35 Spinal Cord Stimulation Using High Frequency and Burst Waveform Variation
Abstract
This chapter will review the burst and high frequency waveform patterns that may be used during spinal stimulation for pain, their mechanisms of action and clinical basis for use. Also to be reviewed will be closed loop systems that are based on evoked compound action potentials.
35.1 Introduction
Spinal cord stimulation (SCS) is widely used to treat chronic neuropathic pain disorders and to a lesser extent heart failure, cardiac arrhythmia, paraplegia, and peripheral vasculopathy. In 1967 C. Norman Shealy, MD published a preliminary study on the treatment of pain by dorsal column stimulation. He described stimulation frequencies of 10–50 Hz, a pulse width of 400 µs and an amplitude of 1V. Waveforms have varied little in the half century since that initial paper and consist of charge-balanced biphasic square-waves at frequencies of 40–200 Hz. Pulse-widths have ranged between 50 and 300 µs. Class I studies have confirmed that such waveforms are effective for the treatment of what has been labeled as failed back surgery syndrome (FBSS) and complex regional pain syndrome (CRPS). However, additional work performed in the past decade has shown that other patterns may be more efficacious and engage other features of the underlying neural circuitry.
First-why does SCS treat neuropathic and not nociceptive pain?
SCS is thought to activate larger axons (> 10µm) located in the periphery of the dorsal columns. The retrograde action potentials generated by stimuli in these larger fibers then reverberate into the dorsal horn circuitry. They activate dorsal horn inter-neuronal pools that inhibit the wide dynamic range neurons carrying pain signals. When nociceptive fibers are activated in the periphery, the circuitry activated by SCS is inhibited, allowing the nociceptive pain signals to continue cephalad.
When larger axons are activated by SCS waveforms, anterograde action potentials are generated. These fibers are responsible for vibration sensation and activation of them is the basis for the perception of paresthesias. While this general mechanistic theory explains the majority of the findings in SCS, it does not explain why SCS may fail to relieve pain. Large fibers represent only one percent of the dorsal column fiber population. Capturing enough of them in the target dermatomes may not be possible with typical programming paradigms if, for example, localized scar tissue is extensive or the cerebrospinal fluid space is wider. Adjacent, untargeted areas may be also be stimulated causing unintended sensations.
35.1.1 Burst Stimulation
Before 2007, Dirk De Ridder, MD in Antwerp, Belgium had only used burst stimulation directed towards the auditory cortex for the treatment of intractable tinnitus. Tonotopic lemniscal auditory pathways fire with tonic signaling. The physiologic response to tonic signalling is slow and possibly graded. Action potentials may be produced throughout the period of stimulation. Extra-lemniscal auditory pathways use signaling bursts. Burst firing is a more powerful activation of the cerebral cortex than tonic firing. Bursting techniques are now being adapted to SCS based on how medial and lateral pain pathways use tonic and burst signaling at the thalamic level. The burst SCS signal (Burst-DR™ St. Jude Medical, Inc., St. Paul, MN) consists of five rectangular pulses at 500 Hz with a PW of one msec and bursting every 25 msec, (40 Hz). The bursts have a shaped envelope and charge balanced portion (▶ Fig. 35.1). The 1 msec pulse-width is much longer more than typical neuromodulation signals. It has been clinically compared to tonic 500 Hz signals. However, if the tonic signal has a < 500µsec pulse-width and subthreshold stimulation is used, it is thought to be comparable to 10,000 Hz high frequency stimulation. Such stimulation appears to be at least equal to traditional SCS parameters and can recover benefit in patients who have failed traditional SCS. Burst stimulation with epidural SCS, does not cause vibratory paresthesias at therapeutic amplitude levels. Paresthesia-free stimuli seem to be better tolerated and the mechanism for its effectiveness and how that mechanism could be optimized is yet to be known.
35.1.2 High Frequency Stimulation
High frequency stimulation, using 10 kHz (HF10, Senza system; Nevro Corp., Redwood City, CA) has been shown to give equal or superior results for pain relief when compared to traditional, lower frequency stimulation. In the large randomized controlled trial (SENZA-RCT) twice as many patients responded to HF-10 than to traditional SCS for back and leg pain. Patients also do not feel paresthesias with high frequency stimulation.
Is the reason for the absence of paresthesias the same with Burst and HF-10?
Are the mechanisms of pain relief the same between Burst and HF-10?
These are fundamental and unanswered questions for which answers need be sought in order to optimize their use and to develop new opportunities for the use of these waveforms in SCS treatment.
35.2 Burst and Burst-DR
▶ Fig. 35.1 shows the waveform for Burst-DR™. There are several unique features to it. The biphasic square wave is charge balanced between bursts and not during the burst. This allows for a cathodic offset during the burst. This may be important because of the lengthy, one msec, pulse width present. Traditional SCS pulse width ranges from 100–300µs. The anodic and cathodic amplitudes of the pulses progressively increase within each burst. It is not known how much such elements affect pain relief. In one study Nathan Crosby, PhD, at The University of Pennsylvania, analyzed recordings from high-threshold and wide dynamic range neurons (WDR) in the rat dorsal horn. WDR neurons can undergo “wind up” a phenomenon that allows response intensity to increase with increasing stimulus frequency. Crosby varied burst stimuli parameters such as the number of burst pulses, pulse frequency, pulse-width, burst frequency, and amplitude. As pulse-width increased from 250µs to 1000µs greater reduction of WDR firing occurred. When the number of pulses per burst increased by up to seven pulses/burst, the firing rate of WDR neurons was further reduced. If pulse frequency was increased up to 500 Hz and amplitude increased up to 90% of threshold, WDR firing rate was progressively reduced. One can conclude that the pulse-width is a significant factor. However, Schu et al.. used a tonic stimulus of 500 Hz and varied the pulse-width based on what pulse-width was needed to maintain stimulation below threshold. This is important because it may be that the delayed charge-balancing of the burst and the longer pulse-width creates an effect on axons within the dorsal column that does not occur with tonic stimulation. Moreover, when patients were treated with tonic 500 Hz stimulation, the stimulation was also below threshold. This implies that activation of larger diameter fibers in the dorsal columns was inadequate, and they were unlikely to obtain adequate pain benefit. Those patients in the tonic group had less pain relief than the lower frequency tonic patients and almost equivalent relief to those in the placebo group. In further support of this analysis, a recent study by Kriek et al., compared 40 Hz, 500 Hz, 1200 Hz and Burst-DR paradigms to treat CRPS. Here, tonic stimuli were programmed to generate paresthesias in the painful areas when they activated larger axons. The study showed no difference between any of the groups in the ability to lessen pain, and there was no preference for the non-paresthesia, burst stimuli. Finally, another study in rats by Crosby et al,. suggested that there was a difference between tonic and burst stimulation using GABA circuitry to obtain pain reduction. GABA antagonists in tonic mode blocked the effects but not in burst conditions.
Clinical findings using the standard Burst-DR waveform have been more promising. Two-year evaluation period data are not available from the Sunburst study but earlier data presented at the 2016 North American Neuromodulation Society meeting, by Dr. Timothy Deer, indicated significant reductions in VAS pain scores over traditional SCS. The magnitude of that difference, though statistically significant, was only 5 mm on a 100 mm scale (43.5 vs 48.7). This is not appreciable to the average patient. Studies have shown that a pain reduction by13 points or less on the scale is not clinically discernible. The original premise of DeRidder that the medial pain pathways, synapsing more predominantly in the anterior insula, might respond better to bursts of action potentials rather than more regular patterns, is to some degree borne out by standardized low resolution brain electromagnetic tomography analysis (sLORETA) study data. sLORETA is an accurate brain localization tool. These data show that patients treated with these waveforms seem to activate centers in the brain within those pathways, but those centers are not activated using non-burst patterns of stimuli. These findings are thought to support the idea that the subjective aspects of pain may be modulated more by burst stimuli. There were only five patients and none of the patients achieved the minimum of 50% pain reduction with either tonic or burst paradigms. Moreover, although each paradigm produced pain improvement, burst had a greater effect. It could be that the sLORETA changes occurred because of the amount of pain benefit and were not related to the pathways activated by the stimuli.
There remain unanswered questions regarding the underlying mechanisms of action for Burst-DR, and while some subjective aspects of pain processing may account for effects of burst, it is likely there are influences from activation of the dorsal column axons and their effect on the dorsal horn circuitry, just as with traditional SCS. These potential effects need to be further elucidated.
35.2.1 High Frequency Stimulation
High frequency refers to SCS at frequencies above 2–3 kHz. Traditional, tonic stimulation has been used at frequencies below 1 kHz. For purposes of this discussion, even frequencies up to the limits of many implanted pulse generators (1200 Hz) will not be considered to be high frequencies. The rationale for this distinction is that there are likely fundamental differences in what occurs at the axonal membrane in dorsal column fibers at these multi-kilohertz oscillations since such rapid alterations of the field encroach on the time constants of the ionic channels themselves. Currently, the only system available for this form of stimulation is the HF-10 system from Nevro, Inc. It uses a charge-balanced 10 kHz waveform. The waveform itself is not a simple, iterative charge-balanced square-wave, however, but instead has two flat pauses, differing in the amount of time and also separating cathode and anode phases (▶ Fig. 35.2).
The HF-10 therapy yields a paresthesia-free stimulus like Burst-DR, but has several other properties. Unlike tonic stimulation paradigms, HF-10 doesn’t lead to pain relief until about 24–72 hours after the onset of stimulation. Amplitudes are in the 0.5–2mA range (most other SCS paradigms utilize between 2 and 5mA), and there is little correlation between where the electrode is located in the medial to lateral plane and how much benefit the patient obtains. Moreover, the best results seem to come from using a longitudinal bipole electrode configuration straddling the T9–10 disk space, regardless of whether the patient has mostly back pain or leg pain or where their spinal cord ends (e.g. within T12-L2). None of these characteristics are either observed or required of the other stimulus types (i.e. traditional tonic or Burst-DR).
The Senza-RCT study consisted of a prospective, randomized review of results with a two-year evaluation period in which 77% of patients had 70% relief in back pain and 73% had 65% relief of leg pain. Despite expert programming of the traditional stimulation patients, the responders with traditional stimulation achieved only 49% relief for back or leg coverage. In responders using traditional stimulation, the benefit only reached 41% and 46% pain reduction for back and leg, respectively. This suggests that traditional stimulation is less than 50% successful, yet the criteria for even placing the permanent system following a trial of stimulation is at least 50% pain relief. Thus, either the patient selection criteria were particularly stringent in this study or patients were not selected well at all. This would undermine either the credibility of the results, or the programming of the patients. Neither of these conclusions seems likely. These results were after two-years of treatment, so one might expect that some benefit from traditional SCS may have been lost. However, analyzing the first year follow-up results shows this was not the case. Results at one year were virtually identical with traditional stimulation, yielding less than 50% benefit; for back or leg, 44% and 49%, respectively. In general, pain relief decreases between 60 and 85% are to be expected with stimulation for back and leg pain,
These concerns notwithstanding, HF-10 does seem to be comparable to traditional stimulation and should be considered as a reasonable alternative or first-line offering. Many patients would prefer not to feel paresthesias. Moreover, it may turn out that some patients respond to one type of therapy, and not as well or at all to others. Even in the same patient, over time, one therapy type may work for a while and then be supplanted by a different one, possibly even avoiding the significant potential of ‘tolerance’ to therapy by cycling through the different waveforms more rapidly, by the minute, hour, day, or longer cycle timeframes.
The mechanism by which HF-10 therapy does yield pain relief, like Burst-DR, is not well understood. The stimulation has either a direct or indirect effect on neurons within the dorsal horn, somehow then affecting the WDR neurons or other dorsal horn processing of pain. The indirect hypothesis suggests that, like traditional SCS mechanisms, there is a modification of dorsal column axons, both suppression and excitation, which leads to eventual ‘indirect’ inhibition of WDR neurons. There may be a combination of these mechanisms at work as well. Several questions need to be addressed by each hypothesis for further progress to be made. These are:
For the ‘direct’ hypothesis:
If electrodes are used only at the T9–10 disk interspace, given the weak field generated by HF-10, how can the dorsal horn neurons affected by the stimulation account for pain relief in other regions not involving neurons at the T9–10 dorsal horn (e.g. foot or lower leg, whose cells are generally located at segments below the T9–10 area and in the conus medullaris)?
For the same reason, how would an electrode over the right T9–10 dorsal horn have much affect at all on the left dorsal horn neurons? Current findings with HF-10, clinically, suggest that using paresthesia mapping of the electrode shows it can be predominantly on one side of the spinal cord and yet yield bilateral pain relief.
How is activity in the dorsal horn using HF-10 known to be ‘direct’ and not related to indirect effects of dorsal column axons, for example, modulating the dorsal horn neurons? Studies suggesting effects on dorsal horn neurons have not isolated them from the rest of the circuitry or eliminated the potential input from dorsal column axons, as occurs in the indirect theory. In fact, the activity of any dorsal horn neurons with HF-10 without such isolation would support either hypothesis.
What basis is there for such miniscule fields that HF-10 creates in the dorsal horn having any effect on neuronal cell bodies, ion channels, synapses or axons?
What basis would there be with a ‘direct’ effect to explain the delays seen in yielding pain benefit of typically over 24–48 hours?
For the ‘indirect’ hypothesis:
The effects of HF-10 on dorsal column axons suggest suppression of large diameter axons and excitation of medium and smaller axons, but only if there is a relatively monophasic field reaching the dorsal column. What is the basis for such a transformation given that the HF-10 waveform is essentially charge balanced?
If large fibers are used in traditional SCS to inhibit WDR neurons, what is the basis of medium and smaller fibers having the same final common pathway?
How can the lack of recordings in the nucleus gracilis during HF-10 stimulation be explained if some axons are thought to be firing action potentials?
How can the delay in therapeutic benefit with the ‘indirect’ hypothesis be explained?
How can broad dermatomal pain benefit be explained using an ‘indirect’ mechanism?
How can the low amplitudes used in HF-10 yield the activity in the dorsal column axons required for the ‘indirect’ hypothesis?
Related posts:
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
