New-onset psychogenic nonepileptic seizures after intracranial neurosurgery: A meta-analysis

S


Introduction
Functional neurological symptom disorder refers to a state in which the normal functioning of brain networks is affected, rather than any structural abnormalities.One of its manifestations is a condition termed psychogenic non-epileptic seizures (PNES), also known as functional or dissociative seizures [1].These seizures are characterized by clinically similar events to epileptic seizures, although they do not have characteristically abnormal ictal electroencephalogram activities [2], which supports the theory that their occurrences are psychological in nature rather than epileptic.Conversion disorders are one of the most prevalent psychiatric disorders; however, other comorbidities such as depression, anxiety disorders, post-traumatic stress disorder, or personality disorders can also co-occur.However, a history of physical or sexual abuse might be considered a risk factor for PNES development, and the majority of patients are reported to be adult women [2].Thus, history of previous trauma or abuse might be higher in people with PNES than in those with epilepsy, and may infer a causative factor [3]. Head injuries, however, were also significantly associated with PNES, suggesting that brain injury may also be a biologically relevant factor its development [4].Nevertheless, in the majority of patients, a number of interacting factors can be identified, and single mechanisms or contributing factors that are sufficient to explain PNES in every case are absent [5].
The most recent systematic review indicated that the incidence and prevalence estimates of PNES are reported to be 3.1 (1.1-5.1) and 108.5 (39.2-177.8)per 100,000 population, respectively [6].However, the true epidemiology and neurobiology of this condition remain unknown [6,7].While approximately 14 million patients require neurosurgical care annually [8], the frequency of PNES following neurosurgical procedures has not been thoroughly examined.
This study aims to conduct a systematic and critical review of the existing literature.We aimed to provide comprehensive analytical estimates regarding the occurrence of PNES based on data sourced from studies focused on intracranial neurosurgery, particularly those pertaining to epilepsy surgery.In turn, this could potentially affect the incidence and prevalence of PNES.This approach may provide insights into the new onset of PNES, specifically after neurosurgical procedures.

Search strategy and selection criteria
This meta-analysis was carried out following the methodological framework outlined in the PRISMA guidelines.Throughout the analysis, two reviewers (GK and DK) examined the titles, abstracts and full-texts of all potentially eligible publications with assistance from a third reviewer (ZU), using broad criteria to allow for the inclusion of any potentially relevant study for further evaluation.They searched four bibliographic databases -PubMed, Ovid, CINAHL, and Cochrane Library-from database inception until December 14, 2023, using a combination of keywords related to PNES and intracranial surgeries, including epilepsy surgeries, without restrictions on language or publication date.Full details of the search strategies are in the appendix (S1 Appendix).Additionally, all co-authors performed manual searches of the reference lists of the included papers and individually identified review articles on the topic to identify additional eligible articles.All the included studies utilized a retrospective observational design with no restrictions, providing data on the occurrence of PNES in patients who underwent various types of intracranial surgery, including epilepsy surgery, but were not preoperatively positive for PNES.Full-text articles were assessed for inclusion by the same reviewers.The authors were not contacted for further information.Studies were included if they reported intracranial/epilepsy surgery and PNES confirmed by video-EEG and by clinical assessment, medical records, or databases with a stated diagnosis.Articles were excluded if they met any of the following criteria: case reports, reviews, systematic reviews, studies lacking clear reporting on PNES following surgery, studies reporting on presurgical PNES only, or studies not specifically reporting on PNES.Any disagreements between reviewers were reconciled among GK, DK, and ZU, and coauthor (RA) was consulted when required.Detailed information on the excluded studies is in the appendix (S4 Appendix).Our meta-analysis utilized data from observational studies.The data that supports the findings of this study are available within the paper and its Supporting information files.

Data analysis
A standardized data extraction form was developed.For the full-text publications that were included, two reviewers (GK and DK) separately retrieved and cross-checked data, including research details, participant characteristics, and data required to compute the pooled estimates of PNES following intracranial/epilepsy surgery.Furthermore, we extracted data on PNES occurrence and stratified them according to the reported surgical area (if such data were available).Acknowledging that the common type of surgery for epilepsy is considered to be temporal resection, we stratified them into mutually exclusive groups (i.e., temporal or either area) and computed odds ratios using fixed-effect models.The primary outcome was the occurrence of PNES.Specifically, PNES following intracranial surgery, manifested by seizure-like episodes stemming from psychological distress rather than abnormal brain electrical activity, was confirmed through clinical evaluation and standardized diagnostic criteria (video-EEG).We employed «Zotero» throughout the review process to manage the records.Although we did not specifically construct an additional technique, the review process allowed us to discover duplicate data in the studies and remove them at this stage if necessary.
We used a scale based on the Strengthening the Reporting of Observational Studies in Epidemiology statement to evaluate the quality of the included articles [9].The quality of each article was rated by two independent investigators (GK and DK) on the basis of five modules: sample population, sample size, participation rate, outcome assessment, and analytical methods.A score of 0-2 was assigned to each module, indicating low, moderate, or high quality (S2 Appendix).All studies were independently rated by DK and GK, and checked by RA to resolve any disagreements.
We performed a meta-analysis of proportions to evaluate the occurrence of newly diagnosed PNES among patients exposed to neurosurgical procedures.Furthermore, we calculated a prediction interval for a proportions (PI) to estimate more robust statistics, indicating the range within which the estimates of a new study would likely fall.Quantitative analysis was performed with random effects models.We entered the reported preoperative psychiatric assessment and Full-Scale Intelligence Quotient (FSIQ) as the mean and standard deviation (SD).We placed the values when they were reported directly.However, for studies that reported only an interquartile range, the SD was estimated using the z distribution.We calculated heterogeneity estimates for the pooled estimates using the I 2 statistic and determined its significance using Cochran's Q test p-value.The I 2 value is defined as the ratio of true heterogeneity to total observed variation, which is not affected by sample size.We report both 95 % CIs and PIs for the pooled estimates.We have provided funnel plots, and their asymmetry was examined through Egger's regression tests to report on the analysis of small study biases.However, given the substantial amount of missing data a metaregression analysis was not performed.Furthermore, we performed a series of sensitivity analyses in assessing the robustness of the findings.
All statistical analyses were performed using the "meta esize" and "metan" packages in STATA 18.This study was registered with PROS-PERO on December 12 2023, identifier CRD42023488611).

Results
A literature search resulted in 1766 studies.After eliminating duplicates, we examined 1625 distinct citations to determine their eligibility.Of these, 86 abstracts and full-text publications were examined, and eight distinct studies (10)(11)(12)(13)(14)(15)(16)(17) were found to be suitable for metaanalysis (Fig. 1), one of which was related to intracranial surgery for brain neoplasms, malformations, abscess, and subdural hematoma, whereas the others included epilepsy surgery in refractory patients (Table 1).More detailed characteristics of included studies are in the Appendix (S5).The publication year ranged from 1995 to 2017.In the meta-analyzed literature, ratings for the risk of bias assessment varied from moderate to high (range 4-9), indicating satisfactory methodological quality.The most common limitations were the absence of analytical methods and sample size calculation approaches (S3 Appendix).
The overall pooled estimate of PNES occurrence (proportions) was 3 % (95 % CI 2 %− 5 %; 95 % PIs 0 %− 11 %) (Fig. 2); (Funnel plot is in the appendix (S8)).The pooled estimates from the meta-analysis indicated significant variability for both effect sizes (I 2 86.18 %; Q < 0.001 and I 2 84.78 %; Q < 0.001 respectively).Examination of the funnel plots suggested a possible small-study bias, however, subsequent Egger's regression test reported on the significant issues (p = 0.002).Additional meta-analysis (S9, S10 Appendices) for mutually exclusive surgery areas included the two largest studies (n = 1895) with a homogeneous population and common methodology.Thus, applying the fixed-effects model resulted in an odds ratio of 2.05 (95 % CI 0.81-5.19)for PNES occurrence following temporal resections.Thereafter, sensitivity analysis using the small-sample (lowest quartile) model for overall pooled estimates yielded results similar to those of the main model (i.e., 95 % CIs substantially overlapped (S12 Appendix)).When we repeated the meta-analyses separately by publication year moderator, the analysis showed that estimates from studies published in 21st century were the  same as those published before (S11 Appendix); however, with substantially less heterogeneity, namely 44.37 % (Q = 0.16) and 88.89 % (Q < 0.001), respectively.Alternatively, the 95 % PIs were substantially wider for the latter group.Furthermore, studies published the period following 2000 have consistently reported on the usage of ictal video-EEG, offering more valid and reliable diagnoses.Thus, the overall heterogeneity can be explained by the publication year.Additionally, Glosser et al. [14] and Krahn et al. [17] reported rounded sample size, namely ~250 and ~700 respectively.Hence, exclusion of those studies resulted in substantial changes of heterogeneity estimates (I 2 31.86 %, Q = 0.21; Egger's test p = 0.22).However, point estimates and precision (95 % CIs, PIs) overlapped substantially (S13, S14 Appendices).

Discussion
Eight PNES-and intracranial surgery-related studies were identified for quantitative synthesis.The pooled estimates showed high heterogeneity.The sensitivity analysis using sample size as a moderator did not improve the model.However, considering a publication year as a factor caused the lower heterogeneity estimates for recent studies.To address potential limitations sourced from the study population, exclusion of studies with «rounded» denominator resulted in the alteration of estimates, but not the precision.Nevertheless, some heterogeneity could be explained by the follow-up period, age at seizure onset, presence of preoperative psychiatric conditions, and FSIQ.Although, to generate a proper hypothesis, we were unable to perform a meta-regression analysis with regard to preoperative psychiatric conditions and FSIQ values.Therefore, there may still be a potentially significant amount of unexplained heterogeneity for these factors.Nevertheless, the narrow 95 % PIs may highlight the usefulness of our estimates.This implies that there may be a need for postoperative monitoring, especially in a cohort of patients with epilepsy receiving temporal resections.
The most prominent and clinically significant finding is that the rate of newly emerging PNES is 3 per 100 surgeries, along with the odds ratio for de novo PNES is two times (105 %) higher following temporal resection compared to resections.Although we did not identify statistical significance in odds ratios for patients with epilepsy in the present meta-analysis (p = 0.13) we found it to be important to report.First of all, there is a wide range of plausible values in the confidence intervals (0.81-5.19).In addition, considering its trend, there is a need to focus on effect estimates rather than relying on the p-value alone.In this sense, we found it interesting to present; however, it needs to be thoroughly evaluated in further studies.
Interpreting heterogeneity, we did not have sufficient data on the presence of psychopathology, low FSIQ in individuals with epilepsy, follow-up period, or age at seizure onset.Psychiatric disorders have become more prevalent with a greater proportion in previous studies [18,19].We do hypothesize that coexisting psychopathology might increase the propensity for PNES development.However, the reported prevalence of co-occurring conditions were not available in most of studies.The underlying issue for this trend is most likely related to sampling strategies, with more registry-based research publishing approaches in the past, which could lead to lower values of comorbid conditions.
It has been suggested that patients with PNES may experience delays in the diagnosis of more than seven years following the manifestations [20], with a significant contribution of younger age, interictal epileptiform potentials, and anticonvulsant medications.Acknowledging the potential causal pathway between brain resections and PNES [21], collection of a detailed post-surgical clinical history and video-EEG monitoring might be essential to establish an appropriate PNES detection strategy in patients with a history of intracranial surgery.In this sense, the estimated occurrence of PNES in people who have undergone resective surgery, especially because of epilepsy, underscores the importance of psychiatric assessments.Given the estimated proportions and annual volume of neurosurgical operations, we expect intracranial surgeries alone will contribute to newly diagnosed PNES cases.Clinicians should maintain a suspicion for PNES in patients presenting with new-onset paroxysmal events following neurosurgery.The prioritization of comprehensive patient history-taking and evaluation of the symptoms are essential.Given the limited availability, costs and relatively low incidence of PNES, video-EEG should be selectively utilized to capture and confirm suspected events rather than routine monitoring for all post-surgical patients.
To the best of our knowledge, this is the first meta-analysis of the PNES incidence following neurosurgical care.The inclusion criteria for both PNES and intracranial neurosurgery, especially epilepsy surgery, were carefully defined, yielding valid estimates that are reflective of the current literature.
Our study had several limitations.Meta-regression analyses were not performed for some essential variables; hence, our results regarding the source of heterogeneity are exploratory and require confirmation.All the studies were retrospective in nature and could have lost participants with PNES that developed much later than the follow-up period ended due to sampling strategies or non-response bias.
A key limitation of the PNES research literature is the huge heterogeneity, spurious perception and challenges in diagnosis of the condition.There is a need for more detailed primary empirical research designs to clarify the observed heterogeneity (e.g., measuring cooccurrence of psychiatric conditions in stratified surgery subgroups by intellectual abilities, clinical profile and neurological background).Given the complexity and variation of how co-occurring psychiatric conditions have been reported in the literature, we were unable to analyze comorbidity loads (i.e., number of co-occurring conditions or severity).Improved consistency in reporting illness loads and patterns of co-occurrence in future primary studies will provide further insights.However, the diagnostic approaches used across the studies were consistent.All studies employed pre-and post-surgical video EEG monitoring for the diagnosis of coexisting or de novo PNES.However, the use of ictal video EEG has been reported mostly in studies published after 2000.
Finally, all findings were derived from observational studies.The causal and developmental relationships underlying such associations require clarification by primary studies that measure the temporal relationships between neurosurgery, especially epilepsy surgery, and the occurrence of a new PNES.Co-occurring psychiatric and intellectual disabilities are common in epilepsy cohorts.Such interplay warrants clinical attention and etiological investigations to uncover shared underpinnings and sources of vulnerability associated with epilepsy surgery.

Declaration of competing interest
None.

Table 1
Characteristics of included studies.