Advances in Interventional Therapies for Painful Diabetic Neuropathy: A Systematic Review

Abstract

Background:

Painful diabetic neuropathy (PDN) is one of the major complications of diabetes mellitus. It is often debilitating and refractory to pharmaceutical therapies. Our goal was to systematically review and evaluate the strength of evidence of interventional management options for PDN and make evidence-based recommendations for clinical practice.

Methods:

We searched PubMed, Scopus, Google Scholar, and Cochrane library and systematically reviewed all types of clinical studies on interventional management modalities for PDN.

Results:

We identified and analyzed 10 relevant randomized clinical trials (RCTs), 8 systematic reviews/meta-analyses, and 5 observational studies of interventional modalities for PDN using pain as primary outcome. We assessed the risk of bias in grading of evidence and found that there is moderate to strong evidence to support the use of dorsal column spinal cord stimulation (SCS) in treating PDN in the lower extremities (Evidence level: 1B+), while studies investigating its efficacy in the upper extremities are lacking. Evidence exists that acupuncture and injection of Botulinum toxin -A provide relief in pain or muscle cramps due to PDN with minimal side effects (2B+/1B+). Similar level of evidence supports surgical decompression of lower limb peripheral nerves in patients with intractable PDN and superimposed nerve compression (2B±/1B+). Evidence for sympathetic blocks or neurolysis and dorsal root ganglion (DRG) stimulation is limited to case series (2C+).

Conclusion:

Moderate to strong evidence exists to support the use of SCS in managing lower extremity pain in patients who have failed conventional medical management for PDN. Acupuncture or injection of Botulinum toxin -A can be considered as an adjunctive therapy for PDN. Surgical decompression of peripheral nerves may be considered in patients with PDN superimposed with nerve compression. High quality studies are warranted to further evaluate the safety, efficacy, and cost-effectiveness of interventional therapies for PDN.

INTRODUCTION

Diabetes mellitus affects about 425 million people worldwide¹ with an estimated annual cost of $327 billion in US alone.² Diabetic sensorimotor polyneuropathy (DSP) is the most common complication, which affects up to 50% of patients with diabetes.^3–5 Approximately 30–50% of patients with DSP develop painful diabetic neuropathy (PDN).^{1, 6} It is estimated that 5.8 to 7.6 million people in the US suffer from PDN,⁷ which is often debilitating.^{1, 6}

Pharmacological therapies have been the mainstay for symptom management in patients with PDN in addition to optimizing the management of diabetes.^{5, 8} The most current first-line pharmacotherapies include gabapentinoids, serotonin-norepinephrine reuptake inhibitors, and tricyclic antidepressants.^{1, 5, 8, 9} However, the efficacy of these medications is limited and many patients respond poorly to these drugs.^{9, 10} Other drugs such as dextromethorphan,^11–13 lacosamide,^14–17 lidocaine infusion,^{18, 19} sustained release formulation of sodium nitrite,²⁰ and inhaled cannabinoids²¹ were reported to be effective but the evidence is far from being conclusive. Therefore, these drugs have not been recommended in practice guidelines by professional societies.⁸

With millions of people suffering from PDN and many of them not responding well to drugs, non-pharmacological interventions have been investigated in recent years. Interventional therapies such as spinal cord stimulation (SCS), acupuncture, Botulinum toxin A (BTX-A) injection, sympathetic nerve blocks, and surgical decompression of specific peripheral nerves have shown promise to improve clinical outcomes of PDN and to decrease the use of drugs and their associated adverse effects. In this systematic review, we seek to review the recent advances in interventional therapies for PDN, to evaluate the evidence (or lack of it) for the application of interventional treatments, to fill the knowledge gap in this emerging field, and to provide evidence-based recommendations in managing patients with refractory PDN.

METHODS

This systematic review is reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) of individual participant data.²²

Search strategy

Electronic databases PubMed, Scopus, Google Scholar, and Cochrane library were searched using the following terms: “painful diabetic neuropathy” AND “randomized blinded controlled trial” OR “pilot study” OR “retrospective study” OR “systematic review” OR “meta-analysis” OR “spinal cord stimulation” OR “acupuncture” OR “botulinum toxin” OR “surgical decompression”. There was no date limitation on literature search and the last search was conducted on April 1, 2021.

Study selection

Inclusion criteria were meta-analyses, prospective randomized controlled trials, prospective pilot studies, retrospective reports, case reports, and case series that evaluated interventional therapies in patients with PDN. Primary outcome of interest was numerical rating or visual analog pain scales. We excluded animal studies and trials that lack adequate study description. Additional literatures were further identified after read the full text of the relevant reports. Four of the authors (LX, JX, ZS, and EC) independently performed the search and extracted data from articles. Any disagreements were resolved by discussion (LX, JX, and JC).

Risk of bias assessment and Data synthesis

Using the Cochrane risk of bias tool (RoB 2),²³ risk of bias of each randomized trial was assessed by one author (JX) and checked by a second author (JC). The risk domains included the randomization process, deviation from the intended interventions, missing outcome data, measurement of the outcomes, selection of the reported results, and the overall risk of bias. All domains were assessed as “Low risk of bias”, “Some concerns”, or “High risk of bias”.²³ Data synthesis was based upon assessments of the quality of individual studies, outcomes assessment, and qualitative analysis. The quality of each individual article included in this analysis was assessed by applying the Cochrane review criteria²⁴ and the Interventional Pain Management techniques - Quality Appraisal of Reliability and Risk of Bias Assessment (IPM – QRB) for randomized trial.²⁵

Analysis of Evidence

The levels of clinical evidence and implications for recommendations of all of the included studies were graded according to “Grading strength of recommendations and quality of evidence in clinical guidelines” described by Guyatt et al²⁶, which has been cited by more than 1300 reports.^{27, 28} (Table 1)

Table 1.

Summary of Evidence Scores and Implications for Recommendation

Score	Description	Implication
1 A+	One or more RCTs of good methodological quality demonstrate effectiveness. The benefits clearly outweigh risk and burdens	Positive recommendation
1 B+	One or more RCTs with methodological weaknesses, demonstrate effectiveness. The benefits clearly outweigh risk and burdens
2 B+	One or more RCTs with methodological weaknesses, demonstrate effectiveness. Benefits closely balanced with risk and burdens
2 B±	Different RCTs or observational studies yield contradictory results better or no better than the control treatment. Benefits closely balanced with risk and burdens, or uncertainty in the estimates of benefits, risk and burdens	Considered, preferably study-related
2 C+	Effectiveness only demonstrated in observational studies. Given that there is no conclusive evidence of the effect, benefits closely balanced with risk and burdens
0	There is no literature or there are case reports available, but these are insufficient to suggest effectiveness and/or safety. These treatments should only be applied in relation to studies	Only study-related
2 C−	Observational studies indicate no or too short-lived effectiveness. Given that there is no positive clinical effect, risk and burdens outweigh the benefit	Negative recommendation
2 B−	One or more RCTs with methodological weaknesses, or large observational studies that do not indicate any superiority to the control treatment. Given that there is no positive clinical effect, risk and burdens outweigh the benefit
2 A−	RCT of a good quality which does not exhibit any clinical effect. Given that there is no positive clinical effect, risk and burdens outweigh the benefit

RCTs: Randomized controlled trials.

RESULTS

Detailed search strategy and results were shown in Appendix. After removing duplicates, we identified 1505 articles. Among which 557 abstracts were screened, 315 full-text articles assessed after reading the abstracts. A total of 267 full articles were reviewed thoroughly, and 27 articles were included in qualitative synthesis and analysis (Fig 1). We summarized the findings in Tables 2 and 3.

Fig 1.

PRISMA Flow Diagram of Literature Search and Selection Process

Table 2.

Advancements in the Pain Management of Refractory PDN

	Level of evidence	Reference	Type of study	N (size)	Results	Side effects
Botox - A	2B+	Yuan et al30	RCT	18	Forty four percent (8/18) of patients in the study achieved a VAS reduction of greater than 3 within 3 months of the injections	Pain at injection site
	2B+	Ghasemi et al31	RCT	18	Thirty percent of patients in the onabotunlinumtoxin A group report no pain after receiving the injections compared to zero percent of the placebo group.	Pain at injection site
	1B+	Lakhan et al32	Meta-Analysis	36	Meta-analysis of the above two RCTs favors Botulinum Toxin A in PDN	Infection at injection site was not statistically significant
	2B+	Restivo et al33	RCT	50	BTX-A (100 or 30 units) significantly improved calf or foot cramp in patients with PDN. Improvements started at one week and maintained up to 14 weeks	Mild pain in the injection site
Acupuncture	1B+	Chen et al34	Meta-analysis	1649	23 of 25 trials reported global improvement by manual acupuncture but there were high risk of bias	No trials reported adverse effects
	1B+	Dimitrova et al35	Meta-analysis	322	Four trials of PDN. Meta-analysis showed an overall benefit of acupuncture for PDN over control. The benefit remains after study heterogeneity was corrected.	No acupuncture-related serious adverse events reported
	1B+	Chao et al36	RCT	40	Acupuncture (once or twice a week) significantly reduced the average weekly pain score compared with usual care at week 12 but pain returns to baseline by week 18. Quality of life and Physical functioning improved in the acupuncture group by week 18.	15 mild side effects in 263 treatments. No serious adverse events were reported
	1B+	Shin et al37	RCT	98	The electroacupuncture (EA) treatment group showed significantly greater improvement on the short form McGill Pain Questionnaire, sleep interference scores, and the EuroQol-5 Dimensions questionnaire. The percentage of patients reporting improvement in the Patient Global Impression of Change was greater in the EA group.	No significant difference in the incidence of adverse events (AEs) and serious adverse events (SAEs) between groups
Surgical decompression of peripheral nerve	2B±	Dellon et al64	Multicenter Prospective	628	Tarsal tunnel release and neurolysis of the medial and lateral plantar and calcaneal tunnels reduced mean VAS from 8.5 to 2.0 at 6 months and remained for 3.5 years after the surgery. No control groups (nonsurgical or a negative Tinel sign) were included.	Not discussed.
	2B+	Liao et al65	Comparative	306	VAS, BPI-DPN, NCV were all improved in the surgical group (214 patients) compared to nonsurgical group (92 patients). Pain relief was better achieved in patients with “focal” than “diffuse” pain pattern. No patients in the surgical group developed new ulcer or required amputation during the 2-year follow-up.	Two patients had wound dehiscence and one had subcutaneous hemorrhage in surgical group. Ten patients developed lower extremity skin ulcer and two underwent amputation.
	1B+	Macare van Maurik, et al66	RCT	42	Surgical decompression of peripheral nerves provided significant limb pain relief in 73.7 % patients at 12 month compared to contralateral non-decompressed limb. Surgery did not change health-related Quality of Life score.	Anticoagulants cause hematoma in 1 patient, wound infection happened in 2 patients at the ankle site.
	1B+	Best, et al69	RCT	22	Surgery provided significant McGill pain VAS reduction at 3 and 6 month but not at 12 month compared to control groups. NeuroQol pain item sensitivity analysis demonstrated a statistically significant change of more than 3 points in pain scores at study end (12 months) compared to baseline.	Post-operative surgical site infection at the tarsal tunnel decompression site happened in 1 patient
	1B+	Tu et al71	Meta-analysis	1,825	Meta-analysis showed surgical patients achieved clinically and statistically significant improvement in VAS and two-point discrimination in lower extremities.	Superficial wound infection, dehiscence, hematoma, ulcer
	1B+	Albers et al72	Systematic review	N/A	Due to questionable definition of PDN, methodological design, issue of blindness, lack of specific controls/sham surgeries, and unexpected similar positive effect of sham surgery, the literature is insufficient to recommend surgical decompression of peripheral nerves for PDN	Not discussed
Sympathetic nerve block or neurolysis	2C+	Cheng et al38	Case report	1	A series of 9 lumbar sympathetic blocks over a 26-month period provided sustained pain relief and improved quality of life over a period of more than two years.	Not reported
	2C+	Ding et al39	Retrospective	90	90 patients were randomly assigned to anhydrous ethanol (AE) neurolysis, thermoradiofrequency (RF), or AE+RF group. Pain starts to return at 3 month after AE chemical blockade, 6 month after RF, one year after combined AE and RF. Total remission rates in AE, RF, and AE + RF group at one year was 66.7%, 73.3% and 93.3%, respectively.	No severe complications were observed.

Table 3.

Advancements in the Pain Management of Refractory PDN - SCS

SCS	2C+	Kumar et al42	Retrospective Study	4	Long-term pain relief in 3 out the 4 PDN patients on an average of 87 month follow-up (not clear whether the 4 PDN followed such long)	Infection, fractured electrode, hardware malfunction, electrode displacements – Not specifically for PDN
	2C+	Tesfaye et al43 Daousi et al44	Observational	10	Neuropathic pain and exercise tolerance improved in 6 patients at 14 months. Prolonged pain relief up to 7 years.	Superficial wound infection
	2C+	de Vos et al45	open-label, prospective	11	Nine patients had significant pain relief and eight reduced pain medication use at 6 months.	Mild infection in one patient resolved with antibiotics. Lead revision in two patients
	2C+	Pluijms et al46	Systematic review	25 total	SCS resulted in ≥ 50% pain relief reduced or withdrew analgesics usage in 60% patients at 1 year.	Infection (14%), technical complications (24%), death from unrelated cause (19%) and remaining complications (10%)
	2C+	Pluijms et al47 Slangen et al48	Prospective open-label cohort	15	Overall success rates of SCS on pain relief were 91% at 12 months, 55% at 24 months, and 64% at 36 months. Improvement in QoL was seen in 64% of the patients at 12 months, 55% at 24 months, and 64% at 36 months	No serious adverse events were encountered
	1B+	de Vos et al49; Duarte et al50	Multicenter RCT	60	Significant reduction in average VAS Pain score and improvement in health-related quality of life at 6 month in SCS group compared to best medical treatment	2 Infections, 1 femur fracture, and 1 cardiac arrest in the SCS group
	1B+	Slangen et al51; van Beek et al52; van Beek et al53	Multicenter RCT	36	At 6 month, 59% of patients in SCS group (vs 7% in best medical treatment group) had at least 50% pain reduction. 65% and 55% of patients in SCS group had ≥ 50% pain relief at 24 month and 5 years, respectively. About 80% of patients still use the SCS treatment after 5 years.	Postdural puncture headache and subdural hematoma in 1 case, infection and autonomic neuropathy in one case
	1B+	Peterson et al59	Multicenter RCT	216	High-frequency SCS was compared to SCS plus conventional medical management in refractory PDN. Responder rate, defined as at least 50% pain reduction from baseline, was 5% (5 of 93) in the CMM group and 85% (74 of 87) in the HF-10 SCS plus CMM group at 6 months. Neurological assessment including motor, sensory, and reflex testing were improved in 60% (52 of 87) in SCS plus CMM group (vs 3%, 3 of 93 in CMM group). 12 and 24 month results coming soon.	3 study-related adverse events for infection, 2 for wound dehiscence, and 1 for impaired healing among 5 of 90 subjects. 2 of 90 implants required explant.
	1B+	Duarte et al54	Meta-analysis	93	Meta-analysis showed significant pain reductions in patients received SCS compared with best medical therapy alone at the 6-month follow-up. The proportion of patients achieved at least a 50% reduction in pain intensity was significantly higher in SCS compared with best medical therapy. SCS also improved health-related quality of life.	Infection in 2 patients, pain from implanted IPG in 2 patients, one coagulopathy; one postdural puncture headache and subdural hematoma
	1B+	Raghu et al55	Meta-analysis	102	RCTs showed that tonic spinal cord stimulation provided greater pain relief than the best medical therapy at six months.	Not analyzed.
	2C+	de Vos et al57	Evaluation Studies	12	Patients who had at least six months of conventional tonic SCS underwent burst stimulation for two weeks. Burst SCS stimulation yielded significantly more pain reduction in 44% patients with little or no paresthesia.	Headaches, dizziness, and the sensation of “heavy legs”, “warm feet”

Risk of bias assessment

The risk of bias assessment is summarized in table 4. Most of the studies were judged to have a “High” risk of bias due to open-label design, lack of placebo-control, small sample sizes, and/or subjective outcome measurement.

Table 4.

Risk of bias assessment in included studies

Study	Risk of bias arising from the randomization process	Risk of bias due to deviations from the intended interventions	Risk of bias in missing outcome data	Risk of bias in measurement of the outcome	Risk of bias in selection of the reported results	Overall risk of bias
Yuan et al30	Low	Low	High	High	Some concerns	High
Ghasemi et al31	Low	Low	High	High	High	High
Lakhan et al32	Low	Low	High	Some concerns	Some concerns	Some concerns
Restivo et al33	Low	Low	Some concerns	High	Some concerns	High
Chen et al34	Low	Some concerns	High	High	Some concerns	High
Dimitrova et al35	Low	Some concerns	High	High	Some concerns	High
Chao et al36	Some concerns	High	High	High	Some concerns	High
Shin et al37	Low	Some concerns	High	High	Some concerns	High
Kumar et al42	High	Some concerns	High	High	Some concerns	High
Tesfaye et al43 Daousi et al44	High	Some concerns	High	High	High	High
de Vos et al45	High	Some concerns	High	High	High	High
Pluijms et al46	High	Some concerns	High	High	High	High
Pluijms et al47 Slangen et al48	High	Some concerns	High	High	Some concerns	High
de Vos et al49;Duarte et al50	Low	Low	Some concerns	High	Some concerns	Some concerns
Slangenet al51; van Beek et al52; van Beek et al53	Low	Low	Some concerns	High	Some concerns	Some concerns
Peterson et al59	Low	Low	Some concerns	High	Some concerns	Some concerns
Duarte et al54	Low	Low	Some concerns	High	Some concerns	Some concerns
Raghu et al55	Low	Low	Some concerns	High	Some concerns	Some concerns
de Vos et al57	High	Some concerns	Some concerns	High	High	High
Dellon et al64	High	Some concerns	High	High	High	High
Liao et al65	High	Some concerns	Some concerns	High	High	High
Macare van Maurik et al66	Low	Some concerns	Some concerns	High	High	High
Best et al69	Low	Some concerns	Some concerns	High	High	High
Tu et al71	High	Some concerns	Some concerns	High	High	High
Albers et al72	High	Some concerns	Some concerns	High	High	High
Cheng et al38	High	Low	High	High	Some concerns	High
Ding et al39	Some concerns	Low	High	High	Some concerns	High

Botulinum toxin-A (BTX-A) injections

BTX-A has been used to treat neuropathic pain as well as spasticity.²⁹ In a RCT of 18 patients with PDN, nine patients received intradermal injection of 50 units of BTX-A over twelve spots in the foot and nine patients received placebo injections.³⁰ After twelve weeks, patients were allowed to cross-over to the other group. At 12 weeks post-injection, 44% (8/18) of patients in the study achieved a visual analog scale (VAS) reduction of greater than 3. Patients reported no major complications.³⁰ In another RCT, 40 patients were randomized to receive intradermal injections of either 100 units of BTX-A or 0.9% saline in twelve spots in the foot.³¹ At three weeks post-injection, 30% of patients in the BTX-A group reported “no pain”, compared to 0% of patients in the placebo group. No major adverse effects were reported.³¹ A meta-analysis based on these two reports favored BTX-A but also cautioned that it might be best used as an adjunctive treatment for PDN because of the small effect size of the studies.³² In a recent RCT, BTX-A was used to treat muscle cramps and its associated pain in the calf or foot in patients with PDN. A total of 50 patients were randomized to intramuscular injection of saline or BTX-A into the gastrocnemius (100 units) or flexor muscles of the foot (30 units).³³ The intensity and frequency of muscles cramps were significantly improved in the BTX-A group, compared with placebo control. Improvements started at one week and maintained up to 14 weeks.

In summary, BTX-A injections showed moderate level of evidence (2B+/1B+) in managing PDN (50 or 100 units intradermal injection for pain; 100 or 30 units intramuscular for pain and muscle cramping).

Acupuncture

Acupuncture has become a part of pragmatic and integrative treatment for PDN. Chen et al reviewed 25 RCTs (1649 subjects) of manual acupuncture for PDN.³⁴ The majority of the trials (23 of 25) reported improvement in global symptoms, but these trials have high risk of bias (methodology or publication). A meta-analysis of 15 studies (among which, 4 RCTs studied PDN) on peripheral neuropathy showed an overall benefit of acupuncture for PDN over sham control.³⁵ The benefit remains after study heterogeneity was corrected. In a recently-published RCT, 40 patients with PDN were randomized to receive usual care only, usual care with acupuncture once or twice weekly for 12 weeks.³⁶ At week 12, the weekly average pain intensity was significantly reduced from baseline in those receiving acupuncture, compared with those receiving usual care only, but the pain intensity returned to baseline levels by week 18. Quality of life scores and physical functioning improved in the acupuncture groups by week 18. No significant differences were observed between the once/week vs. twice/week acupuncture groups. The effects of electroacupuncture for PDN was evaluated in a multicenter, randomized, assessor-blinded, and controlled trial.³⁷ Patients in the electroacupuncture treatment group showed significantly improvement in pain intensity (on 0–10 numerical rating scale [NRS]) than those in the control group. While these results are promising, the authors recognize the methodological problems with standardization, control, blinding, and outcome measures.

In summary, multiple RCTs showed beneficial effects of acupuncture on PDN management. Acupuncture can be considered an adjunctive therapy with minimal side effects for patients with PDN based upon moderate level of evidence (2B+ or 1B+), mainly due to methodological and blinding concerns.

Lumbar sympathetic ganglion block and neurolysis

The sympathetic nervous system may contribute to PDN. Cheng reported a 37-year old male with PDN which was diagnosed by clinical presentations and skin biopsy.³⁸ The patient failed multiple conservative therapies. A series of 9 lumbar sympathetic blocks over a 26-month period provided sustained pain relief and improved quality of life over a period of more than two years. Lumbar sympathetic neurolysis has also been reported to treat PDN. In a retrospective comparative study, ninety patients with PDN were treated with anhydrous ethanol (AE) chemical blockade, radiofrequency thermocoagulation (RF), or combined AE and RF of the lumbar sympathetic ganglion (AE+RF).³⁹ Postoperative VAS were significantly decreased from preoperative baseline. Pain started to return at 3 month after AE chemical blockade, 6 month after RF, one year after combined AE and RF. The complete remission rates for AE, RF, and AE+RF groups at one year were 66.7%, 73.3% and 93.3%, respectively. No severe complications were observed. The authors concluded that radiofrequency thermocoagulation combined with AE chemical blockade of the lumbar sympathetic ganglion was safe and effective in managing PDN.

In summary, there is week evidence (2C+) to support application of lumbar sympathetic ganglion block and neurolysis in selected patients with PDN, who have failed pharmacological and other therapies (see below).

Spinal cord stimulation (SCS)

SCS has emerged as a cutting-edge treatment for chronic refractory neuropathic pain.^{40, 41} In a case series of SCS implants in 276 patients with peripheral neuropathy, 4 patients with PDN had excellent early pain relief and 3 achieved long-term success with an average of 87-month follow-up.⁴² Another case series reports that 8 out of 10 patients with refractory PDN had significant pain relief during SCS trial and after permanent implant at 3, 6, and 14 months.⁴³ By the end of the study (14 months), 6 patients continued to use SCS as the sole treatment for PDN with significant pain relief. SCS used in this small sample size study was paresthesia-based thus blinding was impossible; however, the authors argued against placebo effects by pointing out that the pain relief was lost immediately when there was a lead displacement. The authors speculated that 2 of the patients did not respond to SCS due to loss or gross dysfunction of the inhibitory A-beta fibers. A study with extended follow-ups of up to 7 years after implantation revealed that the pain relief was maintained with few associated complications.⁴⁴ In an open-label, prospective study, 11 patients with refractory PDN in the lower extremities received thoracic SCS implantation and 9 patients had significant pain relief.⁴⁵ Average VAS pain score decreased from 77mm at baseline to 34mm at 6 months after implantation. At the end of the study at 6 months, 8 of the 9 patients significantly reduced pain medication use and 6 patients used SCS as the sole treatment for the PDN.⁴⁵ A systematic review of the above-mentioned 4 studies with a total of 25 patients summarized that, at one year post-implantation, SCS resulted in ≥50% pain relief in 63% of patients with PDN.⁴⁶ Also, analgesics usage was reduced in most SCS-treated patients in the same time. The authors then reported a pilot study in 15 patients with refractory PDN.⁴⁷ Eleven patients achieved clinically significant pain reduction, defined as ≥50% decrease of pain intensity, during a 2-week SCS trial, and 10 patients achieved the treatment objective at 12 months after the SCS implant. Quality of life and neuropathic pain scores were significantly improved at in follow-up visits at 2 weeks and 3 to 12 months.⁴⁷ The effects of SCS were maintained at 12, 24, and 36 months, with 91%, 55%, and 64% patients continued to have ≥50% pain reduction, respectively. Improvement in quality of life at 12, 24, and 36 months was reported in 64%, 55% and 64% of the patients.⁴⁸

The first multicenter RCT of SCS for refractory PDN was reported in 2014.⁴⁹ 60 Patients were randomized to best medical treatment with or without SCS and followed up for 6 months. The average VAS pain score decreased from 73mm at baseline to 31mm in the SCS group whereas there was no changes in the non-SCS group. Patients in the SCS group also had improved quality of life measured with the EuroQoL 5D questionnaires.^{49, 50} Later in 2014, another RCT studied SCS for PDN in 36 patients and reported a trial success rate of 77%.⁵¹ Treatment success at 6 months, defined as ≥50% pain relief, was reported in 59% of patients in the SCS group and 7% in the best medical treatment only group. Pain and sleep were “very much improved” in 55% and 36% respectively in the SCS group, whereas there were no changes in the medical treatment only group.⁵¹ After 6 months, 93% of patients in the medical treatment only group crossed over to SCS treatment with success rates of 65% patients at 24-month⁵² and 55% at 5-year follow-ups.⁵³ About 80% of patients continued to use the SCS treatment after 5 years.⁵³ Recent systematic reviews and meta-analyses of these studies demonstrated that SCS is an effective therapy in reducing PDN pain.^{54, 55}

All of the above-mentioned studies used conventional, paresthesia-based SCS, which makes blinding of the treatments questionable. Some patients may also complaint about discomfort of paresthesia associated with the stimulation. Recently, burst stimulation using five high-frequency pulses at 500 Hz running 40 bursts a second was tested and paresthesia was barely present.⁵⁶ In a study of 12 patients with refractory PDN, conventional tonic SCS was switched to burst SCS for two weeks.⁵⁷ It was found that tonic stimulation reduced the average VAS score from 70mm to 28mm while the burst stimulation further decreased the pain score to 16mm. Eight out of the 12 patients (67%) had additional pain reduction with burst stimulation as compared with tonic stimulation.⁵⁷

Completely paresthesia-free, high frequency SCS (HF-10, 10 kHz) was implanted in 8 patients with PDN in a pilot study. Among the 7 patients who attended the 12-month follow-up, 6 had at least 50% pain relief and 5 demonstrated improvements in sensory and reflex testing.⁵⁸ Peterson et al. recently reported a multicenter RCT study of HF-10 SCS for refractory PDN mainly in the lower extremities.⁵⁹ A total of 216 patients were enrolled; 103 were randomized to conventional medical management (CMM) and 113 to HF-10 SCS plus CMM group. Patients were followed up at 3, 6, 12, and 24 months. At 6 months, patients could opt to cross-over to the other treatment arm if they had insufficient (less than 50%) pain relief. As high as 82% (76 of 93) of patients in the CMM group elected to cross over while no patients in the SCS group did. The proportion of responders with ≥ 50% pain reduction from baseline, was 5% (5 of 93) in the CMM group and 85% (74 of 87) in the SCS group at 6 months. Neurological assessments, including motor, sensory, and reflex testing, were improved in 60% (52 of 87) of patients in the SCS group, in contrast to 3% (3 of 93) in the CMM group. Sleep quality and the Global Assessment of Functioning were also improved in the SCS group but not the CMM group. HbA1c level increased in both groups at 6 months. Study-related adverse events included infection in 3 patients, wound dehiscence in 2 patients, impaired healing in 1 patient, among a total of 90 subjects; 2 out of 90 implants required explant. The study is limited by non-blinding and potential placebo effects, which may be mitigated through the ongoing long-term follow-up.

In summary, these studies provide moderate to strong evidence to support the use of SCS in treating PDN in the lower extremities (Evidence level: 1B+) while the efficacy and safety of SCS for upper extremity PDN remains to be investigated.

Dorsal root ganglion (DRG) stimulation

In a retrospective case series, DRG stimulation with up to four quadripolar percutaneous leads between L2 and L5 were used in 10 patients with refractory PDN in the lower extremities.⁶⁰ Seven patients received permanent implant after successful trials. Four of five patients at 12 month follow-up had an average pain reduction of 64%. In a multicenter retrospective case series using DRG stimulation to treat peripheral neuropathy, 75–100% VAS improvement was reported in two patients with PDN at six-week follow up.⁶¹ One patient completely discontinued the use of gabapentin while there was no change in drug use in the other patient. Interestingly, unilateral DRG leads placed at T12 and S1 significantly improved pain, disability, quality of life in a patient with refractory PDN in both lower extremities. The patient had significant relief of PDN-related symptoms of both feet.⁶² Overall, the evidence of use DRG stimulation for PDN is limited to small sample size case series (2C+).

Surgical decompression of the peripheral nerves

The metabolic derangement in PDN may render peripheral nerves susceptible to compression. Surgical decompression may help relieve pain in select cases.⁶³ In a multicenter prospective study, 628 subjects with diabetes underwent tarsal tunnel release and neurolysis of the medial and lateral plantar and calcaneal tunnels.⁶⁴ The procedures resulted in a significant decrease in mean VAS score from 8.5 to 2.0 at 6 month follow up. The effects were maintained at 3.5 years after the surgery. However, interpretation of these results of this study is compromised due to lack of a control group. A single center prospective study compared a surgical group (214 patients) with a nonsurgical group (92 patients) and reported significant improvements in VAS pain score, Brief Pain Inventory Short Form for diabetic peripheral neuropathy questionnaire, two-point discrimination, nerve conduction velocity and high-resolution ultrasonography (cross-sectional area).⁶⁵ Pain relief was better in patients with “focal” than “diffuse” pain pattern.

A single-center RCT in 42 patients with PDN evaluated unilateral surgical decompression of the tibial, superficial, deep and common peroneal nerves while using the contralateral limb as control (within-patient comparison).⁶⁶ The VAS scores improved significantly in 73.7% patients from a mean of 6.1 preoperatively to 3.5 postoperatively at 12 months, of which 35.7% patients had a decrease of more than 5 points. In a related article, the authors reported 42.5% of the study subjects achieved clinically important difference in VAS at 12 month.⁶⁷ But the surgery did not improve health-related quality of life scores⁶⁷ or the results of nerve conduction studies.⁶⁸ In another randomized single-blind study, 12 patients underwent lower extremity decompression surgery of the common peroneal, tibial, and deep peroneal nerves and 10 patients received usual care for 1 year.⁶⁹ Patients in the surgery group had significant VAS reduction at 3 and 6 month compared to the control groups. Although this benefit waned at 12 month, NeuroQol pain item sensitivity analysis demonstrated a statistically significant change of more than 3 points in VAS pain scores at 12 months compared to baseline. No statistically significant differences in quality of life were detected within or between groups. Patients in the surgical group had over 3 times the odds of rating their pain as “better” compared those in the control group at 12 months. A recent retrospective study in 36 patients with PDN reported that the mean NRS score decreased significantly at 6 days, 6 months, and 12 months after nerve decompression surgery.⁷⁰ At 12 months, 64.7% of patients had at least 50 percent reduction in NRS score from preoperative baseline.

A systematic review identified 8 prospective and 4 retrospective studies with a total of 1825 patients.⁷¹ Meta-analysis showed that surgical decompression achieved clinically and statistically significant improvements in VAS and two-point discrimination in the lower extremities. In contrast, a structured review concluded that the literature is insufficient to recommend surgical decompression of peripheral nerves for PDN, due to questionable definition of PDN, methodological design, issue of blindness, lack of specific controls/sham surgeries, and unexpected similar positive effect of sham surgery (reported only in an Abstract and through the personal communications between the authors).⁷² This conclusion is consistent with the recommendations based on studies published before 2006 by the American Academy of Neurology (AAN) Practice Advisory.⁷³ A RCT to assess cost-effectiveness of surgical decompression of peripheral nerves in the lower extremity compared to non-surgical care is underway.⁷⁴

In summary, large observational studies and small-sample sized RCTs provide moderate to strong evidence (2B± to 1B+) for surgical decompression of peripheral nerves to manage PDN. Well-designed high-quality RCTs are needed to overcome the limitations of published studies and to ascertain the role of surgical decompression in the management of PDN.

DISCUSSION

PDN is often debilitating and refractory to conventional pharmacological therapies including anticonvulsants and antidepressants. For patients who do not respond or unable to tolerate medications, several interventional therapies have been investigated. SCS has been shown to be an effective treatment for neuropathic pain based on numerous clinical trials.^{40, 41, 75} However, not all patients with neuropathic pain respond to SCS. Therefore, a trial stimulation is usually performed before a permanent implant of the stimulator. The success rate of achieving ≥50 % pain reduction during a SCS trial is higher in patients with PDN (up to 77% or 84%)^{46, 51} compared to those with other neuropathic pain syndromes such as postherpetic neuralgia and complex regional pain syndrome (roughly 50%).⁴⁰ The reason for this difference is unclear and could be due to predominantly sensory involvement in PDN. Factors that predict SCS treatment success have not been clearly identified. Preoperative clinical sensory testing failed to identify responders to SCS,⁴⁷ even though modulation of A-beta fiber is thought to be one of the major mechanisms underlying SCS effect and the loss of A-beta (e.g., complete absence of vibration and joint-position) is considered to predict unresponsiveness to SCS.⁴³ There is no difference in baseline severity of neuropathy measured with the Michigan Diabetic Neuropathy Score (MDNS) between responders and non-responders.⁴⁷ However, higher baseline MDNS was associated with higher rate of long-term treatment failure and subsequent removal of the SCS system. In contrast, a higher baseline night NRS pain score was associated with a lower risk of treatment failure in a 5-year follow-up study.⁵³ In a small-size pilot study (15 patients with PDN), the forearm contact heat-evoked potential (CHEP, to test small-diameter nerve fiber function) latencies are increased in patients who responded to SCS treatment (10 out of 15 patients) as compared to non-responders (5 out of 15 patients).⁷⁶

Conventional SCS induces paresthesia to map the painful area and to determine the placement of electrodes. Compared to paresthesia-based conventional SCS, burst SCS or DRG stimulation (both with little or no paresthesia) appears to be more attractive to PDN patients who often suffer from tingling and numbness.⁵⁷ However, evidence supporting the use of burst or DRG stimulation in PDN is limited to case series. The recent multicenter trial using paresthesia-free (HF-10) SCS in 90 patients confirmed its efficacy for PDN.⁵⁹ To date, all SCS studies selected patients with refractory PDN in the lower extremities. It remains to be determined whether SCS could help relieve refractory PDN pain in the upper extremities.

Diabetes is an independent risk factor for surgical site infection for multiple surgical procedures, higher in cardiac than other surgeries.⁷⁷ Although diabetes did not independently increase the rate of infection in patients received SCS,⁷⁸ earlier studies reported that the incidence of infection after SCS in PDN patients (14%)⁴⁶ was much higher than that in the general population receiving SCS (2.45% – 4.5%).^{78, 79} However, the recent HF-10 SCS trial in PDN patients reported three study-related infections out of 90 SCS implants (3%).⁵⁹ This low rate of infection might be due to a relatively tight inclusion criteria (HbA1c ≤10%, BMI<45) in this cohort, although patient BMI was not associated with surgical site infection in a meta-analysis.⁷⁷ It is difficult to make specific recommendations in preventing surgical site infection in PDN patients undergoing SCS procedure since few studies addressed the causes of surgical site infection in patients with diabetes. Nevertheless, general guidelines to prevent surgical site infection should be strictly followed.⁸⁰

The AAN did not recommend surgical decompression for PDN in its practice advisory published in 2006.⁷³ Since then, large observational studies and small RCTs have provided evidence to suggest that surgical decompression of peripheral nerves in the lower extremity can be considered in patients with PDN and superimposed compression of specific peripheral nerves.^{66, 69}Surgical decompression significantly improved pain^{66, 69} but data on whether it would also improve quality of life are conflicting.^{67, 69}

PDN pain might be sympathetically-maintained in some patients,⁸¹ so that lumbar sympathetic block or neurolysis may provide pain relief in the lower extremities.³⁸ The sympathetic block could be a viable option for patients with refractory PDN. It can be considered for patients who are waiting for SCS or surgical decompression of peripheral nerves. Lumbar sympathetic neurolysis should be used with great caution as it may cause severe adverse effects.⁸²

Acupuncture has been reported to be beneficial and cost-effective in help manage chronic pain including myofascial pain, fibromyalgia, and neuropathic pain.⁸³ Acupuncture may modulate the endogenous opioid system, desensitize peripheral nociceptors, suppress release of pro-inflammatory cytokines, and inhibit spinal glial activation.^84–86 Both neuropathic symptoms and nerve conduction study parameters have been shown to be affected by acupuncture.³⁵ In this review, we identified multiple RCTs and meta-analyses that reported clinical efficacy of acupuncture for PDN. Studies on acupuncture for PDN, as for other pain conditions, are often challenged by concerns for methodology, blindness, standardization, and placebo effects. Nevertheless, acupuncture can be considered an adjunctive therapy for patients with PDN without causing significant complications.

A few RCTs of small sample sizes reported that intradermal^{30, 31} or intramuscular³³ injections of BTX-A provided better relief in pain and muscle cramp in the lower limbs as compared to saline injection in patients with PDN. Botulinum toxin may relieve neuropathic pain via different mechanisms such as by inhibiting the release of pain mediators (e.g., substance P, calcitonin gene related gene) from the nerve endings and DRG, deactivating sodium channels, and reducing inflammation.⁸⁷ Botulinum toxin also acts at the presynaptic membrane of the neuromuscular junction to prevent calcium–dependent release of acetylcholine, leading to biochemical denervation and muscle weakness.⁸⁸ The efficacy and safety of intradermal and intramuscular BTX-A injection have been compared and the maximum improvement of wrinkles were similar in both groups but the intramuscular group reported more muscle weakness related side effects.⁸⁹ Besides its peripheral effect, recently studies indicate that botulinum toxin may also cause changes in the central nervous system to modulate sensory inputs.⁸⁸ It has been reported that BTX-A injection significantly reduced the tactile threshold and mechanical pain threshold in patients with PDN.⁹⁰ In addition to PDN, botulinum toxin has been used to treat other types of neuropathic pain with minimal side effects and could reduce the use of invasive or surgical procedures.⁹¹

Conclusions

In conclusion, emerging evidence indicates that interventional therapies can be effective in the management of refractory PDN. Based on systematic review of evidence, we recommend that SCS should be considered to reduce pain and improve quality of life (evidence level: 1B+) for patients with refractory PDN in the lower extremities. Acupuncture or botulinum toxin injection can be considered as an adjunctive therapy for PDN (2B+/1B+). Surgical decompression of specific peripheral nerves in the lower extremity can be considered in PDN patients with superimposed compression of the nerves (2B±/1B+). Lumbar sympathetic block can be used in select patients if the patient had a favorable response to the block (2C+). High-quality RCTs are warranted to further strengthen the evidence for these interventional treatments and to bridge many gaps identified in this review.

Key Points:

Questions:

What options are available to manage patients with PDN beyond pharmaceutical therapies?

Findings:

Emerging evidence indicates that interventional therapies such as SCS, acupuncture, Botulinum toxin A injection, and surgical decompression of lower limb peripheral nerves can provide clinically meaningful pain relief in patients with refractory PDN.

Meaning:

When PDN is refractory to pharmaceutical therapies, clinician should explore interventional management options.

Glossary of Terms:

AAN

American Academy of Neurology

AE

anhydrous ethanol

BTX-A

Botulinum toxin-A

CMM

conventional medical management

DRG

dorsal root ganglion

DSP

Diabetic sensorimotor polyneuropathy

HF-10

High frequency 10 kHz

MDNS

Michigan Diabetic Neuropathy Score

NRS

numerical rating scale

PDN

Painful diabetic neuropathy

PRISMA

Preferred Reporting Items for Systematic Reviews and Meta-Analyses

RCT

randomized clinical trials

RF

radiofrequency thermocoagulation

SCS

spinal cord stimulation

VAS

visual analog scale