**7. Research**

function, see [19, 31]; see [96] for a thorough discussion of the centrally key role of the paraventricular nucleus of the hypothalamus in autonomic dysfunction). In the individual with PTSD, however, cortical appraisal is short‐circuited (red arrows) with repetitive activation of sympathetic nervous system and freeze of fight/flight. We propose that the beneficial effects of HRVB on PTSD symptoms, including attention bias, occur according to the model shown

56 A Multidimensional Approach to Post-Traumatic Stress Disorder - from Theory to Practice

**Figure 3.** Dysregulation of heart rate deceleraton by PTSD and reduction if PTSD symptoms by HRVB.

in **Figure 3**.

**Figure 2.** Model of HRV, orienting, and PTSD.

#### **7.1. Completed research on HRVB and PTSD**

In our meta‐analysis of PTSD and HRV [74], we also examined the effects of treatment of PTSD on various HRV parameters. The first finding was that very few controlled studies examining changes in HRV variables pre‐ and posttreatment on PTSD have been published [97–100]. However, all of these studies employed some form of HRV biofeedback as the treatment intervention. The study by Lande [98] was excluded because HRV data were not included in the study report. Using conservative random effect modeling for the meta‐analysis, a signifi‐ cant increase in RMSSD could be discerned and a decrease in HR was nearly significant (p 1‐ tailed = 0.08).

Our small‐scale controlled study of the co‐occurrence of reduction in HRV parameters and sustained attention in Iraq combat veterans with and without PTSD. We [97] tested the effects of HRVB using as outcomes HRV coherence and a small battery of attentional tests patterned on Mirsky's model of attention [101]. Veterans met with an HRVB professional once weekly for 4 weeks for HRVB. HRV coherence was achieved in all participants, and the increase in coherence ratio was significant post‐HRVB training. Furthermore, significant improvements were observed as increased digit span backwards and fewer commission errors on continuous performance testing, with a significant interaction of training with PTSD on word list learning that demonstrated combat veterans with PTSD were able to benefit from HRVB to a greater degree than veterans without PTSD.

Based on the findings of that small‐scale study, we recently performed a 3‐year study of HRV and HRVB in combat veterans with PTSD, funded by the US Department of Defense. Below are some of the key findings from that study which have not been previously published anywhere else. Operation Iraqi Freedom (OIF)/Operation Enduring Freedom (OEF) veterans 21–45 years old with and without PTSD were recruited from our veterans' hospital outpatient population. PTSD+ veterans receiving standard of care for PTSD were assigned to one of two treatment groups: active HRVB training and sham HRVB training. PTSD‐ veterans served as a baseline control group only and did not receive any HRVB training. The length of training was 6 weekly sessions. A follow‐up assessment was made 8 weeks post‐training to test for persistence of effects (no HRVB was administered during the 8‐week period post‐training until follow‐up). Pre‐training (baseline), post‐training, and follow‐up PTSD symptom levels were assessed by licensed clinical psychologist raters using the Clinician Administered PTSD Scale (CAPS). Raters were blind to the training assignment groups. The study used DSM‐IV‐TR criteria, not DSM‐5 criteria, because the latter were not in existence at that time. Enrollment was planned for 30 PTSD+ veterans in each of the two HRVB groups (active and sham), and 15–20 PTSD‐ veterans in the control group; final results included 29 and 32 PTSD+ combat veterans in the active and sham HRVB subgroups, respectively, and 12 PTSD‐ combat veterans in the control group.

Some of the important findings from this study are summarized here and are being prepared for submission as a research article elsewhere. HRV coherence was quantified as log10 of the peak LF power, thus the measures of HRV analyzed were SDNN, RMSSD, log10 HF, and log10 peak LF. Nonparametric statistical tests revealed that all four pre‐training HRV measures were significantly intercorrelated; overall, SDNN was most strongly correlated to the other three HRV variables, and the largest correlation coefficient with log10 peak LF was SDNN (rho = 0.765, p(1‐tailed)<0.001). Pre‐training SDNN, RMSSD, and log HF were all significantly lower in the PTSD+ compared to the PTSD‐ subgroup (Mann‐Whitney U, all *p*s < 0.020); however, SDNN discriminated best between groups with and without PTSD.

Data showing correlations between HRV variables and measures of PTSD in a sample this size have not, to the best of our knowledge, been previously published. When the four pre‐training HRV variables were tested for associations with pre‐training PTSD, we found that Log10 HF power was most closely correlated with severity of PTSD measured as total CAPS score (*p* = ‐0.370, *p*(1‐tailed) = 0.001); HF power is a traditional measure of parasympathetic activity and consistent with the research hypotheses, the correlation between parasympathetic activity (which indicates vagal tone) was negative. Thus, as vagal tone increased, total PTSD severity decreased. Closer examination revealed that the pre‐training HRV variables associated differentially with the three pre‐training CAPS clusters: intrusive thoughts (e.g., nightmares, daytime memories), avoidance/numbing (e.g., depression, avoidance behaviors), and arousal (e.g., irritability, exaggerated startle). Log10 HF power was also the only HRV variable to significantly correlate with all three clusters (p(1‐tailed) < 0.05, all correlations negative). The time‐domain HRV variables SDNN and RMSSD were both significantly negatively correlated with the arousal cluster. The intrusive thoughts cluster was negatively correlated with log10 HF power, yet was not correlated with either of the time domain variables. The pre‐training coherence indicator, log10 peak LF, did not correlate significantly with CAPS total or any of the clusters, presumably because none of the subjects had received any training at that point in time.

With respect to differences between the active and sham HRVB subgroups, whereas pre‐ training differences in the two HRVB subgroups were nonsignificant (*p* = 0.913), the post‐ training active HRVB active group had significantly higher coherence compared to the Sham group (*p* = 0.007). This is strong evidence that active HRVB training produced coherence in those veterans who received it.

Active HRVB produced increased HRV SDNN and RMSSD post‐training, and reduced PTSD, while sham HRVB produced little or no change. Results showed that the interaction of group (Sham vs HRVB+) x time period of assessment (pre‐, post‐, follow‐up) interaction effect was significant, with clinically significant improvements in PTSD severity in the active HRVB subgroup relative to the sham HRVB subgroup. The mean CAPS score of PTSD+ subgroup receiving active HRVB training improved from 79.4 to 57.3. Within the active HRVB group, the mean PTSD severity did rebound between post‐training and follow‐up 8 weeks later to 60.8, but this increase was not statistically different from the post‐training mean, and at follow‐ up, the PTSD severity mean was statistically lower and clinically improved relative to the pre‐ training mean. Within the sham HRVB group, there were no statistical or clinical improvements in the mean PTSD severity score post‐training or at follow‐up.

#### **7.2. Planned research on HRVB and PTSD: the action cascade**

peak LF power, thus the measures of HRV analyzed were SDNN, RMSSD, log10 HF, and log10 peak LF. Nonparametric statistical tests revealed that all four pre‐training HRV measures were significantly intercorrelated; overall, SDNN was most strongly correlated to the other three HRV variables, and the largest correlation coefficient with log10 peak LF was SDNN (rho = 0.765, p(1‐tailed)<0.001). Pre‐training SDNN, RMSSD, and log HF were all significantly lower in the PTSD+ compared to the PTSD‐ subgroup (Mann‐Whitney U, all *p*s < 0.020); however,

Data showing correlations between HRV variables and measures of PTSD in a sample this size have not, to the best of our knowledge, been previously published. When the four pre‐training HRV variables were tested for associations with pre‐training PTSD, we found that Log10 HF power was most closely correlated with severity of PTSD measured as total CAPS score (*p* = ‐0.370, *p*(1‐tailed) = 0.001); HF power is a traditional measure of parasympathetic activity and consistent with the research hypotheses, the correlation between parasympathetic activity (which indicates vagal tone) was negative. Thus, as vagal tone increased, total PTSD severity decreased. Closer examination revealed that the pre‐training HRV variables associated differentially with the three pre‐training CAPS clusters: intrusive thoughts (e.g., nightmares, daytime memories), avoidance/numbing (e.g., depression, avoidance behaviors), and arousal (e.g., irritability, exaggerated startle). Log10 HF power was also the only HRV variable to significantly correlate with all three clusters (p(1‐tailed) < 0.05, all correlations negative). The time‐domain HRV variables SDNN and RMSSD were both significantly negatively correlated with the arousal cluster. The intrusive thoughts cluster was negatively correlated with log10 HF power, yet was not correlated with either of the time domain variables. The pre‐training coherence indicator, log10 peak LF, did not correlate significantly with CAPS total or any of the clusters, presumably because none of the subjects had received any training at that point

With respect to differences between the active and sham HRVB subgroups, whereas pre‐ training differences in the two HRVB subgroups were nonsignificant (*p* = 0.913), the post‐ training active HRVB active group had significantly higher coherence compared to the Sham group (*p* = 0.007). This is strong evidence that active HRVB training produced coherence in

Active HRVB produced increased HRV SDNN and RMSSD post‐training, and reduced PTSD, while sham HRVB produced little or no change. Results showed that the interaction of group (Sham vs HRVB+) x time period of assessment (pre‐, post‐, follow‐up) interaction effect was significant, with clinically significant improvements in PTSD severity in the active HRVB subgroup relative to the sham HRVB subgroup. The mean CAPS score of PTSD+ subgroup receiving active HRVB training improved from 79.4 to 57.3. Within the active HRVB group, the mean PTSD severity did rebound between post‐training and follow‐up 8 weeks later to 60.8, but this increase was not statistically different from the post‐training mean, and at follow‐ up, the PTSD severity mean was statistically lower and clinically improved relative to the pre‐ training mean. Within the sham HRVB group, there were no statistical or clinical

improvements in the mean PTSD severity score post‐training or at follow‐up.

SDNN discriminated best between groups with and without PTSD.

58 A Multidimensional Approach to Post-Traumatic Stress Disorder - from Theory to Practice

in time.

those veterans who received it.

The basic results presented above provide evidence that HRVB reduces formal DSM‐IV symptoms, yet there remains a gap in our understanding of stimulus appraisal, attention, and orienting aspects of PTSD. The orienting reflex could facilitate attention and perception toward a stimulus on one hand, whereas it could bias attention away from the percept on the other hand. Our planned research on the autonomic stages of the OR in combat veterans with PTSD uses the action cascade, a software program of our own creation. The action cascade is a computerized test that presents the subject with stimulus trials that produce an experimental analog of the naturalistic stages of orienting and response: Rest, Alert, Vigilance, Orienting and Appraisal, and Response Selection and Output. Each trial lasts about 25 s (**Figure 4**). Heart rate and HRV are recorded continuously and simultaneously with task performance on the action cascade by linking the physiological recorder to the computer stimulus presentation program.

**Figure 4.** Action cascade: HRV during stages of rest, alert, vigilance, stimulus orienting/appraisal and response.

We have developed HRV Cascade Action Software to measure HRV during the stages of Rest, Alert, Vigilance, Orienting/Appraisal and Response. Durations of the Rest, Alert, and Vigilance stages vary to reduce the anticipatory predictability of the task. The action cascade is a close analog of the defense cascade paradigm, but modified stimulus valence (e.g., pleasant, unpleasant or fear‐provoking)—which would provoke emotionally laden ANS responding to be instead only informational (Go, No Go) and thereby guiding the action of stimulus appraisal into the cortical (mPFC) portion of the cortico‐limbic circuit controlling autonomic cardiac regulation. The action cascade protocol is in preliminary data collection stage at this time. The working hypothesis, illustrated in **Figure 4**, is that cardiac deceleration will be absent or at least attenuated in the PTSD+ subjects pre‐training HRVB, and this deficit will be normalized or at least improved post‐training HRVB. Results may bridge the gap in under‐ standing the role that ANS dysfunction plays in the adverse effects of PTSD on arousal, attention, and response disinhibition.

#### **8. Summary and conclusions**

Our chapter has reviewed evidence underlying the theory that ANS control of cardiac adjustments to environmental stimulation is a central factor in the symptom complex of PTSD. HRV is measured and quantified in terms of power (variance) and the coherence ratio of parasympathetic to total variance in the tachygram. Understanding of vagus nerve as the major control point of responsivity to environmental stimulation, with inputs and outputs affecting emotions, cognition, and behavior, fits into the evolutionary framework that includes the range of response outputs—fight or flight, freezing, tonic immobility, and affiliation. The neurovisc‐ eral integration model specifies the neuroanatomical networks of vagal afference and efference which control the rhythm of cardiac acceleration and deceleration. The entire system of ANS‐ regulated defense cascade is due to the executive ability of prefrontal cortex. Fear is a normal and adaptively healthy aspect of the defense cascade, well‐understood and modeled by translational models. Dysregulation of the normal fear response by traumatization deranges the ANS and its control of HRV and subsequent defense cascade. As a result, attentional bias both toward and away from reminders and fear‐provoking stimulation occurs. HRVB is theoretically and intuitively beneficial in the restoration of ANS function to adaptive para‐ sympathetic and sympathetic levels. While these complex relations can be heuristically modeled, the reader is cautioned that PTSD is a very heterogeneous and multifactorial disorder and numerous other approaches to modeling and treatment (epigenetic, neuro‐inflammatory, cognitive‐behavioral, to name a few) are certain to add to our understanding and successful treatment outcomes. Our research provides preliminary evidence that HRVB improves HRV and reduces PTSD symptoms, and we intend to further develop our model with an experi‐ mental paradigm.

#### **Author details**

Jay P. Ginsberg\* and Madan Nagpal

\*Address all correspondence to: jay.ginsberg@va.gov

Dorn VA Medical Center and Dorn Research Institute, Columbia, SC, USA
