Reply: The challenges for research on deep brain stimulation and memory

Sir, We thank Drs van den Noort et al. (2015) and Fried (2015) for their interest in our study (Miller et al., 2015) reporting improvement in visual-spatial memory during theta burst deep brain stimulation of the fornix in four individuals with temporal lobe epilepsy, a translational study previously tested in rodents by our team (Sweet et al., 2014). While echoing our cautionary statements regarding interpretation, Dr Fried (2015) applauded the use of a unique clinical opportunity to provide intriguing preliminary data. Dr van den Noort et al. (2015) describe the study as being ‘highly innovative’, ‘a promising direction for various disorders’ and ‘a pioneering study’, adding that ‘it is a strength that the authors of the present study use neuropsychological tests’. Their group also raised some concerns, so we appreciate this opportunity to respond. Dr van den Noort et al. (2015) stated that ‘stimulation settings were different from previous studies, making comparisons . . . impossible.’ It would have been ideal to directly compare paradigms within the same sample, but that would have required electrode placements in different sites, which was not feasible. Using identical outcome measures would have facilitated comparisons, but measures were diverse across studies. Moreover, our design required four equivalent forms to be validated for measuring mesial temporal function and sensitivity to change. In spite of the differences, we achieved results similar to other studies; we believe the consistency among studies using different stimulation settings at different sites within this anatomical network adds robustness to the findings and supports the potential of stimulating this system for treatment of memory disorders. Dr van den Noort et al. (2015) criticize the study for having no statistical analyses and a small clinical sample. Our case series was small in this pilot because of the invasive nature of electrode implantation. While we could capitalize on this unique clinical opportunity, not all patients required placement at the precise target for our research study, limiting the number of potential participants to four, which did not satisfy the assumptions of even nonparametric statistics. To this end, we analysed our data qualitatively and provided full disclosure in tabular and graphic formats for the reader to scrutinize, taking care not to overstate our findings. Dr van den Noort et al. (2015) were concerned that we misled the reader regarding verbal memory. In the sentence they quoted, we attempted to be very transparent: ‘Combining trials within each patient, on active stimulation one improved by 100% whereas the other three patients declined . . . .’ (p. 1837). Furthermore, in the ‘Discussion’ section we concluded: ‘the effect of stimulation on other functions such as verbal memory and naming appears to be much more complex, with considerable variability among patients on stimulation. Therefore, we cannot exclude the possibility that burst stimulation may be detrimental to some types of function in certain individuals’. Finally, Dr van den Noort et al. (2015) wondered why we analysed raw scores rather than standardized scores. Standardized scores show where a person scores relative to a reference group; in Table 1 we provided such scores doi:10.1093/brain/awv295 BRAIN 2016: 139; 1–2 | e13