subjective cognitive impairment, which is in contrast to what has been reported previously.(22) LEV has even been associated with an improved verbal memory in glioma patients,(25) although cognitive functioning was measured objectively instead of subjectively as in our study. Typically, the correlation between subjective and objective measures of cognition is regarded low, with subjective cognitive symptoms being more closely related to emotional and mental symptoms.(40)
Nevertheless, our findings need to be interpreted carefully. In at least 13 patients treatment with LEV, VPA, and/ or topiramate was discontinued or adjusted due to psychiatric adverse effects related to the AED, according to the treating physician. Moreover, only 32 patients used VPA monotherapy and a lack of statistical power might have played a role in the absence of an association between VPA and subjective cognitive impairment. The prevalence of subjective cognitive impairment was twice as high in patients using VPA monotherapy (28%) compared to LEV monotherapy (14%) or no AEDs (16%). Due to the cross-sectional nature of our observational study we cannot establish or refute a definitive causal link between AED use and concurrent depression, anxiety, or subjective cognitive impairment in glioma patients. An ongoing randomized controlled clinical trial also assessing depression, anxiety, and subjective cognitive impairment in patients on LEV versus VPA may contribute to elucidate this issue (ClinicalTrials.gov Identifier: NCT03048084).
A strength of our study is that we included all types of diffuse glioma patients and did not exclude certain patients, such as patients with a (family) history of psychiatric disorder,(18) but instead included this as potential confounder. In addition, we included prescription medications other than AEDs with >1% risk of depression as a relevant confounder, which has not been reported before, and found that this was associated with a higher risk of depression. Although the non-response analysis showed that the percentage of patients in the study population with poor performance status was significantly lower and the mean age higher, the actual differences were not clinically relevant. Therefore, our results can be considered generalizable to the general glioma population.