Authors: Chenlong Liao, Shuo Li, Wenxiang Zhong, Yayuan Tian, Wenchuan Zhang
Categories: Original Research, Painful diabetic peripheral neuropathy, Nerve decompression, Neuropathic pain, Pain interference, Medication burden
Source: Neurology and Therapy
Authors: Chenlong Liao, Shuo Li, Wenxiang Zhong, Yayuan Tian, Wenchuan Zhang
Lower extremity nerve decompression (LEND) for painful diabetic peripheral neuropathy (PDPN) remains controversial, and evidence regarding its long-term effectiveness in real-world clinical practice is limited.
This retrospective real-world cohort study included patients with PDPN treated with LEND or medical therapy alone between 2008 and 2011. Ultra-long-term outcomes were assessed after > 10 years of follow-up. Pain intensity was evaluated using the visual analogue scale (VAS). Composite pain burden and functional impact were assessed with the Brief Pain Inventory for Diabetic Peripheral Neuropathy (BPI-DPN). Psychological symptoms were measured using the Hospital Anxiety and Depression Scale (HADS), and analgesic medication burden was quantified by the Medication Quantification Scale III (MQS-III). Exploratory prognostic factor and subgroup analyses based on pain distribution were performed.
Seventy-six patients in the LEND group and 31 patients in the medical group were available for ultra-long-term analysis. Compared with medical management, LEND was associated with greater long-term pain relief (mean VAS change − 5.63 ± 2.16 vs − 1.03 ± 1.92; p < 0.001) and higher responder rates (≥ 50% pain 65.8% vs 9.7%; p < 0.001). Significant long-term improvements were also observed in BPI-DPN pain severity and pain interference (both p < 0.001), anxiety and depression symptoms (both p < 0.001), and medication burden (MQS-III p < 0.001). Within the LEND cohort, younger age at surgery and lower body mass index were independently associated with greater long-term pain improvement. Both focal and diffuse pain subgroups demonstrated significant improvements in pain and functional outcomes after surgery, with no meaningful differences at ultra-long-term follow-up. Diabetic foot ulcer events occurred less frequently after LEND (0% vs 32.3%; p < 0.001).
LEND demonstrated long-term efficacy in alleviating pain and concurrently improving the pain-related interference and psychological status of patients with PDPN. A Graphical Abstract is available for this article.
This study was retrospectively registrated in Chinese Clinical Trial Registry chictr.org. cn (ChiCTR2500099348), https://www.chictr.org.cn/bin/project/edit?pid=266042.
The online version contains supplementary material available at 10.1007/s40120-026-00913-3.
***Why carry out this study?Painful diabetic peripheral neuropathy causes long-term neuropathic pain, disability, emotional distress, and heavy reliance on pain medications, with limited durable treatment optionsThis study asked whether nerve decompression surgery provides better ultra-long-term outcomes than medical treatment alone for people with painful diabetic peripheral neuropathyWhat was learned from the study?***Over > 10 years of follow-up, about 66% of surgical patients achieved at least 50% pain reduction compared with about 10% of patients treated medicallySurgery was also linked to long-term improvements in daily functioning, emotional symptoms, and reduced use of pain medicationsYounger age and lower body mass index were associated with better long-term pain improvement, and patients with different pain patterns could benefit from nerve decompression surgery
This article is published with digital features, including a graphical abstract to facilitate understanding of the article. To view digital features for this article, go to 10.6084/m9.figshare.31378330.
As is well documented, diabetic peripheral neuropathy (DPN) is one of the most prevalent complications of diabetes, and the reported prevalence varies because of differences in study designs, sampling populations, and diagnostic criteria [1]. The overall risk of DPN increases with age, and the lifetime prevalence has been reported to be > 50% [2]. Notably, approximately 15–25% of patients with diabetes experience neuropathic pain or develop painful DPN (PDPN) [3]. According to data from the International Diabetes Federation, many patients with DPN suffer from neuropathic pain and related psychological disorders such as anxiety and depression [4]. Despite advances in pharmacological and non-pharmacological treatment modalities, effective treatment options for PDPN remain limited [5, 6].
Surgical decompression of lower extremity nerves has been proposed as a promising therapeutic intervention for relieving diabetic neuropathic pain in previous studies [7–9]. The proposed rationale for lower extremity nerve decompression is based on the concept that painful diabetic peripheral neuropathy may result from the combined effects of metabolic nerve impairment and superimposed mechanical compression [10, 11]. The underlying metabolic abnormalities present in diabetes render the peripheral nerves susceptible to external compression by inducing nerve swelling, impaired axonal transport, and microvascular dysfunction [12–14]. In anatomically constrained fibro-osseous tunnels, such vulnerability may lead to focal nerve entrapment, reduced intraneural blood flow, and ongoing nociceptive signaling [15]. Surgical decompression aims to relieve this secondary mechanical stress, potentially improving neural perfusion and reducing ectopic pain signaling. Although decompression does not address the underlying metabolic pathology of diabetes, it may modify pain-related symptoms in selected patients by alleviating reversible compressive components of nerve dysfunction. Encouraging outcomes have also been achieved in several randomized controlled trials [16–18]. However, the follow-up durations of these retrospective and prospective studies primarily ranged from 2 to 3 years, with the longest at 52 months [16]. Long-term follow-up data on the sustained effectiveness and safety of surgical treatment for DPN, a chronic complication of a chronic metabolic disease, are urgently needed. Understanding these outcomes is essential for patient counselling and long-term management. To provide long-term outcomes of lower extremity nerve decompression (LEND) for PDPN, the medical records of patients treated by our surgical team from January 2008 to December 2011 were retrospectively reviewed, and pain symptoms and psychological status were reevaluated using our prior cohort [19]. Rather than focusing on surgical technique, this study examines ultra-long-term pain-related outcomes, functional interference, psychological symptoms, and medication burden from a pain medicine perspective. By offering valuable insights into the durability of LEND effects, this study intends to inform clinical decision-making and contribute to the evolving management strategies for DPN.
Accordingly, the primary aim of this study was to evaluate the ultra–long-term real-world outcomes of lower extremity nerve decompression compared with continued medical management in patients with painful diabetic peripheral neuropathy. Specifically, we sought to assess long-term changes in pain intensity, composite pain burden, pain-related functional interference, psychological symptoms, and analgesic medication use over > 10 years of follow-up. Secondary aims included identifying clinical factors associated with long-term pain improvement after surgery and exploring whether baseline pain distribution patterns were associated with differential long-term outcomes. Through this evaluation, we aimed to provide clinically relevant evidence on long-term outcomes that may support patient counselling and individualized management strategies in routine neurological practice.
This study was approved by the Ethics Committee of Xinhua Hospital, Shanghai JiaoTong University School of Medicine (XHEC-D-2025-014). This study was performed in accordance with the Helsinki Declaration. Owing to the retrospective design, the requirement for written informed consent was waived.
This is a retrospective long-term follow-up study of a previous cohort treated at our institute from January 2008 to December 2011. All patients were treated with either LEND (LEND group) or conventional analgesic medications alone (medical group). Patients with painful diabetic lower-extremity neuropathy reported in our previous study were followed up to investigate long-term outcomes [19] (Fig. 1).Fig. 1The Flow diagram of patient selection and long-term follow-up. LEND lower extremity nerve decompression, PDPN painful diabetic peripheral neuropathy
Each patient in the LEND group underwent one-stage bilateral triple-nerve decompression procedures performed by the senior surgeon (Wenchuan Zhang) using microscope assistance under continuous epidural anesthesia. Neurolysis procedures were performed on each leg, including the common peroneal nerve at the fibular neck, the deep peroneal nerve under the extensor hallucis brevis tendon, and the posterior tibial nerve with its medial and lateral plantar branches at the medial ankle [19, 20]. The medical group largely consisted of outpatients with PDPN who were treated with conventional medication [19, 21]. As nerve decompression is the only difference between LEND and medication groups, conventional analgesic medications, such as calcium channel a2δ ligands (Pregabalin or Gabapentin), serotonin and noradrenaline reuptake inhibitors (Duloxetine), and tricyclic antidepressants (amitriptyline), were prescribed in both the surgical and medical groups as needed before or after enrollment. At the time of enrollment, patients in both groups had a history of analgesic medication treatment. Failure of medication treatment (mostly waned efficacy or side effects) prompted the patients to seek for further therapy. Analgesic drug treatment could be continued in patients in the LEND group as complementary therapy.
The inclusion and exclusion criteria, as well as the clinical evaluations of the visual analogue scale (VAS) and Brief Pain Inventory Short Form for diabetic peripheral neuropathy (BPI-DPN), were described in our prior study [19]. The BPI-DPN was administered by instructing patients to rate pain intensity and the pain-related interference due to diabetes over the past 24 h, following standard instructions [22]. Psychological status, including anxiety and depression, was assessed using the Hospital Anxiety and Depression Scales (HADS), which has been previously applied to patients with PDPN in earlier clinical reports [11]. The Medication Quantification Scale Version III (MQS-III), which co-quantifies three relevant aspects of medications prescribed for chronic nonmalignant pain, namely drug class, dosage, and detriment (risk), was adopted to evaluate medication use and quantify pain [23]. Patients or caregivers were contacted via telephone or outpatient visits to acquire follow-up data. As described in our prior study, patients in the LEND group were divided into focal and diffuse pain subgroups according to their initial pain distribution by two physicians; a senior physician was consulted for the final decision if no agreement on the type of pain distribution was achieved [19] (Fig. 2).Fig. 2The schematic diagram of focal and diffuse pain. Focal pain is defined as pain mainly confined to one to three scattered areas of the legs, dorsum of the feet, heels, toes, or plantar aspect of the feet, while the diffuse pain is defined as pain that is so dispersed along the affected extremities that the exact position cannot be localized
The primary outcome was the long-term change in pain intensity, assessed by the VAS, defined as the difference between baseline and the most recent long-term follow-up. A clinically meaningful pain response was defined a priori as a ≥ 50% reduction in VAS from baseline, and responder rates were compared between groups using chi-square tests. Secondary outcomes included 24-h average pain severity and pain-related functional interference assessed by the BPI-DPN, psychological outcomes assessed using the HADS, analgesic medication burden assessed using the MQS-III, and occurrence of diabetic foot ulcer during long-term follow-up. Exploratory outcomes included prognostic factor analyses for long-term surgical benefit and subgroup analyses based on baseline pain distribution (focal versus diffuse).
All statistical analyses were performed using SPSS version 24.0 (IBM Corp., Armonk, NY, USA) and Prism version 7.0 (GraphPad software, Inc, Chicago, IL, USA). A two-sided p value < 0.05 was considered statistically significant unless otherwise specified. Continuous variables were summarized as mean ± standard deviation (SD) or median (interquartile range, IQR), as appropriate, based on distribution assessed by the Shapiro-Wilk test. Categorical variables were presented as frequencies and percentages.
A two-sample t-test for normally distributed variables or the Mann-Whitney U test for non-normally distributed variables was used to compare baseline demographic and clinical characteristics between different groups or subgroups. Within-group or within-subgroup changes were assessed using paired t-test or Wilcoxon signed rank test. The cumulative incidence of diabetic foot ulcer was analyzed using Kaplan-Meier survival analysis, and differences between groups were compared using the log-rank test. Statistical analysis of categorical data was conducted using the chi-square test. No formal matching or randomization was applied because of the retrospective real-world design. Changes in variables from baseline to the last follow-up were defined as long-term outcomes, while changes in variables from the preoperative assessment to the 2-year follow-up were defined as short-term outcomes. Changes in variables during the later-stage follow-up, defined as the period between the 2-year follow-up and the last follow-up, were also recorded. Since comparisons of short-term outcomes were addressed in a previous report [19], this study compared long-term outcomes and changes in variables during the later-stage follow-up between different groups or subgroups using analysis of covariance, where appropriate.
To assess the robustness of the result, analysis of covariance (ANCOVA) models were constructed with long-term outcome change as the dependent variable, treatment group as the main independent variable, and baseline score as a covariate. Additional models further adjusted for age, sex, body mass index (BMI), diabetes duration, pain duration, and baseline glycaemic control.
Given the long-term follow-up duration, a substantial proportion of patients had died before outcome assessment. Because pain, functional, and psychological outcomes are undefined after death, the primary analyses were conducted among patients who were alive and completed long-term follow-up (complete-case analysis). To assess the potential impact of missing mortality-related data, we performed a conservative worst-case sensitivity analysis, in which deceased patients in both the LEND and medical groups were imputed to have no pain improvement. The primary between-group comparison of VAS change was then repeated under this assumption. In addition, longitudinal models adjusted for baseline pain severity and clinical covariates were examined to assess the robustness of the findings. Detailed results of these sensitivity analyses are provided in the Supplementary Materials.
Within the LEND group, exploratory analyses were conducted to identify baseline factors associated with long-term surgical benefit. Candidate variables were selected a priori based on clinical relevance and included age at surgery, sex, body mass index, duration of diabetes, baseline VAS score, BPI-DPN interference score, and HADS-Anxiety score. Univariable linear regression was first performed with long-term outcome change as the dependent variable. Variables with p < 0.10 in univariable analysis were entered into multivariable linear regression models. To minimize the risk of overfitting given the sample size, the final multivariable models were restricted to a limited number of clinically relevant covariates. Results are presented as regression coefficients (β) with 95% confidence intervals (CI). These analyses were considered hypothesis-generating.
An exploratory subgroup analysis was performed according to baseline pain distribution (focal versus diffuse). Long-term changes in VAS and BPI-DPN scores were compared descriptively within each subgroup. Given the limited sample size, no formal interaction testing was performed, and these analyses were interpreted as exploratory.
Of the 281 patients initially undergoing LEND, 76 (27.0%) were alive and available for evaluation at the current long-term follow-up, while 150 (53.4%) were confirmed deceased, and 55 (19.6%) were lost to follow-up. Of the 86 patients treated with medications in the medical group, 31 (36%) were available for follow-up, while 36 (41.9) patients were confirmed deceased, and 19 (22.1%) lost to follow-up (Fig. 1). The primary reason for loss to follow-up was patient mortality, consistent with the advanced age and comorbid disease burden characteristic of the PDPN population.
Baseline demographic and clinical characteristics were broadly comparable between the surgical and medical groups (Table 1). With the exception of sex distribution (p = 0.048), there were no clinically relavant between-group difference in age, sex, diabetes duration, pain duration, glycaemic control, or baseline VAS (p > 0.05). The median duration from treatment to the current follow-up was > 13 years and did not differ significantly between groups (p = 0.071). Table 1Baseline demographic and clinical characteristics of patients in the surgical and control groupsCharacteristicLEND group (n = 76)Control group (n = 31)pAge at the time of enrollment (years) Mean (SD)51.55 (8.76)49.35 (9.96)0.308 Median53.0046.00Sex, n (%)0.048* Male26 (34.2%)17 (54.8%) Female50 (65.8%)14 (45.2%)Baseline BMI (kg/m^2^), mean (SD)24.60 (2.98)23.40 (3.52)0.178Duration of diabetes at the time of enrollment (years) Mean (SD)3.14 (2.80)4.65 (4.43)0.239 Median2.003.00Duration of pain at the time of enrollment (months) Mean (SD)19.18 (16.37)20.16 (17.87)0.870 Median12.0012.00HbA1c (mmol/l), mean (SD)7.73 (1.08)8.29 (0.99)0.680Follow-up (months) Mean (SD)171.23 (10.90)167.52 (9.71)0.071 Median176.00168.00BMI body mass index, HADS-A anxiety subscale of the Hospital Anxiety and Depression Scales, HADS-D depression subscale of the Hospital Anxiety and Depression Scales, HbA1c glycosylated haemoglobin, SD standard deviation*Statistically significant results (p < 0.05)
In the medical group, pain intensity remained high at long-term follow-up, with only a modest reduction in VAS from baseline (7.74 ± 0.99, mean −0.77 ± 1.87, p = 0.006). In contrast, patients in the LEND group experienced a marked and sustained decrease in VAS scores over time, with a mean long-term reduction of − 5.63 ± 2.16 points from baseline, and VAS scores remained unchanged during the later-stage follow-up (p = 0.608) (Fig. 3A). By comparing the long-term changes in VAS between the two groups, we found that patients in the LEND group experienced more pain relief than those in the medical group (p < 0.001) (Table 2; Fig. 3A). The proportion of patients achieving at least a 50% reduction in pain was substantially higher in the LEND group (n = 50) than in the medical group (n = 3) at last follow-up (p < 0.001).Fig. 3Clinical evaluation of VAS, BPI-DPN, HADS, and MQS-III at different time points in the LEND and medical groups. Patients in the LEND group demonstrated sustained improvement across all domains, whereas minimal long-term change was observed in the medical treatment group. Between-group differences were statistically significant for all measures at long-term follow-up (all p < 0.001). Data shown represent mean ± standard error. Refers to comparisons between the surgical and medical groups at the same time points; # refers to comparisons between the 2-year follow-up and the last follow-up within the same group. ^#^p < 0.05, **^##^p < 0.01, **^###^p < 0.001. BPI-DPN Brief Pain Inventory Short Form for diabetic peripheral neuropathy, LEND lower extremity nerve decompression, HADS-A anxiety subscale of the Hospital Anxiety and Depression Scales, HADS-D depression subscale of the Hospital Anxiety and Depression ScalesTable 2Measurements changes from the last follow-up by groupTime points (years)LEND groupControl groupScoreMean (SD)Change from baselineMean (SD)pScoreChange from baselineMean (SD)pVASBaseline7.95 (1.23)NANA7.74 (0.99)NANA2-PS2.18 (2.60)−5.76 (2.75)< 0.0017.48 (1.29)−0.26 (1.37)0.301Last FU2.32 (1.82)−5.63 (2.16)< 0.0016.71 (1.57)−1.03 (1.92)0.006BPI-DPN mean Pain Severity IndexBaseline5.98 (1.24)NANA6.32 (0.92)NANA2-PS3.35 (1.74)−2.63 (1.86)< 0.0016.55 (1.20)0.23 (0.72)0.084Last FU2.00 (1.64)−3.98 (1.96)< 0.0016.24 (1.32)−0.08 (1.26)0.723BPI-DPN mean Pain Interference IndexBaseline4.83 (0.85)NANA4.61 (0.91)NANA2-PS2.45 (1.45)−2.38 (1.55)< 0.0014.89 (1.12)0.28 (0.78)0.057Last FU1.95 (1.54)−2.88 (1.73)< 0.0015.22 (2.25)0.61 (2.20)0.134HADS-ABaseline9.59 (3.96)NANA9.19 (3.37)NANA2-PS6.29 (2.78)−3.30 (3.40)< 0.0019.48 (4.19)0.29 (5.28)0.762Last FU4.67 (2.07)−4.92 (3.86)< 0.0019.32 (2.81)0.13 (4.60)0.877HADS-DBaseline7.86 (2.77)NANA7.03 (2.07)NANA2-PS5.84 (2.19)−2.01 (2.62)< 0.0016.74 (1.79)−0.29 (1.72)0.354Last FU5.43 (2.16)−2.42 (3.69)< 0.0016.87 (2.75)−0.16 (2.73)0.745MQS-IIIBaseline3.46 (2.49)NANA2.69 (2.66)NANALast FU1.76 (2.62)−1.71 (3.29)< 0.0014.84 (3.41)2.15 (3.35)0.001**NA not available, 2-PS 2 years post-surgery, BMI body mass index, BPI-DPN Brief Pain Inventory Short Form for diabetic peripheral neuropathy, FU follow-up, HADS-A anxiety subscale of the Hospital Anxiety and Depression Scales, HADS-D depression subscale of the Hospital Anxiety and Depression Scales, HbA1c glycosylated haemoglobin, MQS-III Medication Quantification Scale Version III, SD standard deviation, VAS visual analogue scale*Statistically significant results (p < 0.05)
Consistent with the VAS results, significant long-term improvements in pain-related functional interference, as assessed by the BPI-DPN, were observed in the LEND group, (Table 2; Fig. 3B–C). Both BPI pain severity and pain interference scores decreased steadily after surgery and remained significantly lower at long-term follow-up. In contrast, only minimal changes were observed in the medical group. Between-group differences at the last follow-up were statistically significant for both BPI dimensions (both p < 0.001).
At baseline, patients in both groups exhibited substantial psychological burden, with elevated HADS-Anxiety and HADS-Depression scores. At long-term follow-up, HADS-Anxiety scores were markedly reduced in the LEND group (baseline: 9.59 ± 3.96, the last follow-up: 4.67 ± 2.07, mean −4.92 ± 3.87, p < 0.001), whereas little change was observed in the medical group (Table 2; Fig. 3D). A similar, though less pronounced, pattern was noted for HADS-Depression (baseline: 7.86 ± 2.77, the last were 5.43 ± 2.16, mean −2.40 ± 3.70, p < 0.001) (Table 2; Fig. 3E). The between-group differences in psychological improvement remained statistically significant after long-term follow-up (HADS-A, p < 0.001; HADS-D, p < 0.01).
Analgesic medication burden, as quantified by MQS-III, was similar between the LEND and medical groups at baseline. Over the long term, MQS-III scores decreased substantially in the LEND group, reflecting reduced dependence on analgesic medications (p < 0.001). In contrast, the medical group showed an opposite trend in medication burden (Fig. 2F; Table 2). The between-group difference in MQS-III change was statistically significant (p < 0.001).
In analyses adjusted for baseline pain scores and key clinical covariates, the long-term between-group differences in VAS, BPI-DPN, HADS, and MQS-III remained significant (all adjusted p < 0.01; Supplementary Table S1).
During long-term follow-up, no major surgery-related complications were recorded. Minor complications included occasional local scar discomfort, which did not require surgical intervention. Notably, the cumulative incidence of diabetic foot ulceration was lower in the LEND group than in the medical group (0% vs 32.26%; p = < 0.001; Fig. 4).Fig. 4Kaplan-Meier analysis of diabetic foot ulcer occurrence during long-term follow-up. Patients who underwent LEND had a significantly lower cumulative incidence of diabetic foot ulcer compared with medically treated patients (log-rank p < 0.001). Censoring marks represent patients without ulcer events at the time of last follow-up
Exploratory prognostic analyses were performed within the LEND group to identify clinical factors associated with long-term pain relief. In multivariable linear regression models adjusted for baseline pain severity and key covariates, younger age at surgery (β = − 0.073, p = 0.0019) and lower latest BMI (β = −0.159, p = 0.009) were independently associated with greater long-term improvement in VAS scores (Fig. 5-A; Supplementary Table S2).Fig. 5Prognostic factors for long-term surgical benefit. A Prognostic factor for long-term VAS changes in the LEND group. Younger age at surgery (β = −0.073, p = 0.0019) and lower latest BMI (β = −0.159, p = 0.009) were independently associated with greater long-term pain improvement. B Prognostic factor for long-term pain interference in the LEND group. Shorter duration of diabetes (p = 0.025) and higher baseline anxiety scores (p = 0.008) were associated with greater improvement in pain interference. Data represent β coefficients with 95% confidence intervals. BMI body mass index, HADS-A anxiety subscale of the Hospital Anxiety and Depression Scales, HADS-D depression subscale of the Hospital Anxiety and Depression Scales, VAS visual analogue scale
Shorter duration of diabetes (β = −0.160, p = 0.025) and higher baseline anxiety scores (β = −0.009, p = 0.008) were weakly associated with greater improvement in pain interference (Fig. 5B; Supplementary Table S3). None of the baseline covariates was identified to be significantly associated with superior long-term outcome of psychological status (Supplementary Table S4).
In accordance with our prior study, patients in the LEND group were also divided into focal (n = 36) and diffuse (n = 40) pain subgroups according to their initial pain distribution (Fig. 2), with baseline demographics and clinical characteristics summarized in Table 3, showing that no significant difference was noted. Both subgroups experienced significant long-term reductions in VAS and BPI-DPN scores after LEND (Table 4, Supplementary Figure S1). No meaningful differences in long-term pain relief, pain interference, and pain-related psychological changes were observed between patients with focal versus diffuse pain patterns (Table 4). Table 3Baseline demographics and surgical outcomes in the surgical subgroupsCharacteristicFocal pain group (n = 36)Diffuse pain group (n = 40)pAge (years) Mean (SD)50.89 (8.46)52.15 (8.94)0.534 Median5354.5Sex, no. (%)0.878 Male12 (33.3%)14 (35.0%) Female24 (66.7%)26 (65.0%)Preoperative BMI (kg/m^2^), mean (SD)25.21 (2.30)24.05 (3.42)0.089Duration of diabetes at the time of enrollment (years) Mean (SD)2.40 (1.80)3.81 (3.35)0. 270 Median1.753Duration of pain at the time of enrollment (months), mean (SD)0.439 Mean (SD)20.22 (18.92)18.25 (13.85) Median12.012.0HbA1c (mmol/l), mean (SD)7.57 (1.02)7.87 (1.12)0.223Follow-up (months), mean (SD)174.00 (6.77)167.45 (7.39)0.311BMI body mass index, BPI-DPN Brief Pain Inventory Short Form for diabetic peripheral neuropathy, HADS-A anxiety subscale of the Hospital Anxiety and Depression Scales, HADS-D depression subscale of the Hospital Anxiety and Depression Scales, HbA1c glycosylated haemoglobin, MQS-III Medication Quantification Scale Version III, SD standard deviation
Under the worst case scenario imputation for mortality-associated missing data, the estimated treatment effect for VAS improvement remained statistically significant, though attenuated (p = 0.012). This suggests that the observed long-term benefits of surgery on pain relief are robust, even under highly conservative assumptions regarding the outcomes of deceased patients.
Based on a previous retrospective study [19] investigating short-term (2-year follow-up) outcomes, this follow-up study explored long-term changes in pain and psychological status following LEND in patients with PDPN. In this retrospective real-world cohort with > 1 decade of follow-up, we demonstrate that LEND is associated with sustained, clinically meaningful long-term pain relief, improved pain-related functional interference and psychological symptoms, and a durable reduction in analgesic medication burden in patients with PDPN. Importantly, this analgesic benefit persisted beyond 10 years after surgery, a time horizon that has rarely been reported in the literature on surgical interventions for PDPN. Table 4Measurements changes from baseline by subgroupTime points (years)Focal pain groupDiffuse pain groupFocal group vs diffuse group^a^ScoreMean (SD)Change from baselineMean (SD)pScoreChange from baselineMean (SD)p**pVASBaseline7.89 (1.49)NANA8.00 (0.96)NANANA21.28 (1.92)−6.61 (2.02)< 0.0013.00 (2.89)−5.00 (3.11)< 0.0010.010Last FU1.97 (1.38)−5.92 (2.05)< 0.0012.63 (2.12)−5.38 (2.26)< 0.0010.278BPI-DPN mean Pain Severity IndexBaseline5.66 (1.23)NANA6.27 (1.20)NANANA23.17 (1.74)−2.49 (1.90)< 0.0013.52 (1.80)−2.75 (1.84)< 0.0010.549Last FU1.53 (1.27)−4.13 (1.58)< 0.0012.43 (1.82)−3.84 (2.26)< 0.0010.523BPI-DPN mean Pain Interference IndexBaseline4.74 (0.88)NANA4.91 (0.83)NANANA22.13 (1.26)−2.62 (1.44)< 0.0012.75 (1.55)−2.17 (1.63)< 0.0010.058Last FU1.72 (1.44)3.03 (1.71)< 0.0012.16 (1.61)−2.76 (1.76)< 0.0010.496HADS-ABaseline8.22 (3.39)NANA10.83 (4.08)NANANA25.92 (2.96)−2.31 (2.49)< 0.0016.63 (2.59)−4.20 (3.86)< 0.0010.014Last FU4.06 (1.77)−4.17 (3.16)< 0.0015.23 (2.18)−5.60 (4.34)< 0.0010.107HADS-DBaseline7.97 (2.95)NANA7.75 (2.63)NANANA25.50 (2.24)−2.47 (2.60)< 0.0016.15 (2.12)−1.60 (2.59)< 0.0010.148Last FU5.39 (2.51)−2.58 (4.27)< 0.0015.48 (1.81)−2.28 (3.12)< 0.0010.718MQS-IIIBaseline3.56 (2.55)NANA2.38 (2.47)NANANALast FU1.74 (2.89)−1.82 (3.41)< 0.0011.77 (2.40)−1.61 (3.21)0.0030.786NA* not available, BMI body mass index, BPI-DPN Brief Pain Inventory Short Form for diabetic peripheral neuropathy, FU follow-up, HADS-A anxiety subscale of the Hospital Anxiety and Depression Scales, HADS-D depression subscale of the Hospital Anxiety and Depression Scales, HbA1c glycosylated haemoglobin, MQS-III Medication Quantification Scale Version III, SD standard deviation, VAS visual analogue scale*Statistically significant results (p < 0.05)^a^Comparisons were made between the two groups regarding the mean change of each parameter from baseline
The natural course of PDPN varies among individuals, with some experiencing spontaneous improvement while others may manifest persistent or worsening symptoms. In the medical group, while no significant change in short-term VAS scores was noted, a decline in long-term VAS scores was observed, indicating the existence of long-term spontaneous remission of neuropathic pain in PDPN patients (Fig. 3A). However, according to a small longitudinal study undertaken in the UK, 77% of 56 patients with PDPN continued to experience persistent pain after 5 years, suggesting that neuropathic pain in PDPN can completely resolve over time in a minority of cases [24]. The unclear natural course of PDPN may be attributed to the wide variability in observation periods, study populations, and diagnostic methods [25]. Although data regarding the natural history of PDN are limited, it is notable that pain symptoms can improve and even completely resolve despite the ongoing progression of neuropathy. However, pain alleviation can be associated with worsening sensory function [26].
In the LEND group, while both short- and long-term improvements in VAS scores were observed, no significant changes were noted during the later-stage follow-up (Fig. 3A), indicating that surgical efficacy was achieved early and maintained over a considerable period. The opposite changes in MQS-III scores between the surgical and medical groups signal that nerve decompression confers both short- and long-term pain relief in PDPN. The role of LEND in PDPN remains controversial. Early observational studies and small randomized trials suggested potential benefit [17, 18, 27, 28], whereas more recent rigorously designed trials with sham surgery controls, such as the DNND trial [16], reported more conservative results. Our findings do not contradict these randomized trials but rather provide complementary real-world evidence from a high-volume centre with long-term follow-up. Unlike tightly controlled trials, real-world cohorts reflect routine clinical practice, including patient selection, comorbidity burden, and long-term adherence issues. The sustained pain relief observed in our cohort suggests that a subset of carefully selected patients may derive long-lasting benefit from decompression surgery, even if the average treatment effect across all comers is modest.
Of note, multivariable regression analysis identified age and BMI as significant prognostic factors for long-term surgical outcomes as measured by VAS scores, consistent with previously reported risk factors for neuropathic pain in DPN [29, 30]. Age-related degeneration in the structure and function of peripheral nerves, such as a decrease in nerve conduction velocity and regenerative capacity, may contribute to increased susceptibility to neuropathic pain and decreased neural repair capability in older patients with DPN [31]. Although BMI has been identified as a risk factor for painful DPN, the underlying mechanisms linking obesity and neuropathic pain remain debated and unclear. Recent studies proposed that elevated BMI can cause metabolic dysregulation and adverse pathophysiological changes. Increased release of pro-inflammatory cytokines can result from excess adipose tissue, impairing peripheral nerves and altering pain perception [32]. The results of this study further consolidate our prior proposal on potential interactions between mechanical nerve compression and metabolic impairment, indicating that the efficacy of surgical decompression of peripheral nerves for PDPN is based on good management of metabolic impairment in diabetes, as nerve decompression mainly targets nerve entrapment, while metabolic dysfunction is mainly improved by good blood glucose and weight control [11, 33]. These findings are hypothesis-generating and may help refine patient selection in future studies.
Notably, neither significant short- nor long-term changes were noted in the BPI-DPN pain severity index in the medical group. In the LEND group, not only were short- and long-term improvements in the BPI-DPN pain severity index observed, but also consistent improvement was achieved in the later-stage follow-up. While BPI-DPN uses a numeric rating scale similar to that of the VAS, it includes four the worst pain, least pain, average pain, and current pain [22]. Therefore, the BPI-DPN pain severity index might provide a comprehensive evaluation of neuropathic pain throughout the day as the degree of neuropathic pain fluctuates over 24 h. While similar long-term outcomes were also observed in terms of BPI-DPN pain interference indexes in both surgical subgroups, the focal pain subgroup had a significant long-term improvement in BPI-DPN pain severity compared to the diffuse pain subgroup. Given the chronic nature of pain symptoms in patients with DPN, the impact of pain on patients cannot be overlooked. For example, nocturnal exacerbation of pain symptoms is a well-established feature of PDPN [34], and evaluating the impact of PDPN on sleep has significant implications. These findings are clinically meaningful, as long-term reliance on centrally acting analgesics in patients with diabetes is associated with adverse effects, polypharmacy, and impaired quality of life. Furthermore, the lower cumulative incidence of diabetic foot ulcers observed after LEND suggests a potential role of surgical decompression in modifying long-term peripheral nerve vulnerability, although this association should be interpreted cautiously as causal inference cannot be established in this observational study.
Anxiety and depression are commonly associated with pain [5, 35]. Psychological evaluations assist in indirectly reflecting changes in chronic pain intensity as well as the patients’ coping mechanisms. Herein, suspected anxiety and depression were prevalent among PDPN patients, in agreement with the observations of an earlier study [36]. In contrast to the medical group, in which no significant change in psychological status was noted during the entire follow-up period, both short- and long-term improvements were observed in the LEND group. While consistent improvements in HADS-A scores were noted during the later-stage follow-up, no significant change was noted for depression. Interestingly, a mutual influence was identified between HADS-A and pain interference scores in this study. As demonstrated in the linear regression analyses, a higher preoperative BPI-DPN mean Pain Interference score was significantly associated with greater long-term improvement in HADS-A scores, whilst a higher preoperative HADS-A score was identified as a prognostic factor for better long-term improvement in BPI-DPN mean Pain Interference scores (Table S3, S4). These findings collectively suggest that surgical decompression confers long-term benefits in pain relief and pain-related psychological outcomes. The observed improvement in anxiety symptoms and their association with pain-related interference underscore the bidirectional interaction between psychological factors and chronic neuropathic pain.
In line with short-term outcomes reported in a previous study [19], long-term improvements in VAS scores were observed in both subgroups. Although no significant change was noted during the later-stage follow-up, the trends between the focal and diffuse pain subgroups differed, with VAS scores increasing in the focal pain group but decreasing in the diffuse pain group. In terms of the BPI-DPN pain severity index, besides short- and long-term improvements, a continued significant decrease was achieved in both subgroups. Similar outcomes were observed for the BPI-DPN pain interference index. Based on the collective analysis of similarities and differences, it can be proposed that although the efficacy of nerve decompression might be achieved later in the diffuse pain group than in the focal pain group, both subgroups benefited from nerve decompression, and comparable surgical efficacy was achieved over the long term. These findings suggest that diffuse pain distribution alone should not preclude consideration of decompression surgery in otherwise suitable candidates.
An important clinical question is when lower extremity nerve decompression should be considered in the management of PDPN. The findings of the present study do not support routine use of nerve decompression in all patients with PDPN. Rather, they suggest that careful patient selection is likely to be important. In this cohort, younger age at the time of surgery and lower BMI were independently associated with greater long-term pain improvement, indicating that earlier intervention in selected patients may be associated with more favorable outcomes. In addition, all patients included in the LEND group had persistent neuropathic pain despite ongoing medical therapy, suggesting that nerve decompression may be most appropriately considered in patients with unsatisfactory pain relief under optimized pharmacological treatment. Importantly, subgroup analyses demonstrated sustained long-term improvement in both focal and diffuse pain patterns, indicating that baseline pain distribution alone should not be used as a strict exclusion criterion for surgery. Taken together, these observations support considering LEND as a second-line management option in carefully selected patients with PDPN, particularly those with refractory pain and favourable clinical characteristics while acknowledging that definitive indications require confirmation in prospective studies.
Several inherent limitations of this study merit acknowledgment.
First, the retrospective design and non-randomized treatment allocation introduce potential selection bias.
Second, mortality during long-term follow-up inevitably resulted in missing outcome data. To address this, we performed conservative worst case sensitivity analyses, which continued to support the robustness of the primary findings.
Third, the sample size limited the power of subgroup and prognostic factor analyses, and these results should be interpreted as exploratory. We could not conduct sex- and gender-based analyses beyond adjustment/stratification because of sample size limitations.
Fourth, vascular assessments were not available in this study. Although preoperational computed tomography angiography or vascular ultrasonography were routinely conducted to rule out significant peripheral vascular disease, such as arteriosclerosis obliterans and vascular stenosis, post-surgical vascular follow-up data were not systematically collected. Vascular insufficiency, particularly in patients with diabetic foot ulcers, could influence the outcomes of nerve decompression procedures. Future studies incorporating both pre- and post-surgical Doppler arterial studies would help clarify the role of improved blood flow and its potential contribution to nerve regeneration and symptom relief following decompression surgery.
Finally, this was a single-centre study, which may limit generalizability.
LEND demonstrated long-term efficacy in alleviating pain and concurrently improving the pain-related interference and psychological status of patients with PDPN. Younger age and lower latest BM were revealed to be independently associated with pain relief following LEND. Besides, both the focal and diffuse pain groups benefited from nerve decompression, with pain distribution exerting a limited effect on long-term surgical outcomes. These findings support further prospective, multicentre studies integrating surgical intervention into the comprehensive management framework of DPN.
Below is the link to the electronic supplementary material.Supplementary file1 (PDF 394 KB)