FormalPara Key Summary Points

Why carry out this study?

For spinal cord stimulation (SCS) therapies, reliability of long-term efficacy predictions has been criticized because of reliance on patient-reported trial outcomes and the inability to confirm delivery of a stable therapeutic dose to the spinal cord

Evoked compound action potential (ECAP) dose-controlled closed-loop (CL) SCS therapy, with objective confirmation of therapeutic neural activation and pulse-to-pulse adjustment of stimulation amplitude, may provide the ability to improve predictability of SCS trials

We investigated the feasibility of evaluating trial success at the shortest possible duration, postoperatively, immediately after lead insertion and programming with ECAP dose-controlled CL-SCS therapy

What was learned from the study?

This research shows that ECAP dose-controlled CL-SCS responses to therapy can be detected effectively at Day 0, with a high positive predictive value (98.4%) and low false-positive rate (5.6%) for determining trial success

This may allow physicians to confidently shorten the historically antiquated trial duration and provide the opportunity to redefine required SCS trial durations for systems employing ECAP dose-controlled CL-SCS

Shortened trials with ECAP dose-controlled CL-SCS may reduce trial complications and associated failures and support vulnerable patient populations (e.g., the anticoagulated or immunocompromised patient) while maintaining confidence in trial outcomes

Introduction

In many countries, patients are offered or required to experience a temporary screening trial implant to assess the treatment response of spinal cord stimulation (SCS) for chronic pain. SCS trials typically last 3–7 days [1], with success defined by subjective patient-reported feedback of at least 50% reduction in pain symptoms. This requirement was probably adopted after 1984 with the approval of one of the first fully implantable SCS systems [2]. In 2020, the Food and Drug Administration (FDA) also emphasized the importance of conducting a trial [3]. SCS trials are considered useful for evaluation of therapy response and allowing patients to experience the therapy, despite the debate surrounding the effectiveness of trials [1]. There has been, however, surprisingly little investigation into selection of a trial duration that provides the best balance between ensuring SCS success and patient safety [1]. For SCS systems characterized by delayed analgesia [4, 5] or lacking objective confirmation of stable neurophysiologic activation, a screening or trialing period spanning several days or longer is plausibly required. This prolonged trial duration exposes patients to risk of infections [6, 7], wearable trial equipment disconnects and failures, lead migrations [8], need to suspend normal medications (including anti-coagulants) [9], and increases associated patient burden, including work absenteeism [10].

Evoked compound action potential (ECAP)-controlled closed-loop (CL) SCS, using a physiologic closed loop control system (PCLCS[11], per IEC 60601-1-10), allows objective confirmation of therapeutic neural activation and adjusts stimulation amplitude on every stimulation pulse to maintain therapy at a target dose for typical activities of daily living [12] immediately post-programming. ECAP dose-controlled CL-SCS has been shown to provide superior pain relief than fixed-output, open-loop (OL)-SCS in a prospective, multicenter, randomized, self-selected cross-over, double-blind clinical trial with enduring outcomes out to 3 years [12,13,14,15,16]. Would an optimized and reproducible trial, with validated outcome assessments, and the use of objective ECAP dose-controlled CL-SCS offer an improved trial and prediction of therapy response, and when can those signals first be reliably detected?

Here, we investigate the feasibility of evaluating ECAP dose-controlled CL-SCS trial success at the shortest possible duration, postoperatively, immediately after lead insertion and programming. We hypothesize that trial success at this early stage of the trial (Day 0) is associated with success at the end of trial (EOT) for patients experiencing ECAP dose-controlled CL-SCS therapy. We were particularly interested in the positive predictive value (PPV) and false-positive rate (FPR) of the Day 0 evaluation. A high PPV and low FPR would provide confidence for the clinical validity of the evaluation and clearly have clinical relevance.

Methods

Participants with chronic, intractable trunk and/or limb pain were enrolled in the prospective, multicenter ECAP Study (NCT04319887, registered 24 March 2020 on ClinicalTrials.gov, first enrollment 29 October 2020) [17]. All subjects gave their informed consent for inclusion before they participated in the study. The ECAP Study was conducted in accordance with the United States FDA regulatory requirements, good clinical practice (GCP, per ISO 14155), Western Institutional Review Board approval (WIRB), and the Declaration of Helsinki guided ethical principles. This real-world study includes patients with a wide range of pain locations and indications, including chronic back and leg pain, upper limb pain, and conditions such as complex regional pain syndrome (CRPS). All patients underwent a temporary SCS trial with an external stimulator capable of ECAP-based PCLC (Evoke® System, Saluda Medical, Bloomington, MN) connected to one or two percutaneous Evoke leads (12 contacts, all capable of stimulation and ECAP measurement) placed in the dorsal epidural space of the spinal canal. Subjects with a successful temporary trial were eligible for a permanent Evoke System implant. For the observational analysis reported here, patients undergoing a traditional trial had evaluations performed on Day 0 postoperatively and at EOT (Fig. 1). These evaluations were prospectively collected for all possible patients starting from 29 October 2020 (N = 144).

Fig. 1
figure 1

Day 0 and EOT evaluations were performed for a subset of patients undergoing a traditional trial. EOT end of trial, HCP healthcare professional

Full size image

Neural activation, validated by ECAPs and used to set the stimulation parameters, and the ability to enable the PCLC therapy were confirmed, giving confidence that consistent therapeutic activation could be delivered throughout the trial, with real-time adjustment of stimulation amplitude on every stimulation pulse [12]. Data collection did not impact provision of therapy, and patients were not expecting to make decisions about success of the device until EOT. Since the aim of this investigation was to evaluate the feasibility of predicting trial responders on Day 0 for patients with objective confirmation of consistent therapy delivery, only patients for whom PCLC therapy could be established on Day 0 (134/144) were included. Although reasons for not establishing PCLC therapy on Day 0 for the 10/144 patients were not documented, ECAP signals and activation plots could be recorded in these patients. Additionally, patients with confirmed mid-trial lead migration (N = 2) were removed from analysis. Baseline characteristics, pain regions, and etiologies for the patients included in this analysis (N = 132) are shown in Table 1.

Table 1 Baseline characteristics, pain regions, and etiology
Full size table

Success criteria for this investigation were defined as: percentage pain relief (PPR) ≥ 50% (pain intensity test) and responded “Yes” to functional improvement (function test) and willingness to proceed to permanent implant (willingness test). PPR was chosen as the pain test due to its clinical applicability and high degree of correlation with calculated percent pain reduction [18,19,20]. The Day 0 function test was based on the validated Back Pain Functional Scale and Patient Specific Functional Scale assessments [21,22,23]. This involved patients completing a series of posture changes in the clinic mimicking activities of daily living before reporting whether any of the activities were made easier or less painful by their SCS therapy. Activities evaluated included sitting, standing, lying down, coughing, walking, and head forward/backward movements. Patients who passed all success criteria on Day 0 were classified as “Day 0 successes.” Patients failing any of the success criteria on Day 0 were classified as “needing longer to evaluate the therapy.”

PPR was collected on a continuous scale (0–100%) and converted to binary pass/fail based on the 50% pain relief cutoff (PPR ≥ 50% = pass, PPR < 50% = fail). The “yes” and “no” responses to the function test were similarly converted to pass/fail (“yes” = pass, “no” = fail). For the willingness test, patients could respond “yes,” “no,” or “unsure.” Given the need for trials to provide patients with enough experience of SCS to make an informed decision about progressing to a permanent implant [1], “unsure” and “no” responses were both treated as failures (“yes” = pass, “no”/“unsure” = fail). Trials were considered successful at the EOT timepoint if the physician progressed the patient to a permanent implant or if success criteria were met.

For comparison with the Day 0 data, PPR, functional improvement, and willingness test responses were also collected at EOT. We also report additional patient-reported outcomes (PROs) collected as part of the ECAP study [17] at EOT, including Patient-Reported Outcomes Measurement Information System (PROMIS), Patient Global Impression of Change (PGIC), and a holistic score reflecting the overall clinical benefit to patients, calculated as the total number of minimal clinically important differences (MCIDs) across five domains (see Supplementary Materials). Differences in PROs between Day 0 responder types were tested using a chi-square test (proportion difference) or t-test (mean difference). To validate the function test’s reliability in identifying function improvement, we compared function test results against additional EOT PROs (see Supplementary Materials). Appropriate licensure and/or approval for use was obtained for any PROs requiring permission to re-use.

Day 0 classification (successful/need longer to evaluate therapy) was compared to EOT evaluation status (responder/non-responder) for all patients in a 2 × 2 contingency table, and true-positive rate (sensitivity), true-negative rate (specificity), false-positive rate (FPR), and positive and negative predictive values (PPV and NPV, respectively) were calculated. The hypothesis that Day 0 evaluation status is associated with EOT status was tested with Barnard’s exact test.

Objective “dose ratio” and “dose accuracy” metrics were calculated to describe therapy dose levels and accuracy of the feedback loop on Day 0 and at EOT (see Supplementary Materials). Dose ratio is a standardized dose measure that may be compared across patients [24,25,26]. At EOT, additional data describing patients’ at-home usage of the device were available [25]. Therapy utilization represents the percentage of SCS usage expected to be therapeutic (see Supplementary Materials). Guided by previous findings [25], a one-sided Mann-Whitney U test was used to test the hypothesis that median dose ratio and dose accuracy were higher for Day 0 successes versus patients classified as needing longer to evaluate the therapy. At EOT, we tested the hypothesis that dose ratio, dose accuracy, and therapy utilization were all higher for EOT responders versus non-responders. Statistical analysis was conducted in Python v3.10 using the scipy and statsmodels packages. Statistical significance was judged at the 5% level.

Results

Day 0 and EOT evaluations were performed in 144 patients undergoing a traditional trial. ECAP dose-controlled CL therapy with pulse-to-pulse adjustment of stimulation was confirmed in 93% (134/144) of patients on Day 0. Of the 134 patients, two patients’ responses to the pain, function, and willingness tests changed from “yes” to “no” over the course of the trial. Trial lead migration was suspected to have occurred, which was confirmed by comparing x-ray or fluoroscopic images between the trial procedure and EOT. Since the results for these patients at EOT were confounded by lead migration, these patients were removed from analysis. One patient with CRPS had 40% PPR at EOT but experienced functional improvements and was interested to proceed to permanent implant. The trial was considered successful by the physician based on the PPR and functional improvement. This patient was imputed as a true positive.

Of the 132 patients included in this analysis (average trial length 6.4 ± 1.5 days, range 3–12 days), the overall trial success rate was 86.4% (114/132). The PPV was 98.4%, with 60/61 of the Day 0 successes continuing to respond at EOT. Of the patients classified as needing longer to evaluate therapy efficacy, 76.1% (54/71) were EOT responders (Table 2). The FPR of this evaluation was 5.6% (1/18) due to one patient with cervical CRPS who had a successful trial but opted not to proceed to a permanent implant. The true-positive rate (sensitivity) was 52.6% (60/114) while the true-negative rate (specificity) was 94.4% (17/18). The association between Day 0 status and EOT status was statistically significant (P = 0.001). Of note, 12.9% (17/132) of patients reported taking anticoagulant/antiplatelet medications.

Table 2 Contingency table comparing Day 0 status to EOT status
Full size table

Patients passing versus failing the function test at EOT had higher values for PROMIS-29 Physical Function (39.9 vs 35.9, P = 0.037; number of MCIDs improvement 1.52 vs − 0.01, P < 0.001), PROMIS GH Physical Health scores (43.2 vs 35.5, P < 0.001; number of MCIDs improvement 2.05 vs 0.29, P  < 0.001), and the holistic score (9.5 vs 1.1 MCIDs improvement, P < 0.001) and lower values for PROMIS-29 Pain Interference score (56.3 vs 67.3, P < 0.001; number of MCIDs improvement 2.54 vs 0.34, P < 0.001).

Our success criteria (pain, function, and willingness tests) were more conservative than criteria included in typical clinical practice to ensure that the Day 0 evaluation was robust against false positives. On an individual test basis, 80.3% (106/132) and 81.8% (108/132) of patients passed the pain and function tests on Day 0, respectively (not mutually exclusive). The patients who passed Day 0 evaluation had average pain relief of 79.2% ± 17.1% and 81.8 ± 17.0% on Day 0 and EOT, respectively (Pearson correlation coefficient for Day 0 versus EOT PPR: r = 0.66, P < 0.001). At EOT, the Day 0 responders also had high satisfaction rates (100% of patients satisfied with pain relief and therapy), high improvement on PGIC (98.4% of patients improved), and a high holistic response score (10.4 MCIDs improvement). Outcomes for Day 0 success patients versus patients who needed longer to achieve success are displayed in Table 3. At EOT, slightly higher mean PPR was observed for patients who were successful on Day 0 versus those who needed longer to respond (EOT mean PPR: 81.8% vs 75.0%, P = 0.036), but there were no statistically significant differences in responder status, satisfaction rates, or holistic score (all P > 0.05).

Table 3 Pain relief and responder status shown at Day0 and EOT for patients who were successful on the Day 0 evaluation vs those who were successful later. Additional outcomes like satisfaction, PGIC, and holistic response, collected at EOT only, are also displayed
Full size table

Most patients responded “yes” to the willingness test on Day 0 (56.1%) and only a small percentage responded “no” (3.0%), while a substantial percentage responded “not sure” (40.9%). Since most patients classified as needing longer to evaluate the therapy responded “not sure” to the Day 0 willingness test (76.1%), we investigated the possibility of using only the pain and function criteria to assess Day 0 success. This gave a similar PPV to the original analysis (94.7%, 89/94), increased the true positive rate (75.4%, 89/118), but increased the risk of false positives (35.7%, 5/14).

Objective ECAP data analyzed in this investigation are described in the Supplementary Material. Median dose ratio at Day 0 was 1.39 (IQR 1.26–1.53) and 1.42 (IQR 1.23–1.54) for Day 0 successes compared to patients classified as needing longer to evaluate the therapy, respectively (P = 0.374). Median dose accuracy at Day 0 was 2.08 µV (IQR 1.78–2.74 µV) and 2.17 µV (IQR 1.72–2.93 µV) for Day 0 successes compared to patients classified as needing longer, respectively (P = 0.348). At EOT, out-of-clinic dose ratio and therapy utilization, but not dose accuracy, were significantly different between EOT responders and non-responders. Median out-of-clinic dose ratio at EOT was 1.39 (IQR 1.21–1.56) and 1.22 (IQR 1.06–1.41) for EOT responders versus non-responders, respectively (P = 0.037), while median therapy utilization was 98.9% (IQR 81.1–100%) and 94.2% (IQR 52.5–98.4%), respectively (P = 0.026). Median out-of-clinic dose accuracy at EOT was 5.68 (IQR 3.09–8.97) µV for responders and 6.58 (IQR 4.45–8.61) µV for non-responders (P = 0.346).

Discussion

The high positive predictive value (98.4%) and low false-positive rate (5.6%) of the Day 0 evaluation presented here suggest that ECAP dose-controlled CL-SCS, with real-time adjustment of stimulation amplitude on every stimulation pulse, may be utilized for early prediction of SCS trial responders. The single false positive was a CRPS patient who did not want to proceed to permanent implant at EOT despite ≥ 50% pain relief and functional improvement, PGIC improvement, and satisfaction with therapy and pain relief. Physicians and patients can therefore be confident that when a patient has a successful Day 0 evaluation, this is a true success and not due to a placebo effect such that the patient may fail trial evaluation criteria given more time. Moreover, as PPV may be influenced by the overall trial success rate, the specificity of 94.4% further confirms the internal validity of the Day 0 evaluation for predicting trial success.

These results suggest that same day trials, where leads are pulled after successful Day 0 evaluation, may be of utility when spinal cord activation is confirmed using ECAPs, and stable activation is maintained via ECAP dose-controlled CL-SCS therapy such that the in-clinic experience of therapy is representative of how therapy will feel across out-of-clinic activities of daily living [12, 13]. In contrast, shortened fixed-output SCS trials, regardless of waveform, would create risk of false-positive trial outcomes given the inability to confirm neural activation and the dynamic nature of the epidural space, which means that common posture changes and physiologic activity (e.g., respiration and heartbeat) can lead to under- and over-stimulation. Therefore, it makes sense for patients to obtain out-of-clinic experience with fixed-output SCS trial systems to confirm the stimulation levels and therapeutic efficacy that they can maintain when out of the clinic [12, 13].

Same day trials may reduce risks of postoperative infections, lead migrations, complications resulting from stopping antithrombotic medications, equipment disconnections, and other myriad issues associated with traditional trials [6,7,8,9,10]. As occurrence of these risks may undermine therapy efficacy and patient confidence in the therapy, same day trials may also “rescue” trial patients from false-negative trial outcomes that are inherently encountered when evaluating success at traditional 3–7 day durations. The improved efficiency of same day trials may be particularly important for vulnerable patient populations, including the anticoagulated patient. In this study, a substantial number of patients (12.9%) were on anticoagulants similar to rates reported (~ 20%) in other studies [9]. Although clear guidance exists [27, 28], a same day trial may further reduce the risk of ischemic events due to the suspension of anticoagulants.

In this dataset the true-positive rate was approximately 53%, which means that approximately half the patients assessed were identified correctly as trial successes on Day 0, while the other half of the trial successes needed longer to pass all success criteria. This may be related to a few variables, including expectations of the traditional trial (wanting to test drive the device at home over routine daily activities), further compounded by not expecting questions about EOT decisions on the day of the procedure. A substantial percentage of patients classified as “needing longer to evaluate the therapy” reported pain relief and functional improvements on Day 0 but did not respond “yes” to the willingness test (46%). While patient confidence is an important inclusion in any decision to proceed to a permanent implant [1] and may guard against false-positive responses, the drop-off here due to the willingness test is substantially higher than the minority of patients who reported preference for a two-stage procedure allowing “test-driving” an SCS system prior to permanent implant [29]. We hypothesize that a higher percentage of patients would pass the willingness test, and thus Day 0 evaluation, when patients have prior awareness that decisions to proceed to a permanent implant could be made immediately after trial system implant and programming.

Dose level and accuracy of the loop was the same across all patients on Day 0, regardless of their responder status. This highlights that all patients were programmed effectively with ECAP dose-controlled CL-SCS, with consistent pulse-to-pulse activation of the dorsal columns [12, 13]. The dose ratio and dose accuracy levels on Day 0 align with values providing maximum analgesic benefit to patients (dose ratio 1.2–1.6; dose accuracy < 10 µV) [25]. Median dose ratio for all patients on Day 0 was in the middle of this range at 1.4. At EOT, however, responders maintained their dose ratio at 1.4 while non-responders used a significantly lower dose ratio of 1.2, typical of the lower end of the optimal range. While research into evidence-based guidelines for neural dosing prescription is in early stages, these results extend relationships between dose ratio and pain relief previously reported in the permanent implant phase, highlighting the importance of maintaining optimum activation levels, for example via patient education and reprogramming after lead migration, to improve pain relief and associated trial success rates [25, 26].

This study has several limitations. First, we report observational data collected with patients undergoing a traditional trial of variable length. A prospective study is needed to compare long-term outcomes for traditional trials versus an abbreviated trial where patients have their leads pulled on Day 0 after passing success criteria. Second, lead migrations were only confirmed for patients who passed Day 0 evaluations but failed at EOT. Additional lead migrations may have been undetected if patients maintained adequate therapy or if device re-programming mitigated impacts of the migration. These scenarios are not of clear clinical significance [30] and were not further analyzed here. Third, in this investigation data were not available to control for impact of trial procedure anesthetics and/or analgesics on postoperative pain scores. If these significantly impacted the Day 0 evaluation, however, we would expect the FPR to be much higher and the PPV to be lower than observed [31].

We included all prospective patients in our analysis to compare pain relief for patients who responded on Day 0 compared to those requiring a longer period to determine their response. This is in contrast to other studies which only include therapy responders [5]. Interestingly, patients with successful Day 0 evaluations maintained their PPR at EOT (mean PPR 79.2% vs 81.8%) while the responders who needed longer to evaluate the therapy trended upwards in pain relief from Day 0 to EOT (mean PPR 54.9% vs 75.0%). Along with significant pain relief, Day 0 responders had high rates of satisfaction (100% of patients satisfied) and improved quality of life (98.4% of patients improved on PGIC) at EOT. Responder and satisfaction rates and mean holistic scores were not different at EOT between patients who responded on Day 0 compared to those who needed longer to respond. This indicates that Day 0 successes are not different from other SCS responders in terms of outcomes achieved with ECAP dose-controlled CL-SCS but are different in terms of the speed with which a response is achieved. These rapid responders may possess unique characteristics that make them suited to receive rapid relief with SCS and assess their relief quickly after implant, as early as Day 0. Such characteristics should be further investigated to predict rapid responders to SCS therapy.

Conclusion

This study demonstrates that reliable same-day identification of trial responders is enabled by the use of ECAP dose-controlled CL-SCS with objective neurophysiologic confirmation of treatment responses and maintenance of consistent therapy with pulse-to-pulse stimulation amplitude adjustments. Using pain reduction and functional assessments modified from validated PROs, this approach achieves a high PPV (98.4%) and low FPR (5.6%) compared to traditional trial outcomes. Implications from this finding may improve trial efficiency and reduce intra-trial complications while expanding SCS therapy access and decreasing risk for vulnerable patient populations. Further investigation is needed to assess long-term outcomes and true diagnostic efficacy of this technique. This research may have implications for SCS trial predictive success as well as on trial duration requirements.