In this study, we explored the contribution of 18F-FDG PET in the detection of BOSD lesions, alongside MRI. In our cohort of 20 BOSD patients, we found that 18F-FDG PET exhibited focal or regional abnormality consistent with the dysplastic lesion in 95% of cases, contributing remarkably to the diagnosis of BOSD.
Initial review of MR images missed the dysplastic lesion in 20% of cases, while co-registration with PET enabled detection of the BOSD in all cases. The relatively high sensitivity of MRI to detect BOSD lesions in our cohort may relate to improved detection with 3 T MRI and better acquisition methods when compared to previous studies in which negative MRIs have been commonly reported (Salamon et al. 2008; Chassoux et al. 2010). Nonetheless, 18F-FDG PET was more sensitive than MRI in detecting hypometabolic abnormalities consistent with the dysplastic lesion, at least upon initial review, detecting an abnormality in 95% of cases. Indeed, when comparing the prognostic value of MRI and 18F-FDG PET, PET has been shown to be marginally better than MRI in the detection of FCD type IIB lesions, though not for FCD type IIA (Kim et al. 2009).
In this study, the added value of PET was particularly evident in those cases where MRI was initially negative. Following coregistration of PET/MRI, a subtle BOSD was detected in all cases where MRI was initially negative (see Fig. 4). Indeed, others similarly report the high localising value of 18F-FDG PET following PET/MRI coregistration in FCD, particularly in cases where MRI alone was negative or doubtful (Salamon et al. 2008; Chassoux et al. 2010; Desarnaud et al. 2018). The inclusion of PET in presurgical workup is indeed valuable in localising these small BOSD lesions, in line with findings from FCD cohorts.
Despite the localised nature of small BOSD lesions, the PET metabolic abnormality was not necessarily focal in nature across all patients. While the majority of patients exhibited focal hypometabolism, 35% of patients exhibited regional hypometabolism that extended beyond the focal structural lesion. All but one of the patients with focal PET hypometabolism exhibited FCD type IIB pathology, with the one remaining patient being pathology negative, likely due to limited tissue sample in this previously described case (Jackson et al. 2017). In contrast, those with regional PET hypometabolism included cases exhibiting FCD type IIA pathology, in which the abnormality was more complex and extended beyond the visible BOSD.
Localisation of these small epileptogenic lesions is highly important, as removal of the focal lesion results in an excellent post-surgical outcome. In our cohort, a successful surgical outcome was high, with 87% of cases being seizure-free in the follow-up period. This was in line with a previous study in an overlapping BOSD cohort (Harvey et al. 2015), as well as in other FCD cohorts with type II pathology (Chassoux et al. 2010). In previous FCD studies, the primary predictor of unfavourable surgical outcome is reported to be incomplete removal of the dysplastic lesion (Kim et al. 2009; Krsek et al. 2009; Rowland et al. 2012). Accurately identifying and removing the dysplastic lesion is thus of crucial importance. In our cohort, the BOSD was identified on MRI in all three patients who experienced recurrent seizures following surgery; however, it was interesting to note that of the three patients, two exhibited regional, rather than focal, hypometabolism on 18F-FDG PET. It could be that the epileptogenic zone in these two participants extended beyond the structural abnormality that was removed, and that surgical removal was incomplete. Indeed, widespread connectivity disruptions have been reported beyond the focal lesion, in both drug-resistant localization-related epilepsy and pathologically confirmed type II FCD cohorts (Besson et al. 2017; Hong et al. 2019). However, it should be noted that PET has previously shown lower spatial resolution than functional connectivity measures, and the area needed to be resected is likely smaller than the area of PET hypometabolism (Jackson et al. 2017).
There are limitations to this study that should be highlighted. This was a small cohort study, and our estimates of the sensitivity of PET in detecting BOSDs are limited by the small sample size. Moreover, given the retrospective nature of this study, our inclusion criteria limited the study cohort to those in whom a BOSD was identified and who then proceeded to surgery, and we cannot assess the potential false-negative cases who were not referred to surgery. Another limitation to our work relates to the PET analysis, which in this study was limited to visual assessments, given its integration into routine pre-surgical workup. There are various quantitative techniques, including post-processing methods and machine-learning approaches, that have recently exhibited improved detection of FCD lesions (Tan et al. 2018; Liu et al. 2019) and improved detection of BOSD (Besson et al. 2008). Future work could benefit from investigating quantitative methods, particularly those combining MRI and PET, in the detection and associated surgical outcomes in BOSD. Finally, while we focus here on the contribution of 18F-FDG PET in the clinical workup for BOSD, it may be interesting in future to assess the relationship and additional value of electrophysiological data, as has been investigated in other FCD cohorts (Lagarde et al. 2016; Chassoux et al. 2012).
In conclusion, the findings of this work suggest that 18F-FDG PET enables excellent sensitivity in the detection of BOSDs and contributes additional localising value, particularly when there is difficulty in detecting these lesions with MRI. The detection of focal PET hypometabolism in particular is associated with excellent post-surgical outcome, and our findings support the routine use of PET in presurgical workup in patients with a suspected BOSD.