Article Text
PDF
Abstract
Introduction The publication of explant rates has established risk factors and a definitive objective outcome of failure for spinal cord stimulation (SCS) treating neuropathic pain. We present a UK study analyzing explants of electrical neuromodulation devices for different conditions over 11 years in a single center specializing in neuromodulation.
Methods A retrospective analysis was performed using a departmental database between 2008 and 2019. Explants were analyzed according to condition, mode of stimulation and other demographics using logistic regression and Kaplan-Meier graphs with log-rank (Mantel-Cox) test.
Results Out of a total of 1177 patients, the explant rate was 17.8% at 5 years and 25.2% at 10 years. Loss of efficacy was the most frequent reason for explant 119/181 (65%). Multivariant regression analysis indicated patients with back pain without prior surgery had a reduced risk of explant (p=0.03). Patients with SCS systems that had 10 kHz, options of multiple waveforms, and rechargeable batteries also had a decreased risk of explant (p<0.001). None of these findings were confirmed when comparing Kaplan-Meier graphs, however. Contrary to other studies, we found gender and age were not independent variables for explant.
Conclusion These data contribute to a growing list of explant data in the scientific literature and give indications of what factors contribute to long-term utilization of electrical neuromodulation devices.
- spinal cord stimulation
- chronic pain
- back pain
- postoperative complications
Statistics from Altmetric.com
Introduction
Spinal cord stimulation (SCS) has benefited many patients with chronic neuropathic pain.1 2 There is evidence SCS therapy improves multidimensional pain scores and quality of life.3–9 Also noted are a reduction in opioids and an increase in return to work.6 7 9 There is, however, only low-quality evidence of SCS efficacy beyond 2 years.7 The success of SCS has been evaluated subjectively by pain scores, but there are currently no reliable objective outcomes of sustained efficacy. However, explant rates due to a loss of perceived benefit are a definitive objective outcome of failure.
As some SCS systems demand ongoing patient time commitment with recharging, an ineffective device will likely lead to cessation of use and explantation. Additionally, patients reporting discomfort with implantable pulse generators (IPGs) often demand the removal of the system when not achieving a meaningful reduction in pain.10 11
Explant rates have the potential to provide an important indication of therapeutic durability for electrical neuromodulation implantation in general. Unfortunately, there is a paucity of reported explant data and a degree of heterogeneity in the presentation of these data. Two recent studies have been instrumental in establishing a benchmark for presenting retrospective explant data in providing rates of explant per year with survival graphs.10 12 Survival graphs give a more accurate representation of the durability of electrical neuromodulation devices.
Selecting patients for SCS can be challenging. Diagnosing neuropathic pain and potential beneficiaries of SCS under the umbrella of failed back surgery syndrome (FBSS) and complex regional pain syndrome relies on subjective variables and tests with low specificity.13 14 The option of a trial of therapy before full implantation appears logical but adds complexity, and there is little evidence it improves outcomes; it merely predicts initial failure.15 It is well established that patients lose efficacy over time.10 11 16–18 The reasons for this remain unclear, but a significant placebo response in clinical trials may be a factor.19 There is an inherent need to establish enhanced measures of selecting patients before any therapy to minimize healthcare resources and potential complications to patients, including explants.18
Pain physicians still lack consensus regarding the pathogenesis of chronic neuropathic pain and the underlying mechanisms of neuromodulation therapy addressing this pain.20 21 Without a better understanding of these two variables, the failure of this therapy in a proportion of patients is inevitable. Retrospective analyzes of explant rates in selective conditions having a significant component of neuropathic pain, distributed across many demographic variables, have the potential to inform the appropriate patient selection and device developments needed to improve therapeutic effectiveness.
Methods
Study design
This study was a retrospective single-center review to determine the long-term explant rates of patients with dorsal column, dorsal root ganglion, sacral and occipital nerve stimulators (ONS) in neuropathic pain.
Data collection
This study retrospectively reviewed medical records using information from a password protected database of all implants between April 2008 and December 2018. All explants and survival time of the device were recorded up to April 2019 (figure 1). An implanter-blinded review by two personnel of accurate diagnosis, the reason for explant, and mode of stimulation for the implant/explant databases eliminated bias in our data collection.
Patient consort diagram of implants and explants over an 11-year period.
Definitions
Explant was defined as the removal of the leads and IPG without reinsertion of any piece of equipment to maintain a functioning SCS system. Thus, any change of IPG's due to battery depletion or leads due to migration was not considered to be an explant. Any patient converted to a different waveform that required exchanging the IPG or the insertion of additional leads was also not defined as an explantation. Also excluded from the analysis were any patients previously implanted at another center who required an explantation. Patients initially implanted in our center and explanted at another institution were included in the analysis to focus our review solely on ‘in-house’ implantation performance.
Explants were further categorized into the following headings: infection (including suspected infection) related to the device or wound dehiscence, loss of efficacy, a postimplantation clinical requirement for MRI, remission of pain, device-related complication and other. Loss of efficacy was defined as a patient request for the removal of the device due to inadequate pain relief despite multiple reprogramming sessions. In patients who required removal of the device for MRI interrogation to diagnose unresolved pain, the patient was classified as ‘loss of efficacy.’ Remission of pain was defined as absence or resolution of the pain prompting SCS implantation with the device switched off for >3 months. Device-related complications included: battery site pain, cutaneous erosion of device components and stimulator related headaches.
A pain psychologist assesses every patient before a trial of SCS therapy in our center. Following the psychological assessment, if deemed suitable for SCS, patients are referred either to a 2-week preimplant pain management program (PMP) or a 1-day informational ‘tech day.’ Patients were referred to the PMP if they were deemed psychologically vulnerable and prone to catastrophizing. The PMP focuses on acceptance and commitment therapy in the management of chronic pain as well as information on SCS.22 The tech day focused purely on the education of SCS.
Statistical analysis
Statistical analysis was using Stata/IC V.16.0 (StataCorp) and graphs were constructed using Prism Graph Pad V.8.0 (Graph Pad, San Diego, California, USA). Continuous variables are presented in means with SD. Binomial logistic regression was performed to ascertain the effects of age, sex, diagnosis, battery type, lead type and stimulation mode on the likelihood that participants have an explantation of an SCS device. Survival analysis was performed and presented as Kaplan-Meier (KM) curves, and a log-rank (Mantel-Cox) test was performed to compare groups of data.
Results
Our center implanted a total of 1177 patients with a neuromodulation device between April 2008 and December 2018 (figure 1). Explants and time of survival for implants were recorded up to April 2019 (figure 1). Implants were performed by five physicians with the distributions illustrated in table 1. Table 1 illustrates the patient demographics, and table 2 summarizes the identified reasons for the explants.
Patientdemographics and implant characteristics n=1177 patients
The reason for explant of electrical neuromodulation devices in n=182 patients
Overall explant rates are illustrated in figure 2. The overall explant rate was 17.8% at 5 years and 25.2% at 10 years (figure 2). There was no significant difference between implanters with regard to explant rate using KM curves (p=0.21). Loss of efficacy was the most frequent reason for explant 119/181 (65%) (table 2); the rates of explant due to loss of efficacy were 13.3% at 5 years and 17.5% at 10 years (figure 3). There was no significant difference in survival curves between the different conditions relating to a loss of efficacy (p=0.47) (figure 4). We noted a significant difference in the rate of explantation based on the anatomical placement of the lead, as shown in figure 5. Specifically, ONS devices demonstrated a significantly higher risk of early explantation (p<0.025). Conversely, there was also no statistical difference in survival curves between research or non-research patients (figure 6), patient age (figure 7A), patient gender (figure 7B), battery type (figure 7C), preimplant PMP versus tech day (figure 7D) and device and waveforms (figure 8). The majority of patients within the research cohort were using 10 kHz-SCS (46%, 74/160) (figure 6). We compared explants between the research patients using 10kHz-SCS with those patients using 10kHz- SCS who were not research patients, and there was no difference between the groups (p=0.17). The rates of explants due to infection are summarized in figure 9. Half of these occurred following revision surgery 15/30 (50%) and they are described in table 3.
Classification of explants due to infection in 30 patients
(A) Survival analysis (reversed Kaplan-Meier graph) of all explants of electrical neuromodulation with 95% CIs. (B) explant rate per year with numbers at risk.
(A) Survival analysis (reversed Kaplan-Meier estimator) of all electrical neuromodulation devices explanted due to loss of efficacy with 95% CIs. (B) Explant rates due to loss of efficacy per year with numbers at risk.
Survival analysis (reversed Kaplan-Meier graph) of all electrical neuromodulation devices explanted due to loss of efficacy separated by pathology. Log-rank (Mantel-Cox) test was performed to compare groups. There was no significant difference between the different pathologies. Abdominal/loin: n=21, back pain with/without radicular: n=141, CRPS: n=161, FBSS/FNSS: n=502, head pain: n=58, neurogenic bladder: n=25, neuropathic pain: n=62, pelvic pain: n=25. CRPS, complex regional pain syndrome; FBSS, failed back surgery syndrome; FNSS, failed neck surgery syndrome.
Survival analysis (reversed Kaplan-Meier graph) of all electrical neuromodulation devices explanted due to loss of efficacy separated by anatomical lead positions. Log-rank (Mantel-Cox) test was performed to compare groups. There were significant differences between curves with occipital nerve stimulation (ONS) being the outlier (p<0.025). DRG, dorsal root ganglion; SCS, spinal cord stimulation; SNS, sacral nerve stimulation.
(A) Survival analysis (reversed Kaplan-Meier graph) of all electrical neuromodulation devices explanted comparing research and non-research patients. (B) Survival curves of all electrical neuromodulation devices explanted due to loss of efficacy comparing research and non-research patients. Log-rank (Mantel-Cox) test was performed to compare groups. Although a visible difference, there was no statistical difference between groups ((A) p=0.068, (B) p=0.1). (C) The studies in which research patients participated in: EU Study, Virgin Back Study, Senza migraine, Modulate, Non-Surgical Refractory Back pain (NSRBP) are illustrated in different shades of blue and used 10 kHz-Spinal Cord Stimulation (SCS). NU-Burst and CRISP studies involved the use of Burst-SCS with passive recharge. The frequency study involved multiple waveforms and Dragon involved dorsal root ganglion (DRG) stimulation using tonic and Burst-DRG stimulation with active recharge. Saluda mechanism of action (MOA) study was observing physiological responses to SCS. CRISP, comparison of paresthesia mapping to anatomical placement in burst spinal cord stimulation: initial trial results of the prospective, multicenter, randomized, double‐blinded, crossover; EU, European Union; MOA, mechanism of action.
Survival analysis (reversed Kaplan-Meier graph) of all electrical neuromodulation devices explanted comparing: (A) patients aged over and under 65, (B) gender (male vs female), (C) battery type (rechargeable vs non-rechargeable). (D) Explants due to loss of efficacy comparing patients who attended a 2-week preimplant pain management program (PMP) vs a 1-day educational tech day (tech day). Log-rank (Mantel-Cox) test was performed to compare curves and there was no difference between groups.
(A) Survival analysis (reversed Kaplan-Meier graph) of all electrical neuromodulation devices explanted comparing different devices: tonic, high-frequency 10 kHz (10K) and systems with more than one waveform which included Burst (>1 waveform). Log-rank (Mantel-Cox) test was performed to compare groups and there was no difference between groups for all explants (p=0.16). (B) There was also no difference when comparing electrical neuromodulation devices explanted due to loss of efficacy (p=0.22).
(A) Survival analysis (reversed Kaplan-Meier graph) of all electrical neuromodulation devices explanted due to infection. (B) Explant rate per year and numbers at risk for infection.
Univariant and multivariant regression data are illustrated in table 4. From the regression analysis, 13 patients had incomplete data and were excluded leaving 1164 patients for analysis. From the multivariant analysis, an explant was more likely with a non-rechargeable system compared with a rechargeable (p<0.001) (table 4). This was not consistent with the survival curve analysis, however (figure 7C). Patients with back pain with or without radicular pain having had no prior surgery (ie, ‘virgin’ backs) also had a significantly lower risk of explant (p=0.03) (table 4). When compared with tonic stimulation, there was also less risk of explantation for systems using high frequency (10kHz-SCS) (p<0.001), or systems having greater than one waveform (including Burst-SCS with active and passive recharge) (p<0.001) (table 4).
Logistic regression predicting likelihood of spinal cord stimulator explantation (n=1164) based on age, sex, battery type, mode of stimulation, years since implant, research status and diagnosis
Discussion
We present the first study of cumulative explants beyond 10 years in a single center specialising in neuromodulation as far as we are aware. Our explants are illustrated in rate per year, which gives an accurate representation of accumulated time prior to explantation. Data from clinical trials seldom extend beyond 1–2 years, and we have demonstrated with other studies that explants occur beyond this period.10 12 18 The use of survival analysis (inverted KM curves) for explants, although informative, become less accurate over time as the numbers at risk decrease, thus creating much smaller sample sizes. For this reason, explant rates beyond 10 years in our study may not be genuinely reflective or accurate. Our explant rates encapsulating all etiologies and relating purely to loss of efficacy are lower compared with another European study.10 At 5 years, our explant rate was 17.5% compared with 32%.10 Explant due to loss of efficacy was 12.8% at 5 years compared with 19%.7 Contrary to our retrospective study, however, they only included patients who were known to be actively using their device. Additionally, they did not include research patients.10
Except for infection, there is little published data regarding the quantitative perioperative risk of explantation. Knowledge of this variable may be a crucial decision in proceeding with implantation in the first instance. We have included data on pain conditions not previously reported, including head pain and neurogenic bladder. From the survival analysis, we found no difference in explant rates between the different conditions relating to the loss of efficacy. Despite this, there was evidence to suggest chronic cephalgia remains a difficult condition to treat, particularly with ONS.23 24 ONS is indicated for chronic migraine and chronic cluster headache refractory to all available medical therapies, both of which are notoriously difficult to treat.23 25 26 The utilization of neuromodulation for chronic migraine and chronic cluster headache often does not involve a trial of therapy due to efficacy not being achieved in less than 6 months. This may explain higher explant rates.23
The implantation of SCS devices in patients with chronic low back pain who have not had prior surgery (‘virgin’ backs) has increased in many practices. However, there has not been a specific study directly comparing FBSS patients.4 5 7 27 Our data suggest a decreased risk of explant in patients with chronic back pain without prior surgery compared with FBSS in the regression analysis. The actual efficacy of SCS for ‘virgin’ back patients, however, is currently under investigation in a randomized placebo-controlled trial.28 The current evidence for SCS remains much stronger for patients with FBSS.29
Our regression analysis data also suggests high-frequency 10 kHz-SCS and systems with more than one waveform were less likely to be explanted compared with systems with tonic stimulation only. The survival analysis, however, illustrated no statistical difference between these newer devices. A possible reason for improved survival using newer treatment paradigms in SCS, including high-frequency and multiple waves, is the avoidance of ‘therapy habituation’.30–32 The potential to mitigate therapy habituation with pulse dosing or duty-cycling regimens with paresthesia-free waveforms appears attractive, but the mechanisms of action required are unknown and speculative at best. Habituation has yet to be scientifically defined, explained and validated concerning SCS in chronic pain patients. More targeted clinical research is required to explain this phenomenon and provide evidence that altering waveforms or neural dosing is beneficial to the longevity of neuromodulation devices. With most commercially available systems now providing multiple waveforms, the impact on explant rates will need close observation to validate a multiwaveform approach to neuromodulation therapy.
Several explant studies have highlighted risk factors for explantation relating to patient selection, including preceding higher healthcare cost, comorbidities, increased age and female gender.10 18 In our study, we found no difference in explant rates due to age or gender.
We have illustrated research patients caused the greatest divergence in explant rates compared with standard patients (figure 6), but this was not statistically significant. There are a few reasons why research patients appear to enjoy a better rate of SCS survival than the general patient population. Research SCS patients tend to undergo more rigorous selection criteria, have less psychiatric or psychological comorbidities, and no outstanding medicolegal issues or perceived secondary gain issues. An improved patient selection ability comes with clinical experience and enhances the need for Neuromodulation centers of excellence.18 33 It has been suggested low volume centers experience a higher rate of SCS explantation.18 However, an acceptable explant rate or failure of therapy has yet to be defined. By having overly stringent criteria, we may not select patients who would benefit from SCS.
When comparing patients who attended the preimplant PMP versus the educational tech day, there was no difference in explants between the two groups. There is evidence to suggest patients having psychological comorbidities or are prone to catastrophizing have less favorable outcomes with SCS.34 However, a study that combined cognitive behavioral therapy with neuromodulation devices demonstrated psychological outcomes could be enhanced in selected patient populations.35 Our results complement these findings by illustrating patients with psychological problems undergoing a preimplant PMP can achieve similar outcomes with electrical neuromodulation to standard patients (figure 7D).
Another finding from our study concerns the explant rate due to infection. Infections in the first year of an implant in other studies range from 1% to 10%.12 36–38 Our infection rate in the first year was low but subsequently developed a higher comparable rate in later years.10 12 18 39 Half of these explants involved patients who had undergone revision surgeries to reposition or change batteries or leads. These results suggest patients undergoing revisions are more prone to infection and may be due to the existence of scar tissue or prolonged surgical time.12 All infections occurred at the lead incision or IPG pocket sites, or both; there were no cases of meningitis.
Two previous studies suggested patients with SCS devices using rechargeable batteries experienced a higher rate of early explantation.10 39 We found different outcomes from our survival data and regression analysis with the latter suggesting rechargeable systems have a lower risk of explant. The decision on which type of battery to use will sometimes depend on the system manufacturer, as some do not provide non-rechargeable systems. Where both types of battery are available, the decision on which battery to implant depends more on patient factors, including the patient's ability to recharge appropriately. Studies have suggested using non-rechargeable systems may be beneficial due to lower battery consumption with lower dosing regimens.30 However, there have been reports to suggest larger, non-rechargeable cells may have a higher incidence of pocket pain38 40 and are more expensive.41 The primary reason for selecting a type of IPG should still be made on a case-by-case basis with allowance for the system to be used.30
Limitations
This study has several limitations, most importantly, its retrospective design. There are many variables that we could not control or obtain because of its retrospective nature, and some of these may be highly relevant to explant rates, including cost, number of encounters, medication usage, work status, time to implant from original surgery and comorbidities. Although we routinely follow-up our patients by telephone clinics, we may not have captured all explants in the entire cohort or confirmed all patients continued use of their device. Only two of our patients underwent explantation at a different center: one due to the recurrence of a chordoma requiring surgical excision and another following an infection of the device. A study by Pope et al demonstrated 20% of patients had their SCS explanted by a different provider from the original implanter.39 However, Pope et al conducted their study in a very different healthcare environment from the UK, where neuromodulation services are more limited and centralized.
Conclusion
Explant rates in patients selected for electrical neuromodulation give a long-term reliable, objective outcome of success or failure of the device. Our data suggest that patients with low back pain without prior surgery and patients with access to novel waveforms tend to demonstrate more favorable outcomes.
An acceptable explant rate has yet to be defined. However, the publication of further data from multiple centers could contribute to defining the acceptable limits of explant experiences in the future. Excluding ONS for head pain and non-rechargeable systems, we did not statistically determine any specific variables that increase the risk of explant for patients with neuromodulation devices. Contrary to other published studies, we found no difference in explant rates related to age or gender. More research is required, especially in a prospective manner, to determine outcomes of neuromodulation device implantation accurately. Identifying the factors most responsible for device explant are necessary to improve patient selection, techniques of device implantation, programming and device maintenance to achieve optimal satisfactory and durable therapeutic efficacy.
References
Footnotes
Contributors JR conceptualized the study and design with input from AA-K, OA-K and SH. AA-K, TS, SP, DP and NP performed the cases. OA-K, SH, JR and SW were involved in acquisition of data. JR and OA-K analyzed and interpreted the data. JR and SW performed statistical analysis. JR drafted the manuscript with input from TLY. All authors read, contributed and approved the final version of the manuscript.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests AA-K is an investigator with Nevro, Boston Scientific, Abbott and Medtronic and has stock in Micron. SP is and investigator with Saluda Medical. TLY is a consultant with Boston Scientific, an investigator with Nevro, Abbott and Boston Scientific and chief medical officer of Meghan Medical.
Patient consent for publication Not required.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement No data are available. The deidentified data are available on request from Jonathan Royds (jonathan.royds@gstt.nhs.uk)