Not a Slam-Dunk: Assessing the Role of EEG/fMRI in the Presurgical Evaluation

Association of EEG-fMRI Responses and Outcome After Epilepsy Surgery ¹

Andreas Koupparis, Nicolás von Ellenrieder, Hui Ming Khoo, Natalja Zazubovits, Dang Khoa Nguyen, Jeffery A Hall, Roy W R Dudley, Francois Dubeau, and Jean Gotman. Neurology 2021;97(15):10. doi: 10.1212/WNL.0000000000012660

Objective

To assess the utility of EEG-fMRI for epilepsy surgery, we evaluated surgical outcome in relation to the resection of the most significant EEG-fMRI response.

Methods

Patients with post-operative neuroimaging and follow-up of at least 1 year were included. In EEG-fMRI responses, we defined as “primary” the cluster with the highest absolute t-value located in the cortex, and evaluated three levels of confidence for the results. The threshold for low confidence was t ≥ 3.1 (P < .005); the one for medium confidence corresponded to correction for multiple comparisons with a false discovery rate of .05; and a result reached high confidence when the primary cluster was much more significant than the next highest cluster. Concordance with the resection was determined by comparison to post-operative neuroimaging.

Results

We evaluated 106 epilepsy surgeries in 84 patients. An increasing association between concordance and surgical outcome with higher levels of confidence was demonstrated. If the peak response was not resected, the surgical outcome was likely to be poor: for the high confidence level, no patient had a good outcome; for the medium and low levels, only 18% and 28% had a good outcome. The positive predictive value remained low for all confidence levels, indicating that removing the maximum cluster did not ensure seizure freedom.

Conclusion

Resection of the primary EEG-fMRI cluster, especially in high confidence cases, is necessary to obtain a good outcome, but not sufficient.

Classification of Evidence

This study provided Class II evidence that failure to resect the primary EEG-fMRI cluster is associated with poorer epilepsy surgery outcomes.

For years, the Montreal Neurological Institute (MNI) group has investigated EEG/fMRI in generalized and focal epilepsies as an electrophysiological research tool in epilepsy and as a possible contributor to the presurgical evaluation of patients with treatment-resistant epilepsies. Initially, they and others developed the data collection and analysis methods and then initiated numerous studies of patients including evaluations of hemodynamic response curves in generalized and focal epilepsies and of the brain regional differences in such responses. Later, they provided the initial glimpses into the blood oxygenation level dependent (BOLD) signal responses to interictal and ictal events in patients with focal epilepsies including a description of positive and negative responses (i.e., activations and deactivations). They provided an estimate of concordance between EEG source localization and BOLD signal peak locations in EEG/fMRI studies. ² Finally, they extended their investigation to estimating the concordance between the EEG/fMRI-derived BOLD signal changes and to the sEEG evaluation of the brain region exhibiting the BOLD peak. ³ These and other studies have laid ground for the next step which is answering the important question of whether these combined neuroimaging techniques can provide useful information in guiding surgical decisions and determining outcome of surgical intervention(s). This is where the current and several other papers come into play.

The first question many will ask is “what is EEG/fMRI?" There are several reviews and editorials that address the combination of these techniques hence there is no need to dive very deeply into the topic. However, the important aspect of this multimodality technique is that it combines two imaging methods with complementary advantages. ⁴ EEG provides excellent temporal but relatively poor spatial resolution, while fMRI provides the opposite: relatively poor temporal yet excellent spatial resolution. ⁵ There is a multitude of methods and software packages that may be used to process EEG/fMRI data. Analytical approaches vary from study to study and from one group to another. The strength of the methods used in this study is that they were previously validated in many EEG/fMRI studies. Also important are meticulous data collection and handling methods and a clear description of the process that is provided in sufficient detail in the manuscript. ¹ The drawback is that this multimodality technique, beyond few very limited and fortuitous studies, rarely captures ictal events and the data analyses need to rely, for the most part, on BOLD signal changes related to interictal epileptiform discharges (IEDs). This is because it would be virtually impossible (or at least impractical) to keep collecting EEG/fMRI data until a seizure had occurred because of the scanner time needed as well as because of the motion artifact generated by seizure events that would make the data useless. Of the limited data available, at least one study showed some discordance between the BOLD signal changes induced by IEDs and those induced by seizures. ⁶ Another study reported on IEDs and ictal events being captured but it did not provide a comparison of BOLD signal changes between IEDs and ictal event(s). ⁷ Further, yet another previously mentioned study indicated a 77% concordance between EEG/fMRI primary BOLD cluster location and the ictal onset zone determination with sEEG. ³

One of the most important aspects of the current study is the sample size, which is the largest to date in a study of this kind. While the authors clearly indicate that after division into three groups the number of participants per group was smaller, the N of 84 EEG/fMRI patients and N of 106 surgeries is very respectable. ¹ This underscores the two important aspects of this line of research. One is that large numbers of participants are necessary to address the important surgical outcome questions, and the second is that it takes a very long time to collect such data by a single center (here 13 years). ¹ How does this study compare to some of the other EEG/fMRI surgical outcome studies? The previous study by the same group (some of the participants of that study are likely included in the current report) included 35 patients with lesional or non-lesional epilepsy who successfully underwent EEG/fMRI and had focal fMRI activation identified (out of 47 who participated).^1,8 In this earlier study, the degree of concordance between the BOLD peaks and surgical outcomes was similar to the present study. The other conclusion of that study was that partial concordance or partial discordance of the EEG/fMRI results and resection site were best considered inconclusive in regards to the post-surgical outcome prediction while the highly concordant and discordant results had high value for predicting favorable or unfavorable outcomes, respectively. ⁸ Another study reported on 118 patients but only 50% of them captured IEDs and in only 10/59 the EEG/fMRI results were considered to have “critical impact” on the surgical decision. ⁷ While these 10 patients had good seizure outcomes at 1 year, it is not clear how the other patients in whom EEG/fMRI results were not considered crucial for surgical decision-making related to the long-term surgical outcomes. Further, this study did not specifically evaluate the relationship between the BOLD activation region, resection of that region, and the long-term outcome. In another study, of the 76 patients who underwent presurgical evaluation that included EEG/fMRI, 21 had EEG/fMRI activations and 10 underwent surgery—6/7 patients with the maximum BOLD signal resection were seizure free vs 0/3 who did not have the area of maximum BOLD signal area resected. ⁹

So, what are the important points? The results of the present and the previous studies underscore several significant issues. For one, it takes a long time and enormous effort to collect large datasets amenable to the analyses of outcome predictions. Second, while a portion of patients will have IEDs during the EEG/fMRI, many will not or will have insufficient numbers of IEDs to visualize an area of BOLD signal peak with high significance. This suggests that EEG/fMRI may make its most important contributions to the presurgical evaluation in enriched cohorts of patients with high frequency of IED on routine or ambulatory EEG. Third, the concordance between EEG source localization and EEG/fMRI BOLD signal peaks or between EEG/fMRI and sEEG is not absolute, as is often seen in complex surgical cases with discordant presurgical tests. These discrepancies need to be investigated and resolved before reliance can be placed on EEG/fMRI for presurgical evaluation. Fourth, the differences between BOLD signal localization in response to IEDs and ictal events—need to be investigated and explained. Finally but maybe foremost, EEG/fMRI is a research technique. Thus, lack or limited access, non-standardized methods of data collection, processing, and presentation make the implementation of this technique for clinical purposes difficult and still years away. Despite the very encouraging results of the present study, until multi-center, adequately powered, prospective outcome trials with well-defined inclusion and exclusion criteria and outcome measures using standardized EEG/fMRI methods are completed, this elegant multimodality data fusion technique will remain largely a research tool with very limited clinical applications.