Long-term efficacy of high-frequency (10 kHz) spinal cord stimulation for the treatment of painful diabetic neuropathy: 24-Month results of a randomized controlled trial

Aims: To evaluate the long-term efficacy of high-frequency (10 kHz) spinal cord stimulation (SCS) for treating refractory painful diabetic neuropathy (PDN). Methods: The SENZA-PDN study was a prospective, multicenter, randomized controlled trial that compared conventional medical management (CMM) alone with 10 kHz SCS plus CMM (10 kHz SCS+CMM) in 216 patients with refractory PDN. After 6 months, participants with insufficient pain relief could cross over to the other treatment. In total, 142 patients with a 10 kHz SCS system were followed for 24 months, including 84 initial 10 kHz SCS+CMM recipients and 58 crossovers from CMM alone. Assessments included pain intensity, health-related quality of life (HRQoL), sleep, and neurological function. Investigators assessed neurological function via sensory, reflex, and motor tests. They identified a clinically meaningful improvement relative to the baseline assessment if there was a significant persistent improvement in neurological function that impacted the participant’s well-being and was attributable to a neurological finding. Results: At 24 months, 10 kHz SCS reduced pain by a mean of 79.9% compared to baseline, with 90.1% of participants experiencing ≥50% pain relief. Participants had significantly improved HRQoL and sleep, and 65.7% demonstrated clinically meaningful neurological improvement. Five (3.2%) SCS systems were explanted due to infection. Conclusions: Over 24 months, 10 kHz SCS provided durable pain relief and significant improvements in HRQoL and sleep. Furthermore, the majority of participants demonstrated neurological improvement. These long-term data support 10 kHz SCS as a safe and highly effective therapy for PDN. Trial Registration: ClincalTrials.gov Identifier, NCT03228420.


Introduction
Globally, the number of people with diabetes has quadrupled in the last two decades [1].Diabetic neuropathy (DN) is a frequent long-term complication of diabetes, with an estimated lifetime prevalence exceeding 50% [2].Typically starting in the feet and gradually ascending to the lower legs [3], symptoms classically manifest as numbness, tingling/ paresthesia, loss of protective sensation, impaired balance and weakness, and reduced response to stimuli such as cold temperature and pinprick [3].Over time, as the feet become progressively insensate, the risk of injury increases, leading to a greater likelihood of falls, fractures, foot ulceration, and lower limb amputation [3][4][5][6][7].
Painful diabetic neuropathy (PDN), affecting up to 25% of people with diabetes [8], is characterized by painful burning, lancinating, tingling, and/or shooting sensations, which can be severe and continuous, particularly at night [3].Consequently, individuals with PDN suffer significantly reduced health-related quality of life (HRQoL), impaired functionality, and comorbidities such as sleep disorders and depression/anxiety [9].
Long-term management of PDN usually includes the prescription of oral neuropathic pain medications [3].However, the effectiveness of these medications is relatively low, as measured by the number needed to treat (NNT; the number of patients that need to be treated before we expect one extra patient to be a treatment responder versus the control treatment) [10][11][12][13][14]. Additionally, many patients discontinue these medications, primarily due to inadequate pain relief and/or intolerable side effects [15].Although the FDA has recently approved an 8% capsaicin patch as an alternative treatment, its effectiveness is also limited [13,16].Overall, existing pharmacotherapies fail to provide long-term pain relief for the majority of PDN patients.
Spinal cord stimulation (SCS) is a safe and reversible nonpharmacological therapy for chronic neuropathic pain.In the traditional form of SCS, low-frequency pulses  are applied to the spinal cord to induce continuous paresthesia over the painful area, thereby masking the pain sensation.A more recent SCS innovation uses high-frequency 10 kHz pulses ("10 kHz SCS") to relieve pain without paresthesia, which may be more comfortable for patients [17,18].Recent results from a large randomized controlled trial (RCT; "SENZA-PDN study") show that 10 kHz SCS provides substantial pain relief in PDN patients, with many demonstrating improved sensation on neurological examination [19,20].Based on these findings, the FDA approved 10 kHz SCS for this indication in 2021.The American Association of Clinical Endocrinology and experts in a recent Clinical Compendia Series from the American Diabetes Association have also recommended 10 kHz SCS therapy as a treatment for refractory PDN pain with a high level of evidence [3,21].
Given the severe and chronic nature of PDN, any analgesic treatment must provide robust, durable pain relief.Here, we present the results of the SENZA-PDN study after 6 and 24 months of treatment with 10 kHz SCS.

Study design
The SENZA-PDN study was a multicenter, prospective, randomized, open-label clinical study conducted in the United States at 18 centers.A detailed description of the study design and procedures has previously been published [22].Prior to study initiation, Institutional Review Board (IRB) approval was obtained (Western IRB or local site IRB, when applicable).The study was conducted following the Declaration of Helsinki and Good Clinical Practices, reported under the Consolidated Standards of Reporting Trials (CONSORT) guideline [19], and registered with ClinicalTrials.gov(NCT03228420).

Participants
To be eligible for study enrollment, patients had to have PDN symptoms for 12 months or longer that were refractory to current or prior treatment with a gabapentinoid and at least 1 other class of analgesic drug, experience lower limb pain intensity of ≥5 cm on a 10-cm visual analog scale (VAS), hemoglobin A1c (HbA1c) ≤10% (86 mmol/mol), and be suitable surgical candidates.Exclusion criteria included upper extremity pain intensity due to diabetic neuropathy of ≥3 cm on a 10-cm VAS, body mass index >45 kg/m 2 , daily opioid dosage >120 mg morphine equivalents, and not an appropriate candidate for implantation or study involvement based on a recent psychological assessment.

Randomization
Patients who met eligibility requirements and provided written informed consent were randomized (1:1) to conventional medical management (CMM) alone or 10 kHz SCS plus CMM (10 kHz SCS+CMM).Randomization procedures have previously been reported [19].At the end of the 6-month randomized phase, patients could cross to the other treatment arm if they had less than 50% pain relief from baseline, were dissatisfied with their treatment, and the investigator agreed that the switch was appropriate.Patients who received 10 kHz SCS therapy (including the original 10 kHz SCS+CMM recipients and CMM-to-10 kHz SCS+CMM crossover patients) were followed up for 24 months postimplantation.

Interventions
Interventions have been described previously [19,20].In brief, investigators administered CMM to all enrolled patients according to their standard of care for PDN and diabetes.Individuals receiving 10 kHz SCS underwent a 5-7 day trial with an external stimulator connected to 2 percutaneous leads placed in the posterior epidural space.Patients who experienced ≥50% pain relief from baseline (ie, trial success) were eligible to receive a permanent 10 kHz SCS system (Nevro Corp.) [17].
Patient safety was also evaluated by monitoring adverse events and performing a comprehensive neurological evaluation at baseline and selected follow-up visits.As previously reported, independent neurologists developed this examination and trained the investigators to assess motor, sensory, and reflex function [19,20].At each follow-up neurological examination, investigators used their clinical judgment to determine if there was a clinically meaningful deficit, no change, or a clinically meaningful improvement from baseline in each motor, sensory, and reflex function.From these results, overall neurological status was classified as improvement, maintenance, or deficit via the following criteria: improvement (improvement in at least 1 function without any deficit in any other function), maintenance (maintenance in all functions), or deficit (deficit in any function).

Statistical methods
A previous publication describes and reports the sample size calculation [22].The current statistical analysis evaluated the results at the end of (1) the randomized phase, ie, 6 months after baseline, and (2) the postimplantation phase, ie, 24 months after implantation with a permanent 10 kHz SCS system (this includes the original 10 kHz SCS+CMM group and the CMM-to-10 kHz SCS+CMM crossover cohort).Statistical analyses were conducted separately for each phase.
For the randomized phase, results were evaluated for the CMM alone group and the 10 kHz SCS+CMM group.For the postimplantation phase, results were evaluated for (1) implanted participants from the original 10 kHz SCS+CMM group, (2) implanted participants from the CMM-to-10 kHz SCS+CMM crossover cohort, and (3) all implanted participants.We defined the preimplantation time point as study baseline in the original 10 kHz SCS+CMM group and the end of the randomized phase in the CMM-to-10 kHz SCS+CMM crossover patients.
In addition to raw scores and percentage change from baseline or preimplantation, we evaluated lower limb pain intensity VAS data at 6 and 24 months according to responders (≥50% pain relief from baseline or preimplantation) and profound responders (≥80% pain relief from baseline or preimplantation).Using standard methods, we also calculated the NNT [29].
Results are presented for all available data at each study visit.Statistical analyses were conducted in SAS (Version 9.4, SAS Institute Inc., Cary, NC).To statistically evaluate time and group effects for continuous and ordinal variables, missing data was first imputed via a multiple imputation procedure using all available nonmissing data for that outcome.Each imputed dataset was then analyzed with a repeated-measures linear model, and these results were summarized using the MIANALYZE procedure.The repeated-measures model included time and group as fixed effects and subject as a random effect, with an autoregressive correlation structure of order 1.Similar logistic and nonlinear repeatedmeasures models were used for proportions and percentage change in outcomes over time, respectively.However, imputation was not possible for 6 participants who withdrew prior to any neurological/sensory assessments follow-up.Consequently, these participants were classified as not having neurological or sensory improvement.

Study population and demographics
The study investigators enrolled patients between August 28, 2017, andAugust 23, 2019 (Fig. 1).A total of 113 participants were randomized to receive 10 kHz SCS+CMM and 103 to CMM alone, with optional crossover at 6 months if specific criteria were met.
Baseline characteristics for the 216 randomized participants have previously been reported [19,20].In brief, the participants had a mean age (SD) of 60.8 years (10.7), and 63% were male.Median durations (interquartile range) of diabetes and PDN symptoms were 10.9 (6.3-16.4) and 5.6 years (3.0-10.1),respectively.Baseline characteristics were comparable between the randomized groups.Baseline neurological exam results (reflexes, motor strength, and sensory function) were also similar between the randomized groups, with the only significant difference noted in a higher percentage of normal responses for light touch in the CMM alone group (73.9% vs 65.0% for the CMM alone and 10 kHz SCS+CMM groups, respectively; P <.001, Fisher's Exact Test).
Among the 113 participants originally assigned to 10 kHz SCS+CMM, 104 completed the temporary stimulation trial, with 98 (94.2%) reporting pain relief of at least 50% from baseline (ie, trial success).Of the 90 patients who received a permanent system, 88 completed 6 months of follow-up, and 84 completed 24 months of follow-up.At 6 months, none of the original 10 kHz SCS+CMM recipients elected to cross over to the CMM alone arm.
In the control group (CMM alone), 95 participants completed the 6-month follow-up, with 83 eligible to cross over to the 10 kHz SCS+CMM arm.Of these, 77 (92.8%) crossed over and underwent a temporary stimulation trial, with 73 (94.8%) reporting trial success.Fifty-eight of the 64 patients who underwent permanent implantation completed 24 months of follow-up after their preimplantation assessment.
A total of 181 patients underwent a 10 kHz SCS trial, including the original 10 kHz SCS+CMM recipients and those who received 10 kHz SCS+CMM after crossing over from CMM alone, ie, the CMM-to-10 kHz SCS+CMM crossover group.Of these, 171 (94.5%) had a successful trial, 154 underwent permanent implantation, and 142 completed 24 months of follow-up (Fig. 1).

Safety
Through 24 months, no stimulation-related neurological deficits occurred, and no devices were explanted due to lack of efficacy.Among 154 permanently implanted participants, 7 (4.5%)experienced a study-related serious adverse event (SAE), and 8 (5.2%) had a procedure-related infection.Three infections resolved with standard treatment, while 5 required explantation (3.2% of all implanted patients).Of the 5 explanted patients, 4 exited the study, while 1 continued participation after reimplantation.One additional explant occurred as a precaution for an unrelated infection.Five other participants (3.2%) underwent revision surgery to reposition or replace the implantable pulse generator (IPG), and 3 (1.9%)had their leads repositioned or replaced due to migration.All participants with repositioned/ replaced system components remained in the study.One additional infection occurred during the stimulation trial that resolved with standard treatment.

Postimplantation phase-Investigators assessed neurological function versus
study baseline in all implanted patients.After 24 months of 10 kHz SCS, 92 of 140 implanted individuals (65.7%; 95% CI, 57.5%-73.1%)exhibited a clinically meaningful improvement over study baseline in sensory, motor, or reflex function, without worsening in any category.Most of the neurological gains were observed in sensory function (65.0% of participants assessed to have improved; 95% CI, 56.8%-72.4%;91 of 140).Additionally, the reported neurological and sensory improvement outcomes were similar between the original 10 kHz SCS+CMM group and the CMM-to-10 kHz SCS+CMM crossover cohort, with the initial 10 kHz SCS recipients showing higher improvement rates that reached statistical significance for neurological function at 24 months postimplantation (P =.048 for neurological improvement and P =.076 for sensory improvement; Fig. 4C and Fig. 4D).
3.6.Health-Related quality of life 3.6.1.Randomized phase-Health-related quality of life significantly improved after 6 months of 10 kHz SCS, based on the EuroQol 5-Dimensional 5-Level (EQ-5D-5L) questionnaire.The mean EQ-5D-5L index value increased by 0.130 (95% CI, 0.097-0.163;P <.001) from baseline in the 10 kHz SCS recipients, while the index value for the control participants declined by 0.031 but was not significantly different from baseline (P =.090; Fig. 5A).Additionally, at the 6-month assessment, the 10 kHz SCS recipients had significantly higher HRQoL than the control participants (P <.001).

Discussion
The SENZA-PDN study, the largest RCT to evaluate SCS in PDN patients to date, found that 10 kHz SCS provided significant and durable pain relief as well as improved HRQoL and sleep over the 24-month study period.Comparable to the 6-and 12-month study outcomes, the 24-month results confirm the robust and long-lasting beneficial effects of 10 kHz SCS in PDN patients [19,20,30].These are important findings, given the severe, chronic, and debilitating nature of PDN and the lack of effective treatment options for this patient population.Moreover, investigators observed neurological improvement in a remarkably high proportion of patients after 10 kHz SCS treatment.To our knowledge, no other SCS treatment for PDN has exhibited this potentially disease-modifying effect.For each cohort evaluated, HbA1c did not show a statistically significant change over the 24-month follow-up period, supporting that the study results were not due to changes in glycemic control.

10 kHz SCS versus traditional low-frequency SCS
Compared to previous RCTs that evaluated traditional paresthesia-based low-frequency SCS (LF-SCS) in similar PDN patients with a similar study design and control treatment [31,32], 10 kHz SCS treatment resulted in a greater reduction in pain at 6 months than LF-SCS (76% vs 41%-55%, respectively) and a higher responder rate for 50% pain relief (85% vs 53%-69%, respectively).By 24 months, the differences in pain reduction (80% vs 39%, respectively) and responder rate (90% vs 41%, respectively) were even more pronounced [33].Evidence from a 6-month indirect comparison of RCT stimulation arm outcomes in PDN supports this observation, with the analysis finding significantly higher pain relief and responder rate for 10 kHz SCS over LF-SCS [18].
In the same long-term LF-SCS RCT, several HRQoL and sleep parameters did not differ from CMM at 6 months and did not show significant improvement over baseline at 24 months [32,34].In comparison, patients who received 10 kHz SCS had significantly better HRQoL and sleep at 6 months than those treated with CMM, and these benefits lasted for the full 24 months.In particular, the increase of 0.146 in the mean EQ-5D-5L index value with 10 kHz SCS is highly clinically relevant, based on an estimated minimally important difference (MID) of 0.03 to 0.05 in patients with type 2 diabetes mellitus (ie, 2.9 to 4.9 times the MID) [46].The long-term improvement in sleep with 10 kHz SCS is also important because patients with chronic pain repeatedly indicate that sleep is a high-priority treatment outcome [47][48][49].

10 kHz SCS versus pharmacotherapy
The effectiveness of 10 kHz SCS compared to pharmacotherapy also appears very favorable.The NNT values for duloxetine, pregabalin, and gabapentin in PDN and neuropathic pain (including PDN pain) for the outcome of ≥50% pain relief versus placebo are reported to range from 5 to 16 at follow-up of 16 weeks or less [10][11][12][13].Tricyclic antidepressants, mainly amitriptyline, appear to be more effective (NNT, 3.6); however, amitriptyline is often poorly tolerated [13].
In contrast to pharmacotherapy, the SENZA-PDN study had very low NNT values for ≥50% pain relief versus CMM alone at 1, 3, and 6 months, with values of 1.4, 1.2, and 1.3, respectively.These NNT values indicate that PDN patients who are suitable candidates for 10 kHz SCS have a much higher likelihood of treatment success than with CMM alone.Of course, it should be kept in mind when reviewing the NNT values that the control arms in the SENZA-PDN study and the pharmacotherapy trials were not identical, even if the overall study populations, timeframe, and outcome measures were similar.

Safety
The incidence and type of procedure-related complications were comparable to those reported in the SCS literature for all patient populations [50].SAE rates were very low, and lead migration, IPG/lead revision surgery, and explant rates were at the lower end of the reported ranges for SCS [50].Considering that patients with diabetes can be more susceptible to infection, it was reassuring to find that the incidence of infection during the SENZA-PDN study (5.2%) was within the range for SCS in other indications (3.4%-10.0%;mean, 5.2%; excluding PDN studies) [50].Furthermore, the rate of explant due to infection in the current study was low, at 3.2%, and no devices were explanted due to lack of efficacy.Overall, the safety results support that patients with diabetes have no additional risk of complications with SCS.

Neurological/Sensory improvement
The neurological benefits observed after treatment with 10 kHz SCS are another noteworthy finding.Relative to their baseline examination, 66% of the patients had improved neurological status at the 24-month assessment, with improvement most often in the sensory domain.A recent small, open-label clinical study demonstrated statistically significant increases in intraepidermal nerve fiber density in the lower limb after 6 and 12 months of 10 kHz SCS [51], providing a potential mechanism for the observed improvements in sensory function.Interestingly, participants with sensory improvement at 24 months had higher preimplantation HbA1c compared to those with no change or reduced sensory function at 24 months (HbA1c, 7.7% vs. 7.2%; P =.021 via t-test), which suggests that patients with higher preimplantation HbA1c may be more likely to experience sensory improvement with 10 kHz SCS.These outcomes merit further exploration, given the increased risk of foot ulceration and lower limb amputation associated with diabetic neuropathy [6,7].Based on our review of the literature, we are unaware of other SCS therapies associated with this potentially disease-modifying effect of 10 kHz SCS [52].

Study limitations
This pragmatic study aimed to provide high-level evidence to aid clinical decision-making, albeit with some limitations.Comparing an implanted SCS device with CMM made it impossible to blind participants and study personnel, leading to a risk of biased outcomes and a possible placebo effect.However, the results through 24 months with 10 kHz SCS were consistent, indicating minimal placebo effect.
Another limitation is the potential impact of missed visits on study results.However, relatively few visits were missed despite conducting the study during the COVID-19 pandemic.Furthermore, we found minimal impact across study outcomes when we accounted for missing data using a multiple imputation method.For example, the mean reduction in pain was 79.9% among all implanted participants at 24 months using all available data and 79.7% using all available data plus imputed data.The corresponding responder rates were 90.1% and 89.5%, respectively.When interpreting the observed neurological improvements in this study, it is important to acknowledge certain limitations.Firstly, although the examination was developed as part of an RCT, it was not based on a validated scale.Additionally, the assessments of neurological status were conducted by trained clinicians who relied on their clinical judgment to determine changes.Despite standardized training, there is a possibility that inter-assessor variability may have influenced the test results.Furthermore, it should be noted that the investigators were not blinded to the treatment allocation when evaluating the neurological outcomes.Lastly, the assessment of neurological outcomes was performed relative to the study baseline examination for all patients in the study.Specifically, the 24-month postimplantation assessment in the CMM-to-10 kHz SCS+CMM crossover participants corresponded to 30 months after the study baseline exam, while the original 10 kHz SCS recipients were evaluated at 24 months after the study baseline exam.It is possible that the crossover cohort clinically deteriorated during the extra 6 months of CMM-only treatment prior to implantation, which may have contributed to their lower overall neurological and sensory improvements and the between-group difference in neurological outcome that achieved statistical significance at 24 months (P =.048, original 10 kHz SCS recipients vs CMM-to-SCS+CMM crossover group).While both groups still had meaningful neurological improvements, these outcomes suggest that delaying treatment for indicated patients could result in further neurological decline over time.

Conclusions
The results of this randomized controlled trial demonstrate robust, high-level evidence that 10 kHz SCS provides significant and durable pain relief in patients with refractory lower limb PDN pain.In addition, 10 kHz SCS treatment resulted in significantly improved sleep and HRQoL at 24 months, as well as neurological improvement in the majority of patients, providing a comprehensive solution for PDN management.The remarkable 24-month responder rate of 90% demonstrates that the therapy is highly effective and an important nonpharmacological therapy option.Furthermore, consistent pain relief for the original recipients of 10 kHz SCS and the implanted crossover participants (who received 10 kHz SCS after 6 months of CMM) supports that indicated patients should be treated without delay.

Fig. 3 .
Fig. 3. Patient-Reported Neuropathic Pain.The Douleur Neuropathique 4 Questions (DN4) scale is a validated neuropathic pain measure, with a score of ≥4 consistent with a clinical diagnosis of PDN.(A) Mean DN4 scores and (B) proportion of participants with DN4 <4 during the 6-month randomization phase.(C) Mean DN4 scores and (D) proportion of participants with DN4 <4 during the 24-month postimplantation phase.Error bars indicate 95% CI; *P <.001 vs baseline or preimplantation; # P <.001 vs CMM alone.