The mind and body are inevitably intertwined. Current research increasingly targets interoception (i.e., the sense of the internal bodily state; Craig, 2002) as an indispensable source of sensory input that significantly influences mental processes (Azzalini et al., 2019). The heartbeat-evoked potential [HEP, (Schandry et al., 1986)] is regarded as an electrophysiological index of the cortical processing of cardiac interoceptive information (Park and Blanke, 2019). Higher HEP amplitudes have been found when participants focus on interoceptive states (Petzschner et al., 2019). At the same time, altered HEPs have been observed during multiple mental processes (Schulz et al., 2013; Luft and Bhattacharya, 2015; Sel et al., 2017; Park et al., 2018; Richter and Ibáñez, 2021) as well as in clinical conditions (Terhaar et al., 2012; Schulz et al., 2015; Wei et al., 2016; Lutz et al., 2019; Pang et al., 2019). Importantly, although the exact nature of the interplay between HEPs and cardiac dynamics is still unclear (Coll et al., 2021), changes in HEPs seem related to cardiac interoception rather than variation in cardiac signaling (Park and Blanke, 2019).

While HEP is the most used neural indicator of interoception and has often been investigated in isolation, the brain receives ascending signals from diverse inner organs (Critchley and Harrison, 2013). Breathing, for example, generates an abundance of ascending signals (Heck et al., 2017), associated with modulatory effects on other interoceptive functions, including cardiac processing (Critchley and Harrison, 2013). Indeed, the respiratory and cardiovascular systems are firmly coupled (Berntson et al., 1993; Ben-Tal et al., 2012; Dick et al., 2014a). Surprisingly, though, knowledge regarding the interplay between cardiac and respiratory interoceptive processing is scarce. In a behavioral study (Garfinkel et al., 2016), for instance, individual interoceptive awareness, referring to metacognitive insight into one’s ability to accurately detect internal bodily signals (Garfinkel et al., 2015), was significantly correlated across the heart and breathing dimensions, hinting at partially overlapping mental representations. This is supported by brain imaging studies (Hassanpour et al., 2016, 2018), highlighting a common cortical substrate, the right mid-insula, in the representation of sympathetically-induced changes in cardiorespiratory states. Integration of cardiorespiratory afferents at the cortical level could encode the physiological arousal, involving cardiorespiratory coupling by default (Ben-Tal et al., 2012; Dick et al., 2014a), that necessitate an attentional shift toward interoceptive signaling to initiate adaptive behavior (Critchley and Harrison, 2013) and possibly also related to emotional awareness (Critchley and Garfinkel, 2017) and decision making (Herman and Tsakiris, 2021). Electroencephalography (EEG) studies (Immanuel et al., 2014; Baumert et al., 2015) showing altered HEP waveforms during sleep in children with sleep-disordered breathing establishing a pathophysiological link between disrupted respiratory patterns and cardiac neural processing in populations with cardiopulmonary dysfunctions. Importantly, though, these studies have included solely sleep periods free from abnormal respiratory events in their analysis. Thus, the question of whether altered breathing patterns can affect the neural processing of cardiac information arises.

As far as the authors are aware, only one study (MacKinnon et al., 2013) has investigated how changes in ventilatory pattern relate to cardiac interoception by evaluating the effects of resonant breathing (i.e., slow diaphragmatic respiration) on HEPs in healthy participants. Specifically, the authors found an increased HEP amplitude during resonant breathing compared to baseline in one out of three recording EEG channels, providing preliminary evidence on the modulatory role of respiratory signals in HEP activation. This could expand the existing understanding of cortical mechanisms underlying breathing perceptions (von Leupoldt et al., 2010a,b, 2011; Chenivesse et al., 2014), as well as aid in elucidating the relationship between respiration and higher-order mental functioning, such as bodily self-consiousness (Adler et al., 2014; Allard et al., 2017) and facial recognition (Vinckier et al., 2018). It is also noteworthy that many neurological conditions (Cragg et al., 2015; Chen et al., 2017; Racca et al., 2020) are associated with disrupted cardiorespiratory processes. Thus, a cardiac interoceptive neural marker relating to breathing might be of considerable clinical relevance in interpreting possible interoceptive irregularities in such pathophysiological conditions.

Epilepsy is a neurological disorder linked to altered patterns of both cardiac and ventilatory activity (Moseley and Britton, 2014; Myers et al., 2018; Druschky et al., 2020), which is often encountered in HEP literature due to methodological considerations (Kern et al., 2013; Babo-Rebelo et al., 2016; García-Cordero et al., 2017; Park et al., 2018). A common activation procedure in clinical settings for the diagnosis and classification of seizure disorders is voluntary hyperventilation (VH), involving distinct electrographic responses associated with the physiological mechanisms inherent to VH (Godoy et al., 2017; Rana et al., 2023). Specifically, VH is related to EEG morphologic changes with brain wave activation manifested as a buildup of delta and theta activity (Siddiqui et al., 2011; Acharya and Acharya, 2021), a phenomenon observed also in non-epileptic individuals (Salvati and Beenhakker, 2019). At the same time, VH may trigger epileptiform discharges and clinical seizures in an appreciable portion of susceptible patients (Rana et al., 2023), although this seems to be an infrequent, as well as an epilepsy-specific, occurrence (Holmes et al., 2004; Kane et al., 2014; Alghamdi et al., 2021) that also appears to alter with increasing age (Marsden et al., 2012; Ahdab and Riachi, 2014). The processes contributing to VH-induced EEG changes remain unclear (Rana et al., 2023), with several studies suggesting a multifactorial combination of numerous concomitant physiological changes following VH (Son et al., 2012; Assenza et al., 2015; Kanaan et al., 2022; Wennberg, 2022; Rana et al., 2023), also introducing the crosstalk between cortical and subcortical structures that may involve interoceptive components and their response to altered breathing patterns (Salvati and Beenhakker, 2019; Salvati et al., 2022). Thereby, exploring cardiac interoceptive indicators and their alterations in epileptic patients during VH could aid in better understanding the disorder, but also interoceptive communication in general.

Here, we investigated how VH affects HEPs in epilepsy. Appertaining to “proof of concept” research, HEPs were derived from already recorded EEG scalp data of epilepsy patients, who have previously undergone routine EEG (rEEG) assessment with VH. VH constitutes a standard activation protocol in neurodiagnostic settings (Sinha et al., 2016) and, therefore, any obtained effects could be validated in subsequent studies with larger samples by retrospectively reviewing patient records and relating findings to specific symptoms. Following a quasi-experimental methodology, we compared HEP amplitudes across pre-VH and post-VH conditions in a sample of young and elderly adults. As VH-induced changes in breathing patterns should call for heightened attention to interoceptive afferents (von Leupoldt et al., 2010a; Petzschner et al., 2019), the main hypothesis was that HEP amplitudes will be significantly increased following the VH intervention, as compared to baseline. To the knowledge of the authors, this is the first study examining the effects of VH on HEPs.

Next to the principal research question, the current work also investigated whether VH effects on HEP waveforms differ as a function of age and sex. The motivation for such analysis was based on previous studies suggesting differences in interoceptive processing as a function of age (Khalsa et al., 2009; Murphy et al., 2018) and sex (Grabauskaitė et al., 2017). With regards to age, it was expected that young subjects will exhibit overall greater HEP waveforms in comparison to elderly participants, as interoceptive capability is thought to decrease with age (Murphy et al., 2018; but see Kamp et al., 2021). It was, furthermore, anticipated that age will interact with VH condition, with HEP amplitude being higher following VH, compared to baseline, and this effect being stronger for younger participants, thus denoting an increased interoceptive regulation for young adults (Khalsa et al., 2009; Murphy et al., 2018), in comparison to older individuals. Concerning sex, on the other hand, it was speculated that females will generally show greater HEP amplitudes, in comparison to males, as female participants display overall heightened interoceptive awareness (Grabauskaitė et al., 2017) and have also previously demonstrated higher HEPs compared to men (MacKinnon et al., 2013), however, no interaction effect was expected.

The present study aimed at investigating HEP responses following a VH intervention in epilepsy patients. Whilst the main analysis over the total sample suggested no difference across pre-VH and post-VH, there was a modulation of HEP as a function of VH condition observed within the elderly subgroup, but in the opposite direction than hypothesized. Specifically, HEPs for elderly participants were higher during pre-VH as compared to post-VH in a time window of 530 to 560 ms after the R-peak, denoting decreased cardiac interoceptive processing following breathing disruptions in senior individuals. This effect was spread predominantly over frontal sites, suggesting variations in visceral, as opposed to somatosensory, afferent representation of cardiac information (Schandry and Montoya, 1996). Importantly, the late timing of detected differences and the lack of discrepancies in ECG amplitude across VH conditions for senior participants, as well as within the time window of the identified cluster, favor the interpretation of acquired variations in HEP amplitude in terms of altered cardiac interoception as opposed to fluctuations in cardiac dynamics (Park and Blanke, 2019) or mechanical cardiorespiratory coupling (Billman, 2011).

With respect to the overall null effect, potential explanations might relate to the confounding influence of various mental (e.g., shifts in attention and/or emotional state) and physiological (e.g., changes in bodily arousal) processes on the HEP (Park and Blanke, 2019; Petzschner et al., 2019; Coll et al., 2021) that could result in fluctuations of the HEP measure at the individual level, thereby contaminating true effects (Verdonk et al., 2021). Indeed, recently, HEP’s reliability over time has been called into question by a study on trait mindfulness and interoceptive sensibility, showing poor HEP reliability across experimental sessions during rest and lack of reliable associations to self-reported body awareness, possibly relating to fluctuations in cardiac arousal at rest (Verdonk et al., 2021). This raises methodological concerns with regards to the validity of the relationship between HEPs and interoception (Coll et al., 2021), especially when it comes to resting-state measurements (Verdonk et al., 2021). Nonetheless, an alternative explanation of the null finding could pertain to variability across subgroups with respect to age-related differences on interoceptive (Murphy et al., 2018; Kamp et al., 2021), as well as autonomic (Voss et al., 2015), processes that, in turn, could have affected statistical power (Kananen et al., 2020). The inclusion of a control group, as well as of increased sample sizes (Goulet and Cousineau, 2019), in future studies could possibly aid in resolving the point of issue.

In light of the foregoing considerations, although speculative, observed reductions in HEP amplitude following VH for elderly participants might reflect the interference of VH-elicited sensory signals from respiratory afferents (Lee and Yu, 2014) with the transmission of cardiovascular ascending information, thereby, resulting in decreased cardiac signaling at the cortical level (MacKinnon et al., 2013). In particular, VH might modify respiratory-related neural input through changes in carbon dioxide (CO₂) sensing (Guyenet and Bayliss, 2015), as breathing during VH progressively eliminates CO₂ from the lungs, forcing arterial CO₂ to drop below normal levels, a phenomenon known as hypocapnia (Godoy et al., 2017). In turn, altered ventilatory signals may interact with cardiac afferents during the integration of interoceptive information (Critchley and Harrison, 2013), thus, modulating HEPs (Baumert et al., 2015). Notably, respiratory interoception might be disrupted for elderly participants, since aging is associated with a progressive decline in lung physiology and function (Sharma and Goodwin, 2006) that could potentially affect CO₂ sensing (Dempsey and Smith, 2014). Indeed, elderly individuals exhibit altered breathing patterns and arterial CO₂ tension during exercise hyperpnea, maintaining CO₂ levels at resting normocapnic range by increasing minute ventilation in order to achieve a CO₂ exchange ratio comparable to that of younger subjects (Dempsey and Smith, 2014). This suggests alterations in the sensing mechanism of respiratory CO₂ exchange that could be implicated in homeostatic deficiencies and corresponding CO₂ afferent signaling (Guyenet and Bayliss, 2015). At the same time, as aging is associated with autonomic imbalances (Voss et al., 2015), compensatory mechanisms responding to pulmonary stressors might be impaired in older individuals (Porta et al., 2014), possibly occasioning breathing instability (Dempsey and Smith, 2014) after VH cessation in elderly individuals. Accordingly, altered CO₂ sensing could interact with sympathovagal dysautonomia in elderly participants, thereby prolonging respiratory signaling after VH cessation and consequently resulting in diminished cardiac interoceptive processing, possibly also signifying disturbances in overall cardiorespiratory coupling (Druschky et al., 2020).

With regards to control analyses, HRV was significantly lower after VH cessation in both subgroups, in line with HRV-related autonomic imbalances observed post-VH (Neginhal et al., 2017; Sinha et al., 2020). Notably, though, alterations in HEP waveforms seem to generally be independent of changes in HRV (MacKinnon et al., 2013; Luft and Bhattacharya, 2015; Schulz et al., 2015; Marshall et al., 2017; Lutz et al., 2019; Kamp et al., 2021), suggesting that HEPs do not represent variability in cardiac dynamics, but rather interoceptive processes that may be differentially recruited (MacKinnon et al., 2013; Luft and Bhattacharya, 2015) in response to HRV fluctuations (Park and Blanke, 2019). Consistent with this assertion, there were no variations in HEP waveforms within the younger subgroup. In contrast, RMSSD, an index of vagally-mediated alterations in HRV (McCraty and Shaffer, 2015), was significantly lower after VH for young participants, suggesting a post-VH shift toward sympathetic control in young epilepsy adults that is consistent with similar observations in young healthy participants (Sinha et al., 2020), whilst indicating disrupted autonomic cardiac control for elderly individuals (Voss et al., 2015). However, as it has been previously reported that RMSSD is not correlated with HEP amplitude during rest, emotional stimulation, or breathing manipulations (MacKinnon et al., 2013), it is unlikely that RMSSD differences alone drive the incongruencies in HEP results acquired for the young and elderly subgroups.

Assuredly, concerning prospective caveats, the potential influence of changes in cardiac dynamics, even to a partial extent, on the acquired HEP modulations as a function of VH cannot be readily excluded (Coll et al., 2021), given also the obtained variation in HRV across VH conditions within the elderly subgroup. Importantly, though, the generation of HRV is a highly complicated process involving the regulation of vagal afferents by the intrinsic nervous system of the heart, as well as their integration with autonomic efferent signals within innately produced cardiac activity (McCraty and Shaffer, 2015). Therefore, variations in HRV might indicate disruptions in overall neurocardiac function, involving both altered heart-dependent signaling (MacKinnon et al., 2013) and sympathovagal fluctuations (Neginhal et al., 2017; Sinha et al., 2020). Consistent with this speculation, behavioral studies (Knapp-Kline and Kline, 2005; Owens et al., 2018; Leganes-Fonteneau et al., 2021) linking various HRV measures to heartbeat detection tasks support the interrelatedness of HRV with the perception of interoceptive cardiac information, which is, in turn, connected to HEP responses (Park and Blanke, 2019), although the precise role of HRV in afferent cardiac communication remains obscure (Coll et al., 2021). Still, there is evidence that HRV may be implicated in regulating the strength of the connection between the heart and the brain (MacKinnon et al., 2013), perhaps by modulating interferences in ascending vagal pathways from heart-related autonomic efferent information through baroreflex functioning (Rahman et al., 2011; Leganes-Fonteneau et al., 2021) to ensure the maintenance of homeostasis (Critchley and Harrison, 2013), thus affecting cardiac afferent traffic (McCraty and Shaffer, 2015). Of late, however, it has been also proposed that interoceptive afferents might control the mediation of homeostasis by the autonomic system and hence moderate HRV parameters (Owens et al., 2018), which could be associated with the capacity to engage autonomic reflexes in response to stressors (Owens et al., 2018; Park and Blanke, 2019). Crucially, HRV assessment in this study was based on a geometric method (Vollmer, 2015) that is more robust against breathing-induced HR changes compared to spectral and time-domain metrics (Billman, 2011; Shaffer and Ginsberg, 2017), including RMSSD (Druschky et al., 2020; Sinha et al., 2020), suggesting that acquired HRV differences indeed represent changes in neurocardiac dynamics rather than fluctuations in ventilation parameters.

With respect to age-group comparisons, most results were in the predicted direction, with young participants displaying higher HEPs than older adults at early time points both preceding (310–365 ms after R-peak) and succeeding (290–360 ms after R-peak) VH performance. The detected effects were distributed over left parietal and temporal regions, possibly indicating decreased somatosensory afferent processing in elderly subjects (Malandraki et al., 2011). Upon initial inspection, these findings are inconsistent with those obtained in one recent study (Kamp et al., 2021), published after the preregistration of the current research, where older adults exhibited increased HEPs at rest, compared to young adults, in a community sample. This age-related effect was observed in a late temporal window (455–595 ms after R-peak) and over frontal areas, denoting increased visceral, as opposed to somatosensory, afferent processing (Schandry and Montoya, 1996) and, in that sense, matching the acquired HEP effect as a function of VH condition within the elderly subgroup. Although this discordance in findings may relate to the use of dissimilar populations and/or HEP analysis paradigms, it is important to note that there were confounding differences in ECG amplitude for observed HEP differences between age groups in the present study, possibly relating to confounding CFA influences, so it is unclear whether acquired HEP differences between age groups represent age dissimilarities in cardiac parameters per se, early age-related variations in the neural processing of cardiac signals, or a combination of both (Kamp et al., 2021). In any event, obtained findings match the posterior HEP waveform increases in earlier time intervals (180–320 ms after R-peak) observed for young participants at rest by Kamp et al. (2021), though early age differences were no longer significant after controlling for CFA in their study, suggesting a major CFA influence for detected HEP modulations as a function of age group in the current case. Admittedly, the only acquired age differences that appear least confounded by the CFA, as evident by the lack of variation in ECG amplitude between age groups within the cluster time-window, concern the augmentation in HEP waveforms at 280–305 ms after R-peak for older individuals succeeding VH termination, with this effect spread over right temporal sites and a single right frontal electrode. This is in agreement with similar CFA-uncorrected findings for elderly subjects in early temporal intervals (Kamp et al., 2021), but again not entirely dissociable from CFA-related noise. Still, despite the negative correlation of HR with HEP amplitude at rest observed in older, but not younger, age (Kamp et al., 2021), there were no HR differences between age groups at baseline or after VH cessation, signifying the existence of additional sources of variability between findings in the aforementioned and present study.

Concerning sex, findings were in the opposite direction than anticipated. In total, males exhibited higher HEPs at 560–585 ms after R-peak compared to females following VH, suggesting augmented cardiac interoception for male participants after breathing perturbations. This contradictory finding may relate to the investigation of a clinical group, as opposed to a healthy sample (MacKinnon et al., 2013; Grabauskaitė et al., 2017). Alternatively, it may reflect the confounding influence of age-dependent differences in female sex hormones that are associated with altered baroreceptor sensitivity (Fu and Ogoh, 2019), which could also explain the partial overlap in observed variability between sex groups. Indeed, the time interval and location of the effect conform to the central representation of baroreceptor-mediated cardiac afferents (Gray et al., 2007). In any case, however, the late timing of detected differences and lack of variation in ECG amplitude between sex groups succeeding VH, as well as within the time window of the identified cluster, suggest that the identified alterations in HEP amplitude reflect genuine neural activity rather than mere cardiac artifacts. At the same time, acquired HR differences are consistent with higher HR reports in healthy women compared to age-matched healthy men (Moodithaya and Avadhany, 2012), yet do not mirror post-VH variations in HEP amplitude, so it is unlikely that the obtained effect represents dissimilarities between sexes in cardiac fluctuations. So, overall, acquired HEP modulations seem to verily index variations in cardiac interoception, independent of whether they actually reflect age-or sex-related differences, although additional research, involving healthy control groups and monitoring of hormonal influences, is necessary in order to unravel the actual mechanism underlying the obtained effects.

Since this preliminary study was meant to be a retrospective exploration of an epileptic patient group, aiming to establish the core rationale for further investigation of interoceptive neurobiological markers in clinical populations, the inclusion of a healthy population group in future studies would be mandatory to control for extraneous factors (Hunter et al., 2014). Besides not including a control group, our study did not incorporate an explicit measure of respiratory interoception, thereby focusing on the effects of altered respiratory patterns, rather than breathing-related interoceptive signaling per se (von Leupoldt et al., 2010a), on cardiac neural processing. Such a metric would have been conducive in further exploring the relatedness of breathing to cardiac interoception. Notably, a perceptual threshold breathing and metacognition task, termed the Filter Detection Task, has been developed recently for the quantification of breathing-related interoceptive dimensions that can be easily integrated in clinical settings due to its simple setup (Harrison et al., 2021) and, thus, may constitute an important asset for future studies. Interestingly, effects of breathing on HEP amplitude appear to be modulated by respiratory phase, at least in spontaneous breathing conditions, with increased HEP waveforms observed during exhalation in healthy adults at rest, indicating an optimization of interoceptive cortical processing of heart-related signaling across the ventilatory cycle (Zaccaro et al., 2022). On the other hand, a study showing decreased HEPs during expiration in children with sleep-disordered breathing (Baumert et al., 2015) suggests this optimizing mechanism might be impaired in clinical conditions accompanied by alterations in cardiorespiratory interactions. In this regard, it would be interesting to investigated whether HEP responses are respiratory phase-dependent in altered breathing conditions in healthy, as well as in clinical populations.

Finally, further restrictions could relate to the confounding nature of VH due to its cerebro-vascular effects on brain physiology and metabolism (Gouvea Bogossian et al., 2021). Indeed, HEP amplitude during resting-state in magnetoencephalography measurements co-varies with stroke volume differences over occipital and bilateral occipito-temporal areas in the time-window of 500–700 ms post R-peak (Buot et al., 2021). This effect disappeared with an ICA correction procedure of cardiac components (Oostenveld et al., 2011), although these results may vary depending on the statistical computation as well as HEP recording method applied (Buot et al., 2021). It is, moreover, unclear, whether the relationship between ICA-uncorrected HEPs and stroke volume is indeed mediated by the cardiac artifact or if it reflects genuine neural activity dependent on stroke volume due to transient arterial baroreceptor activity masked by the ICA correction (Buot et al., 2021). At the same time, changes in respiration rhythm affect cortical hemodynamics and thus may act as a physiological confounding factor in heart-related interoceptive awareness (Candia-Rivera et al., 2022), at least when applying the heartbeat-counting task (Desmedt et al., 2018). On the other hand, as the heartbeat-counting task may be contaminated to a large extent by non-interoceptive processes (Desmedt et al., 2018), it remains unclear whether fluctuations in breathing rate primarily affect cardiac interoception and, if so, to what extent, warranting thereby further exploration, whilst carefully controlling for differences in physiological parameters between conditions. In this regard, additional study limitations comprise the lack of pulmonary measures, such as breathing frequency, as well as the absence of capnographic evaluation, since such assessments are not incorporated in the usual clinical EEG (Sinha et al., 2016). It is unclear how respiratory rate influences RMSSD (Shaffer and Ginsberg, 2017), while literature regarding ventilatory effects on the HRV metric implemented in this study is, to the knowledge of the authors, lacking. So, the inclusion of such an estimate would be meaningful for controlling potential pulmonary confounds, as well as quantitatively appraising VH execution.

Nonetheless, breathing frequency alone does not guarantee that VH is effectively performed, as hyperventilation due to increases in minute ventilation can also ensue from heightened air tidal volume while pulmonary rate remains unaffected (Ritz et al., 2009). In this regard, CO₂ metrics would have been beneficial in determining the VH level achieved (Coppola et al., 2010), as well as in linking CO₂ decreases to HEP amplitude. Still, the omission of capnographic assessment should not adversely affect acquired HEP results, as changes in VH-induced EEG activity do not seem to differ substantially between CO₂-controlled and routine VH (Zwiener et al., 1998). Furthermore, routine VH during clinical EEG examinations appears to evoke seizures or exacerbation of interictal discharge rather infrequently (Holmes et al., 2004; Kane et al., 2014; Alghamdi et al., 2021), whilst the mechanism underlying this phenomenon remains unclear (Salvati and Beenhakker, 2019). There is, in one respect, evidence that VH-elicited changes in EEG recordings involve patient-specific sensitivities to CO₂ sensing (Salvati et al., 2022; Rana et al., 2023), although the absolute values of partial arterial CO₂ seem to be less significant than the acute changes in blood CO₂ when it comes to the provocation of spike-wave activity (Son et al., 2012). Thus, it remains unknown whether concrete arterial CO₂ levels are involved in VH-induced epileptiform discharges. In any case, the association between VH-specific EEG observations and CO₂ changes is still debated (Rana et al., 2023) and may involve additional factors, including sympathovagal imbalances (Lotufo et al., 2012; Assenza et al., 2015; Bermeo-Ovalle et al., 2015) and cardiorespiratory dysfunction (Akyüz et al., 2021).

Despite the abovementioned limitations in study design, our findings suggest a modulatory effect of altered breathing on the neural processing of heart-related interoceptive signaling in epilepsy patients. HEPs could, thus, provide a valuable marker of aberrant neurocardiac interactions in epilepsy that may be of particular relevance for clinical symptoms, such as functional seizures (Koreki et al., 2020) or sudden unexpected death in epilepsy (Moseley and Britton, 2014), possibly relating to systemic dysfunction within the cardiovagal circuitry (Druschky et al., 2020) and overall sympathetic overactivity (Myers et al., 2018). The transdiagnostic role of HEPs in other neuropsychiatric conditions and how that diverges from healthy neurobiological processes remains to be explored. In any case, due to VH’s physiologically activating nature it is difficult to dissociate neurocardiac interactions from sympathetically-driven effects, so further research is necessary before drawing any definitive conclusions. In that respect, it would be helpful to investigate alternative VH protocols, such as prolonged VH interventions (Craciun et al., 2015) and post-VH measurements (Son et al., 2012; Neginhal et al., 2017), and how these relate to VH-induced HEP alterations. The exploration of cardiorespiratory dysautonomia through stimulation of non-chemosensory peripheral channels (Hassanpour et al., 2016, 2018) and its potential implications in HEP modulations by breathing could (Dick et al., 2014b) also possibly help in the disentanglement of CO₂-mediated and autonomically-prompted effects on cardiac interoception. Additionally, it would be important to examine other forms of aberrant respiration and their relationship to cardiac interoception, like dyspnea, a CO₂-related interoceptive breathing sensation linked to several mental irruptions (Allard et al., 2017; Vinckier et al., 2018). Investigation of the effects of cardiovascular input on respiratory interoception would also be beneficial, as there is accumulating evidence that baroreceptor activation can modulate breathing (Barnett et al., 2021).

Finally, epilepsy subcategories generally differ in terms of viscero-vegetative and somatosensory symptomatology (Aljafen, 2020), while they present variations in autonomic regulation (Myers et al., 2018) and VH-triggered epileptiform discharges (Guaranha et al., 2005; Abubakr et al., 2010) as well. In that sense, different epilepsy forms could also vary with regards to interoceptive irregularities. Psychogenic seizures, for instance, are associated with distinct disruptions in interoceptive processing that predict both dissociative symptoms and the occurrence of functional seizures (Koreki et al., 2020). In that respect, interoceptive neural markers, such as the HEP, could potentially constitute valuable diagnostic tools in the identification of interoceptive abnormalities that may be related to various epilepsy subtypes (Petzschner et al., 2019). Of note, insular epilepsy (Jobst et al., 2019), an underrecognized epilepsy form and great mimicker of psychogenic non-epileptic seizures and other epilepsies with temporal semiology, could be of particular interest to future studies on cardiorespiratory interoception due to the implications of the insular cortex in the representation of concurrent changes in cardiorespiratory states (Hassanpour et al., 2016, 2018) and overall interoceptive processing (Azzalini et al., 2019), as well as its potential involvement in seizure-related cardiopulmonary dysfunction and associated patho phenomenology of sudden unexpected death in epilepsy (Jobst et al., 2019).

Concluding, breathing constitutes a vital interoceptive dimension, which possesses a fundamental role in driving neuronal activity (Heck et al., 2017; Herrero et al., 2018), possibly also relating to the modulation of cortical activity by cardiac afferent signals (McCraty and Shaffer, 2015). The present study contributes to current knowledge surrounding the interplay of breathing and cardiac interoception by providing preliminary evidence of ventilatory-related HEP modulations in elderly epilepsy patients. This could have considerable implications for HEP’s clinical significance as a prospective biomarker in the assessment of aberrant interoceptive signaling (Khalsa and Lapidus, 2016), establishing the rationale for extending this line of research to large-scale studies in various neuropsychiatric populations. Available databases involving patient records of routine VH assessment may constitute a valuable asset in this regard, whilst the application of software, such as Automagic (Pedroni et al., 2019), would allow the automated processing of large EEG data. Notably, as ventilation is situated at the crossroads of exteroception and interoception (Davenport and Vovk, 2009), a potential translatability of HEP waveforms to respiratory signaling could assist in elucidating the neural underpinnings of their intricate relationship (Marshall et al., 2017; Salvato et al., 2020).

The datasets presented in this article are not readily available because according to data protection guidelines patient datasets are not publicly available. Further inquiries can be directed to the corresponding author/s.

The studies involving humans were approved by Cantonal Ethics Committee Zurich (approval number: 2019-00890). The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

NS: Data curation, Formal analysis, Investigation, Methodology, Software, Writing – original draft, Writing – review & editing. MW: Conceptualization, Funding acquisition, Methodology, Resources, Software, Supervision, Validation, Writing – review & editing. LI: Conceptualization, Methodology, Project administration, Resources, Supervision, Writing – review & editing. BL: Conceptualization, Funding acquisition, Methodology, Resources, Supervision, Writing – review & editing.