Implantable Peripheral Nerve Stimulation for Peripheral Neuropathic Pain: A Systematic Review of Prospective Studies

Peripheral nerve stimulation (PNS) has been utilized for over 50 years with accumulating evidence of efficacy in a variety of chronic pain conditions. The level and strength of evidence supporting the use of PNS for peripheral neuropathic pain remains unclear. The purpose of this review is to synthesize data from prospective studies on the efficacy of PNS for neuropathic pain as it pertains to pain intensity, neurological deficits/neuropathy (e.g., weakness, sensory deficits, gait/balance), and other secondary outcomes (quality of life, satisfaction, emotional functioning, and adverse events). In compliance with the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines, this review identified articles from MEDLINE(R), EMBASE, Cochrane Central Register of Controlled Trials, Cochrane Database of Systematic Reviews, and Scopus. Overall, per the Grading of Recommendations Assessment, Development and Evaluation (GRADE) criteria, pooled results demonstrate very low quality or low quality of evidence supporting modest to substantial improvement in pain and neurological function after PNS implantation for treatment of peripheral neuropathic pain. PNS for phantom limb pain was the only indication that had moderate level evidence. Future prospective and well-powered studies are warranted to assess the efficacy of PNS for peripheral neuropathic pain.


Introduction
The utilization of electrical stimulation of peripheral nerves has been documented for over 50 years. It was first described in the treatment of trigeminal neuralgia, and later utilized for chronic cutaneous neuropathic limb pain [1,2]. A decade later, the first clinical studies of implantable peripheral nerve stimulation (PNS) were performed, revealing pain reduction, increased function, and quality of life with use [3,4]. Unfortunately, the early adoption of PNS was poor due to early studies demonstrating modest effectiveness with the potential for concerning complications [5,6]. With surgical advancements in nerve dissection and visualization, techniques then evolved to allow for open lead implantation

Data Extraction
The following data was extracted from each included article: general study characteristics (study design, funding, number of participants, mean age of participants) and intervention data (waveform, stimulation settings, stimulation type) and all presented outcome data with timeframes of assessment. The primary outcome of interest for this review was the change in pain intensity related to peripheral neuropathic pain after PNS implant. Secondary outcomes of interest included changes in neurological function, changes in quality of life, emotional functioning, and adverse events. Two authors (M.M. and M.Y.J.) independently extracted data, while a third author (S.C.) resolved any discrepancies.

Assessment of Risk of Bias
Risk of bias was assessed using either the Newcastle-Ottawa Quality Assessment Scale (NOS) or the Cochrane Risk of Bias Tool (C-ROB). Observational studies were assessed using the NOS while the assessment of randomized controlled trials (RCTs) was completed using the C-ROB. For the NOS, studies were evaluated based on selection (representativeness of the exposed cohort, selection of the non-exposed cohort, ascertainment of exposure, demonstration that outcome of interest was not present at the start), comparability (comparability of cohorts on the basis of the design or analysis), and exposure/outcome (assessment of outcome, follow-up long enough for outcomes to occur, adequacy of follow-up of cohorts). A maximum of four stars can be obtained for the selection domain, while a maximum of two and three stars can be obtained for the comparability and exposure/outcome domains respectively. For each domain, a greater number of stars obtained indicates a lower risk of bias. The C-ROB assesses studies for bias based on the following domains: Selection, Performance, Detection, Attrition, Reporting, and other biases. Each domain could receive a score of high risk, low risk, or unclear risk. All bias assessments were independently completed by two authors (S.C. and M.M.) with a third author adjudicating any discrepancies (R.S.D.).

Quality Assessment
The Grading of Recommendations Assessment, Development, and Evaluation (GRADE) approach was used to assess the overall quality of evidence of PNS for treatment of pain intensity for each type of neuropathic pain (primary outcome only). The GRADE assessment uses standard criteria to evaluate the certainty of evidence as being of very low, low, moderate, and high.

Results
Our search results yielded 1380 citations. After duplicate screening, 778 citations had their title and abstracts screened for eligibility. After the initial screen based on title and abstract alone, 40 full-text articles were retrieved and assessed for their eligibility. We included 14 studies and their characteristics are shown in Table 1. Figure 1 (PRISMA) shows the results of the search and reasons for exclusion. Eleven studies were prospective observational studies/case series [14][15][16][17][18][19][20][21][22][23][24] while three [25][26][27] were randomized controlled trials (RCTs).   Small non-significant decreases in depression scores (BDI-II). Improvement in quality of life with the assessment of the patient global impression of change in the second week and fourth week of follow up. Overall satisfaction score with the study was 9.6 cm, on a scale from 0 to 10 cm. All patients (100%) responded by selecting 10/10 (with 10 meaning "complete likelihood") as to their likelihood for wanting to undergo similar treatment with a permanent device.

Type of Neuropathic Pain
3.1.1. Complex Regional Pain Syndrome (CRPS) Complex Regional Pain Syndrome (CRPS) is defined as a painful condition that is disproportionate in time or degree to the usual course of any known trauma or other lesion. The pain is regional (not in a specific nerve territory or dermatome) and usually has a distal predominance of abnormal sensory, motor, sudomotor, vasomotor, and/or trophic findings 1 . Three studies [14,18,24] evaluated the use of PNS in patients diagnosed with 3.1. Type of Neuropathic Pain 3.1.1. Complex Regional Pain Syndrome (CRPS) Complex Regional Pain Syndrome (CRPS) is defined as a painful condition that is disproportionate in time or degree to the usual course of any known trauma or other lesion. The pain is regional (not in a specific nerve territory or dermatome) and usually has a distal predominance of abnormal sensory, motor, sudomotor, vasomotor, and/or trophic findings 1 . Three studies [14,18,24] evaluated the use of PNS in patients diagnosed with treatment-resistant CRPS. Frederico et al. [14] included seven patients with CRPS I and three patients with CRPS II. At 12-month follow-up after PNS implantation, visual analog scale (VAS) score, Neuropathic Pain Scale (NPS), and Short Form-12 (SF-12) physical and mental component scores were analyzed. VAS, NPS and SF-12 improved by 57.4% ± 10% (p = 0.005), 60.2% ± 12.9% (p = 0.006), and 21.9% ± 5.9% (p = 0.015), respectively. Eight of the 10 patients showed a pain reduction > 50% on the VAS scale whereas the remaining two had a >30% reduction in pain intensity. No adverse events were reported. In a prospective clinical trial [18] involving six patients with CRPS, PNS of the tibial nerve was performed. From a baseline VAS score of 7.5, follow-up data revealed reduced VAS scores after 1 month (2.6, p = 0.03), 3 months (1.6, p = 0.03), and 6 months (1.3, p = 0.02). Secondary endpoints of the average McGill score before surgery was 23.8, 11.0 (p = 0.45) after 1 month, 6.3 (p = 0.043) after 3 months, and 4.5 (p = 0.01) after 6 months. Only 1-2 h of active stimulation with 10 to 20 Hz was sufficient and provided analgesia lasting 24 h in this cohort. Lastly, 30 patients with CRPS (median nerve affected, 7 patients; ulnar nerve, 10 patients; radial nerve, one patient; common peroneal nerve, five patients; and posterior tibial nerve, seven patients) underwent PNS implantation to the affected nerve [24]. The authors reported a reduction in pain intensity from 8.3 ± 0.3 preimplantation to 3.5 ± 0.4 (56.7% ± 5.0% reduction) at the latest follow up (p < 0.001). Furthermore, there was an increased level of activity by 63.3% ± 21.8% with four patients increasing employment from unemployed to full-time employment, two from unemployed to part-time employment, and two from part-time to full-time employment.

Shoulder Pain
Implantation of PNS in patients with chronic shoulder pain was evaluated in two studies [16,20]. In a multi-site case series [16], five patients with poststroke shoulder pain received PNS to the axillary nerve. Using the Brief Pain Inventory Short Form (BPI-SF3), there was a reduction in pain intensity by 69.2% at 6 months (95% CI [1.8-5.5], p = 0.003), 84.6% at 12 months (95% CI [2.6-6.3], p = 0.0002), and 69.2% at 24 months (95% CI [1.7-5.5], p = 0.003) compared to prior device placement. All five participants experienced a 50% or greater pain reduction at 6 and 12 months after PNS, and four experienced at least a 50% reduction at 24 months after PNS. In a single-center, unblinded case series [20], PNS of the terminal branches of the axillary nerve in ten patients with chronic shoulder pain due to subacromial impingement syndrome was performed. Seven patients completed all outcome assessments with the primary outcome measure being BPI-SF3. There was a significant reduction in pain (BPI-SF3, F(1, 66) = 12.9, p < 0.01). After 16 weeks following implantation, average pain intensity among subjects was 8.2 (±standard error [SE] 1.1) and 4.2 (±SE 1.1). Apart from benign granuloma formation seen in seven patients, the authors reported no adverse events.

Phantom Limb Pain (PLP)
Phantom limb pain is defined as the perception of pain in the amputated portion of the limb after amputation. On the other hand, residual limb pain (RLP) is pain originating from the part of the limb that remains after an amputation. Three studies [22,25,26] evaluated PNS in patients with PLP. The first was a randomized, placebo-controlled, double-blind trial [26]. The primary outcome of treatment responders was defined as a ≥50% reduction in average daily pain score during weeks 1-4 of the treatment period in their RLP and PLP. The primary safety outcome was the occurrence of device-related and procedure-related adverse events assessed at all follow-up visits. Nine participants in the treatment group and six in the placebo group completed the 12-month follow-up period. At 12 months, 67% (6/9, p = 0.001) of participants receiving PNS treatment had sustained reductions of ≥50% in average pain in RLP and PLP over the week prior to the 12-month visit. No participants in the placebo group (0%, 0/14) reported ≥50% reductions in average weekly pain at the end of the placebo period. After crossing over to receive 4 weeks of active stimulation, the placebo group did report significant improvement in average PLP (33% reduction from baseline, p = 0.027) compared with placebo treatment during weeks 1-4. There were no serious or unanticipated study-related adverse events.

Post-Surgical Pain
Two studies [19,27] evaluated the effectiveness of PNS in patients with post-surgical pain. The first study [19] involved 29 patients with neuropathic, chronic postherniorrhaphy groin pain. Twenty-one patients (72.4%) presented with pain consistent with ilioinguinal nerve involvement. A total of seven patients received PNS implantation. After 3 months of follow-up, a significant reduction in pain from 8/10 to 2/10 on the NRS scale was observed (p < 0.001). Only one patient failed PNS therapy.
The second study [27] was a prospective multicenter, randomized, double-blind, partial crossover, three stage group (upper extremities, lower extremities, trunk) sequential study. The primary outcomes were pain relief and adverse events. Pain relief was measured by average pain at rest using a numerical rating scale (NRS) at three months. Safety was determined by assessment of adverse events during the one-year study period. Ninety-four patients with chronic, intractable posttraumatic/postsurgical pain were implanted and then randomized to the treatment (n = 45) or the control group (n = 49). The primary effectiveness endpoint (≥30% decrease in the NRS pain score without any upward titration of the patient's pain medicine regimen), three months after randomization to treatment, demonstrated that patients receiving active stimulation achieved a statistically significantly higher response rate of 38% vs. the 10% rate found in the control group (p = 0.0048). The overall mean pain reduction from baseline to three-month follow-up was 27.2% in the treatment group vs. 2.3% in the control group (p < 0.0001). Of the 94 subjects included in the study, 15 subjects did not participate in the 6-and 12-month follow-up, and an additional 33 did not follow up at the 12-month visit, representing an attrition rate of 51% (48/94). The authors reported no serious adverse events related with the device, but did note 14 adverse events in the treatment cohort and 13 adverse events in the control cohort. These events typically occurred and resolved early within the first three months of the study, and were largely localized to the stimulation area or site of surgery and were superficial in nature (e.g., skin rash, redness, soreness).

Mononeuropathy
Five studies [15,17,21,23] used PNS therapy for focal mononeuropathies. One case series [15] of 39 patients used PNS for focal mononeuropathies where several different nerves were targeted, with the axillary nerve (n = 18 patients) being the most frequent. The average percent reduction of VAS pain scores ranged from 29 to 100%, and the magnitude of effectiveness varied by the nerve stimulated. Notably, all three patients who received PNS of the lateral femoral cutaneous nerve experienced a 100% change in VAS from 8.3 prior to implant to 0 after implant. The effect on activity was also noted to improve by 72% in all patients. Moreover, 89% of those implanted with a PNS observed a greater than 50% reduction in opioid consumption. In a prospective case series [17], 23 patients with painful mononeuropathy secondary to leprosy underwent PNS implantation. Follow up visits were conducted at 1, 3, 6 and 12 months after PNS implant. After a seven-day trial, it was found that 10 patients reported a >50% pain reduction on the VAS scale and the neuropathic pain scale. After 12 months, 6/10 had a pain reduction of >50% or greater. Seven patients with intractable post-traumatic brachial plexus lesions [21] received a quadripolar electrode lead placed directly on the sensory peripheral branch of the main nerve involved, proximal to the site of lesion, into the axillary cavity. The mean baseline NRS was 9/10, indicating moderate to severe pain intensity before surgery. Pain intensity decreased from an NRS of 9 ± 1.15 before surgery to 2.14 ± 1.57 at the 6-month follow-up and to 2.57 ± 1.13 at the 12-month follow-up (p < 0.001). Lastly, eight patients with treatment-resistant carpal tunnel syndrome (CTS) [23] had PNS to stimulate the median nerve. The primary endpoint was pain relief near the median nerve and device safety. Overall, 2/10 patients (20%) experienced a >30% decrease in pain. Mean average pain scores were reduced from 6.7 pre-implant to 6.2 post-implant. In addition, 9/10 (90%) experienced 17-100% reduction in pain intensity on day 5 (at the end of stimulation) versus baseline, with an average pain reduction of 44.2%. After explant, pain scores returned to baseline, increasing 36.8% to 45.6% relative to the average reduced pain scores with daily stimulation. Three adverse events were reported, all of which were mild, unrelated to the device, and resolved uneventfully.

Bias Assessment
The risk of bias assessment of cohort studies is summarized in Table 2. An adequate follow-up period was determined to be at least six months, and 95% of total participants remaining under observation at the primary endpoint of the study was deemed adequate (e.g., <5% patients who dropped out). As none of the studies selected controls, a maximum of three stars could be awarded when evaluating for selection bias via the NOS due to an absence of a non-exposed cohort. No study was evaluated for comparability due to the same reason. With the exception of 4 studies [15,18,22,24], all began with a trial phase, after which only patients who met pre-determined response criteria progressed to the more permanent form of PNS. As such, all calculations regarding the duration of followup and percentage lost to follow-up were derived from the time and number of patients who entered the second phase of the respective studies. Apart from the absence of a control group, all studies demonstrated a low risk for selection bias. Five of the studies demonstrated moderate bias risk pertaining to outcomes, owing to a high percentage of patients lost to follow-up [15,17,23,24,27]. Figure 2 shows the C-ROB assessment of the three included RCTs [25][26][27]. All three studies demonstrated low risk for bias in all domains except attrition bias due to an attrition rate of 57% [25], 58% [26], and 57% [27].

Author
Year Selection Comparability Outcome

Figure 2.
Cochrane Risk-of-Bias Assessment of prospective trials. Figure 2. Cochrane Risk-of-Bias Assessment of prospective trials.

Quality of Evidence
Assessment using GRADE found that there was overall low-quality evidence supporting reduced pain intensity of peripheral neuropathic pain after treatment with PNS. While all included studies were prospective in nature, only three were RCTs, thus reducing the quality of evidence. Stratifying the GRADE quality of evidence assessment by type of peripheral neuropathic pain, low-quality evidence supported reduced pain intensity with PNS treatment for CRPS, shoulder pain, post-surgical pain, and mononeuropathies, and moderate-quality evidence for PLP. A summary table with the GRADE assessment is displayed in Table 3.

Discussion
This systematic review appraised evidence on changes in pain intensity in peripheral neuropathic pain after treatment with implantable PNS systems. Specific peripheral neuropathic pain syndromes included were CRPS, shoulder pain, PLP, post-surgical pain, and mononeuropathies of the extremities. PNS for peripheral neuropathic pain is a wellresearched area of neuromodulation, with a plethora of available literature. As such, we aimed to only assess prospective studies and exclude all retrospective data. Across 14 prospective studies and one 12-month follow-up analysis of those prospective studies, overall findings suggest that there is low-quality evidence supporting that PNS has the ability to provide clinically meaningful pain relief for peripheral neuropathic pain. The majority of patients experienced at least a 30% reduction in pain, although it was common for patients to report greater than 50% pain relief. This reduction in pain was consistent across all types of peripheral neuropathic pain syndromes. These findings align with results from recent reviews of PNS treatment in chemotherapy-induced peripheral neuropathy [28] and other peripheral neuropathies [6,[28][29][30][31].
SCS and DRG-S are also being utilized for managing patients with peripheral neuropathic pain. The literature has shown that in cases of appropriate patient selection, SCS can achieve a success rate in the range of 50-100%, approaching and even surpassing that of PNS [8,32]. However, the available evidence for DRG-S therapy for painful diabetic neuropathy (PDN) and polyneuropathy highlights low-quality GRADE evidence in pain reduction [9]. On the contrary, in the management of CRPS, a randomized, prospective trial showed clinical and statistical significance in pain relief, postural stability and mood improvements favoring DRG-S versus SCS therapy [32].
The exact mechanism by which PNS modulates peripheral neuropathic pain remains a subject of further inquiry and investigation. The hypotheses detailed in current literature can largely be divided into peripherally and centrally acting mechanisms [33]. The Gate Control Theory [34], originally proposed in 1965, hypothesized that the stimulation of large, myelinated, sensory nerve fibers exerts an inhibitory effect on the transmission of nociceptive information from smaller nerve fibers via the activation of dorsal horn interneurons. The Gate Control Theory remains the underlying foundation for the hypotheses attempting to explain the peripherally acting mechanism of PNS and has been demonstrated in both human [35] and animal [36] studies. Further research has suggested that PNS induces lower concentrations of neurotransmitters and local proinflammatory molecules in the peripheral nervous system, and that this effect plays a role in the modulation and attenuation of pain [37].
Centrally acting mechanisms also include alterations of neurotransmitter levels in the central nervous system, specifically in the serotonergic, glycinergic, and GABAergic pathways [33]. An additional centrally acting mechanism is the interference effect induced by PNS on long nociceptive fibers and pathways [38], specifically the medial lemniscal pathway, mediated by the inhibition of wide dynamic range neurons [39]. While all the mechanisms mentioned above likely contribute to the effect of PNS, the mechanisms targeted by high frequency and low intensity stimulation (inhibition of large fiber spinothalamic afferents), and low frequency and high intensity stimulation (activation of antinociceptive systems) [40] may explain how the utilization of a spectrum of frequency and intensity combinations, as seen in the studies included in this systematic review, all produce a positive pain reduction effect.
Of the four studies included in this systematic review that demonstrated the ability of PNS to reduce neuropathic pain experienced by patients who suffered some form of traumatic nerve injury, three [22,25,26] of them recruited patients who underwent amputation procedures and one recruited patients with brachial plexus and upper extremity nerve injuries [21]. The authors query the utility of neuromodulation interventions to aid in the process of nerve regeneration and reinnervation following mechanical nerve injuries, which has recently gained more attention as evidence has emerged supporting this potential mechanism [41].
Building upon decades of preclinical research that demonstrated the ability of electrical stimulation to accelerate axonal regeneration proximal to the site of nerve injury, four RCTs (3 of which were double blinded) demonstrated that a single session of low frequency electrical stimulation perioperatively was able to improve the outcomes of patients suffering from ulnar [42] and median nerve [43] (increase in motor unit number estimation), digital nerve [44] (multimodal sensory function), and spinal accessory nerve [45] (functional outcomes) injuries. The promising findings of these studies provide an opportunity for the incorporation of brief, perioperative, low frequency electrical stimulation into the standard of care for surgical peripheral nerve injury management. The exciting prospect of a multimodal approach to peripheral nerve injury management combining electrical stimulation, end-to-end nerve autografts, and local administration of FK506 (Tacrolimus), a drug that has also proven the ability to accelerate nerve regeneration, to the site of coaptation in transected nerves may further improve patient outcomes in the future [46].
Strengths of this systematic review are the inclusion of only prospective studies, query of multiple databases, and appraisal of both bias risk and evidence quality [47]. However, the findings of our study should be interpreted while taking into account some notable limitations. A common study design of included studies was prospective case series, which lack generalizability. Three [25][26][27] of the included studies had a partial crossover design, hence introducing the possibility of carryover effect and the subsequent impact on the outcome. None of the three studies implemented a reasonable washout period [48]. The subjective nature of our primary and secondary outcomes in addition to challenges with choosing a specific study population for every study may introduce ambiguity in conclusions. Another limitation is that several studies included and analyzed only patients who showed improvement in their pain [16,17,26]. We did not assess opioid and non-opioid analgesic consumption in our review. The relatively short follow-up periods of included studies also add uncertainty regarding the long-term effects and potential adverse effects from PNS therapy. Variations in implantation sites, techniques, level of proceduralists' expertise, and waveform settings are all confounding factors that may markedly impact the results. This degree of clinical and methodological heterogeneity also averted the possibility of performing a meta-analysis. There are numerous causes and inducers of peripheral neuropathy that were not accounted in this systematic review including traumatic neurovascular injuries, infections (e.g., human immunodeficiency virus infection), inflammatory causes (e.g., Guillain-Barre syndrome), hereditary disorders (e.g., Charcot-Marie-Tooth disease), and systemic disease (e.g., diabetes mellitus). Even rare causes such as pelvic malignancy and other pelvic pathologies may induce peripheral neuropathy [49,50].
PNS utilization in standard clinical practice is increasing as evidence continues to grow, supporting its mechanism of action, safety profile and clinical efficacy. Future advancements in device technology, surgical technique, waveform delivery and electrical programming will likely open the possibility of neural target optimization, allowing better understanding of responders to this therapy, thereby improving patient selection, and optimizing longitudinal efficacy and safety. For future directions, PNS indications may expand from targeting only localized neuropathic pain to more diffuse and complex painful syndromes. Although additional research is needed, this emerging therapy may have the potential to significantly change practice patterns and could substantially impact patient satisfaction and quality of life in patients suffering from intractable chronic neuropathic pain. Finally, loss of efficacy from neuromodulation interventions has been described and strategies to salvage efficacy from implanted neuromodulation devices warrant future investigation [51].

Conclusions
This review highlighted low-quality GRADE evidence supporting the use of PNS therapy to treat peripheral neuropathic pain. Further, studies highlight promising data on improvement in neurological function, quality of life, satisfaction, and emotional functioning after PNS therapy for peripheral neuropathic pain.