On 2015 Sep 22, DANIEL BARTH commented:

Lack of appropriate controls leads to mistaking absence seizures for post-traumatic epilepsy

Krista M. Rodgers^1, F. Edward Dudek^2, and Daniel S. Barth¹

¹ Department of Psychology and Neuroscience, University of Colorado, Boulder, CO 80309

² Department of Neurosurgery, University of Utah School of Medicine, Salt Lake City, UT 84108

We are disappointed in the rebuttal to our paper by D’Ambrosio and colleagues, which we believe does not productively address or clarify key issues. We decided that our response would be most useful to readers if we first describe here what prompted us to publish our paper; these issues are discussed briefly in a Letter to the Editor in the Journal of Neuroscience and also discussed in a detailed response hosted at http://arxiv.org/abs/1509.05802.

Barth and Rodgers were funded from a DoD grant to use the fluid percussion injury (FPI) model to investigate interventional strategies for post-traumatic epilepsy (PTE). We used a different but well-documented FPI protocol (i.e., with a Picospritzer; (Frey et al., 2009; Rodgers et al., 2012, 2014). The FPI procedure provided macroscopic and histopathological evidence for brain injury that appeared similar if not identical to traditional FPI. We maximized severity so that with aggressive animal care we only had 10% mortality; however, greater severity enhanced mortality. Thus, we attempted to induce severe FPI and replicate the results of others, including the D’Ambrosio group.

We were initially pleased to replicate nearly all of the data described by D’Ambrosio and colleagues. We recorded epileptiform electrographic events (EEEs), the hallmark of the short non-convulsive PTE “seizures” described by D’Ambrosio et al. (see Figs 2&7 from D’Ambrosio et al., 2009), with features nearly identical to many of the events shown in their publications, including spike-and-wave morphology, high frequency of occurrence, short duration, lack of frequency progression during the event, lack of post-ictal suppression, “spindle-like” amplitude fluctuation with longer events, and time-locked behavioral interruption (i.e., “freezing”) with accompanying facial automatisms. However, with randomization and blind procedures for our experimental and control groups (i.e., sham surgical and non-surgical controls), we became increasingly concerned that virtually every animal showed EEEs, even though some of the animals were obviously controls. Furthermore, in none of the animals could we find clear evidence of a focal onset to the EEEs; nearly all of the EEE onsets were synchronous on all electrodes. We realized that we were not recording FPI-induced epileptic seizures; instead, it appeared we had brief, absence-type seizures and/or other oscillatory activity.

D’Ambrosio et al. argue that the SWDs we record in young adult (< 6 mo) rats are extremely rare and cast doubts on our methods for identifying SWD with supervised pattern recognition. Their statement is not an accurate reflection of the literature on SWD. While more difficult to identify due to their short (1-2 sec) duration, numerous authors in several papers have seen SWDs in young as well as adult rats. For example, Pearce et al. (2014), in Scharfman’s group, have reported that SWDs are present in 20% of young (2-3 months old) uninjured rats, a paper cited by the D’Ambrosio et al. rebuttal as evidence for their rarity in young animals. It is difficult to imagine quantifying brief SWD in the young animals with anything other than the supervised pattern recognition (support vector machine or SVM) we deployed. D’Ambrosio et al. suggest that the use of young rats with a low background SWD would simplify the task of identifying EEE. We strongly recommend the opposite. EEEs should be induced in older rats (> 6 mo) with abundant SWDs so that the two phenomena can be quantitatively compared, and compared to age-matched controls. This has never been done. Discriminating EEEs and SWDs in the same rats is essential if EEEs are to be validated as a model of PTE.

In terms of identifying the seizures resulting from traumatic brain injury (i.e., PTE), D’Ambrosio and colleagues essentially claim that the properties of duration and waveform of the electrographic events are not clinically relevant, and consider that the critical feature is a focal onset. Clearly, the presence or lack of a focal onset is a meaningful part of the constellation of data needed to assess epileptic seizures, but the electrographic waveforms and their duration have been important criteria for decades to distinguish different types of seizures, particularly separation of absence seizures from complex partial seizures (i.e., more recently called focal dyscognitive seizures, Berg et al., 2010). The arguments by D’Ambrosio and colleagues (D’Ambrosio et al., 2009; D’Ambrosio and Miller, 2010) concerning what is clinically relevant are based on their own effort to re-define “What is a seizure?”; these are simply opinions by D’Ambrosio and colleagues not a formal consensus report from a group of unbiased experts (Dudek and Bertram, 2010). The argument of focal onset to distinguish EEEs does not reflect PTE in humans, which is not a focal variant of absence epilepsy, identical in every aspect to absence seizures except a focal versus generalized onset. For this reason we propose in our paper that EEE are not an etiologically realistic rat model of human PTE.

In our interactions with Dudek throughout our experiments, we became familiar with the publications of Kevin Kelly (Kelly, 2004), who >10 years ago expressed concern about the EEE events of D’Ambrosio and colleagues, in large part because of Kelly’s own observations in control animals. Based on publications from Kelly et al, plus recent work from the Scharfman lab (Pearce et al., 2014), we decided that it was our scientific responsibility to publish our “negative” results, even though we fully anticipated that we would be criticized - if not ridiculed - by some researchers. We firmly believe that anyone considering use of the FPI model and EEEs, as described by D’Ambrosio and collaborators, should carefully read our paper and proceed with extreme caution. Their own description of the FPI method in their rebuttal strongly suggests that the traditional FPI device (and general procedures) is capricious and unreliable, even though other authors (Kharatishvili et al., 2006a, 2006b; Kharatishvili and Pitkanen, 2010; Shultz et al., 2013) besides D’Ambrosio and colleagues have reported genuine PTE.

On 2015 Sep 04, RAIMONDO D'AMBROSIO commented:

We find that the authors' conclusions are invalidated by poor experimental design, application of inappropriate epilepsy diagnosis criteria, and disregard of antecedent literature that is incompatible with their thesis. These issues are addressed briefly in a Letter to the Editor in the Journal of Neuroscience, and thoroughly discussed in a document found at http://arxiv.org/abs/1509.01206

First, there is no evidence that the authors managed to master the FPI technique to induce any epileptogenesis. They simply failed to induce any of the convulsive and non-convulsive epileptic seizures induced by classic FPI by many different independent laboratories. The discussion in ArXiv details why the nominal matching of the parameters of the pressure pulse used by the authors cannot be used to claim equivalency of injury.

Second, the authors' report of frequent bilateral SWDs in virtually all male Sprague Dawley rats by 3 months of age is unprecedented and in conflict with all other reports we know of, which originate from many different independent laboratories. Thus, their claim that young male Sprague Dawley rats commonly exhibit SWDs is unfounded.

Third, the authors do not use valid diagnostic criteria to distinguish focal epilepsy from absence epilepsy, but dwell instead on the morphology of the ECoG discharge which is not recognized as clinically valid to differentiate these seizure types. In the clinic, diagnosis of focal epilepsy is based on the identification of a pathological brain area from which seizures consistently precipitate (the epileptic focus) and the analysis of the consistent focality, lateralization and spread of the seizures. Using the clinically recognized criteria, it is apparent that the authors were dealing with absence epilepsy in both their control and FPI rats. The analysis in ArXiv details why the authors' diagnostic criteria are wrong, why they err in asserting that it is established that short duration and spectral power at 7-9Hz is uniquely characteristic of rodent absence epilepsy, and why the SWDs described here are clearly distinct from the focal seizures observed after effective FPI.

Fourth, the study systematically fails to acknowledge published data that are inconsistent with their conclusion, including: 1) focal seizures induced by FPI are not sensitive to drugs (valproate and carisbamate) that potently control genetic SWDs, 2) FPI-induced focal seizures can be potently prevented by mild focal cooling of the perilesional neocortical epileptic focus, which proves both their acquired nature and their origin from an epileptic focus, and 3) spectral power in the theta band is common in acquired epilepsy in both rats and humans.

As a result of these many flaws, this study simply amounts to a comparison of absence epilepsy in control rats with the same absence epilepsy in FPI rats. Contrary to the authors’ assertion, this study has no discernible implication for the translational relevance of the FPI model of PTE, or for any of the many other etiologically realistic acquired epilepsy models that are currently being used in a growing number of laboratories and that induce non-convulsive seizures with a dominant ECoG spectral power in the theta band.

On 2015 Sep 04, RAIMONDO D'AMBROSIO commented:

We find that the authors' conclusions are invalidated by poor experimental design, application of inappropriate epilepsy diagnosis criteria, and disregard of antecedent literature that is incompatible with their thesis. These issues are addressed briefly in a Letter to the Editor in the Journal of Neuroscience, and thoroughly discussed in a document found at http://arxiv.org/abs/1509.01206

First, there is no evidence that the authors managed to master the FPI technique to induce any epileptogenesis. They simply failed to induce any of the convulsive and non-convulsive epileptic seizures induced by classic FPI by many different independent laboratories. The discussion in ArXiv details why the nominal matching of the parameters of the pressure pulse used by the authors cannot be used to claim equivalency of injury.

Second, the authors' report of frequent bilateral SWDs in virtually all male Sprague Dawley rats by 3 months of age is unprecedented and in conflict with all other reports we know of, which originate from many different independent laboratories. Thus, their claim that young male Sprague Dawley rats commonly exhibit SWDs is unfounded.

Third, the authors do not use valid diagnostic criteria to distinguish focal epilepsy from absence epilepsy, but dwell instead on the morphology of the ECoG discharge which is not recognized as clinically valid to differentiate these seizure types. In the clinic, diagnosis of focal epilepsy is based on the identification of a pathological brain area from which seizures consistently precipitate (the epileptic focus) and the analysis of the consistent focality, lateralization and spread of the seizures. Using the clinically recognized criteria, it is apparent that the authors were dealing with absence epilepsy in both their control and FPI rats. The analysis in ArXiv details why the authors' diagnostic criteria are wrong, why they err in asserting that it is established that short duration and spectral power at 7-9Hz is uniquely characteristic of rodent absence epilepsy, and why the SWDs described here are clearly distinct from the focal seizures observed after effective FPI.

Fourth, the study systematically fails to acknowledge published data that are inconsistent with their conclusion, including: 1) focal seizures induced by FPI are not sensitive to drugs (valproate and carisbamate) that potently control genetic SWDs, 2) FPI-induced focal seizures can be potently prevented by mild focal cooling of the perilesional neocortical epileptic focus, which proves both their acquired nature and their origin from an epileptic focus, and 3) spectral power in the theta band is common in acquired epilepsy in both rats and humans.

As a result of these many flaws, this study simply amounts to a comparison of absence epilepsy in control rats with the same absence epilepsy in FPI rats. Contrary to the authors’ assertion, this study has no discernible implication for the translational relevance of the FPI model of PTE, or for any of the many other etiologically realistic acquired epilepsy models that are currently being used in a growing number of laboratories and that induce non-convulsive seizures with a dominant ECoG spectral power in the theta band.

On 2015 Sep 22, DANIEL BARTH commented:

Lack of appropriate controls leads to mistaking absence seizures for post-traumatic epilepsy

Krista M. Rodgers^1, F. Edward Dudek^2, and Daniel S. Barth¹

¹ Department of Psychology and Neuroscience, University of Colorado, Boulder, CO 80309

² Department of Neurosurgery, University of Utah School of Medicine, Salt Lake City, UT 84108

We are disappointed in the rebuttal to our paper by D’Ambrosio and colleagues, which we believe does not productively address or clarify key issues. We decided that our response would be most useful to readers if we first describe here what prompted us to publish our paper; these issues are discussed briefly in a Letter to the Editor in the Journal of Neuroscience and also discussed in a detailed response hosted at http://arxiv.org/abs/1509.05802.

Barth and Rodgers were funded from a DoD grant to use the fluid percussion injury (FPI) model to investigate interventional strategies for post-traumatic epilepsy (PTE). We used a different but well-documented FPI protocol (i.e., with a Picospritzer; (Frey et al., 2009; Rodgers et al., 2012, 2014). The FPI procedure provided macroscopic and histopathological evidence for brain injury that appeared similar if not identical to traditional FPI. We maximized severity so that with aggressive animal care we only had 10% mortality; however, greater severity enhanced mortality. Thus, we attempted to induce severe FPI and replicate the results of others, including the D’Ambrosio group.

We were initially pleased to replicate nearly all of the data described by D’Ambrosio and colleagues. We recorded epileptiform electrographic events (EEEs), the hallmark of the short non-convulsive PTE “seizures” described by D’Ambrosio et al. (see Figs 2&7 from D’Ambrosio et al., 2009), with features nearly identical to many of the events shown in their publications, including spike-and-wave morphology, high frequency of occurrence, short duration, lack of frequency progression during the event, lack of post-ictal suppression, “spindle-like” amplitude fluctuation with longer events, and time-locked behavioral interruption (i.e., “freezing”) with accompanying facial automatisms. However, with randomization and blind procedures for our experimental and control groups (i.e., sham surgical and non-surgical controls), we became increasingly concerned that virtually every animal showed EEEs, even though some of the animals were obviously controls. Furthermore, in none of the animals could we find clear evidence of a focal onset to the EEEs; nearly all of the EEE onsets were synchronous on all electrodes. We realized that we were not recording FPI-induced epileptic seizures; instead, it appeared we had brief, absence-type seizures and/or other oscillatory activity.

D’Ambrosio et al. argue that the SWDs we record in young adult (< 6 mo) rats are extremely rare and cast doubts on our methods for identifying SWD with supervised pattern recognition. Their statement is not an accurate reflection of the literature on SWD. While more difficult to identify due to their short (1-2 sec) duration, numerous authors in several papers have seen SWDs in young as well as adult rats. For example, Pearce et al. (2014), in Scharfman’s group, have reported that SWDs are present in 20% of young (2-3 months old) uninjured rats, a paper cited by the D’Ambrosio et al. rebuttal as evidence for their rarity in young animals. It is difficult to imagine quantifying brief SWD in the young animals with anything other than the supervised pattern recognition (support vector machine or SVM) we deployed. D’Ambrosio et al. suggest that the use of young rats with a low background SWD would simplify the task of identifying EEE. We strongly recommend the opposite. EEEs should be induced in older rats (> 6 mo) with abundant SWDs so that the two phenomena can be quantitatively compared, and compared to age-matched controls. This has never been done. Discriminating EEEs and SWDs in the same rats is essential if EEEs are to be validated as a model of PTE.

In terms of identifying the seizures resulting from traumatic brain injury (i.e., PTE), D’Ambrosio and colleagues essentially claim that the properties of duration and waveform of the electrographic events are not clinically relevant, and consider that the critical feature is a focal onset. Clearly, the presence or lack of a focal onset is a meaningful part of the constellation of data needed to assess epileptic seizures, but the electrographic waveforms and their duration have been important criteria for decades to distinguish different types of seizures, particularly separation of absence seizures from complex partial seizures (i.e., more recently called focal dyscognitive seizures, Berg et al., 2010). The arguments by D’Ambrosio and colleagues (D’Ambrosio et al., 2009; D’Ambrosio and Miller, 2010) concerning what is clinically relevant are based on their own effort to re-define “What is a seizure?”; these are simply opinions by D’Ambrosio and colleagues not a formal consensus report from a group of unbiased experts (Dudek and Bertram, 2010). The argument of focal onset to distinguish EEEs does not reflect PTE in humans, which is not a focal variant of absence epilepsy, identical in every aspect to absence seizures except a focal versus generalized onset. For this reason we propose in our paper that EEE are not an etiologically realistic rat model of human PTE.

In our interactions with Dudek throughout our experiments, we became familiar with the publications of Kevin Kelly (Kelly, 2004), who >10 years ago expressed concern about the EEE events of D’Ambrosio and colleagues, in large part because of Kelly’s own observations in control animals. Based on publications from Kelly et al, plus recent work from the Scharfman lab (Pearce et al., 2014), we decided that it was our scientific responsibility to publish our “negative” results, even though we fully anticipated that we would be criticized - if not ridiculed - by some researchers. We firmly believe that anyone considering use of the FPI model and EEEs, as described by D’Ambrosio and collaborators, should carefully read our paper and proceed with extreme caution. Their own description of the FPI method in their rebuttal strongly suggests that the traditional FPI device (and general procedures) is capricious and unreliable, even though other authors (Kharatishvili et al., 2006a, 2006b; Kharatishvili and Pitkanen, 2010; Shultz et al., 2013) besides D’Ambrosio and colleagues have reported genuine PTE.