Long-Term Efficacy and Safety of Treatments for Chronic Peripheral Neuropathic Pain: A Systematic Review and Narrative Synthesis

Abstract

Introduction

Peripheral neuropathic pain (PNP) is a chronic condition that often requires long-term treatment. In 2025, the Neuropathic Pain Special Interest Group (NeuPSIG) updated its recommendations for treatment of neuropathic pain based on data from double-blind, randomized controlled trials (RCTs) with a duration of ≥ 3 weeks. To complement the existing body of evidence, we reviewed the long-term efficacy/safety of NeuPSIG-recommended treatments for PNP.

Methods

We conducted a systematic literature review including clinical trials and observational studies of any design. Two literature searches were performed using PubMed and Embase (December 2024, May 2025). We included studies that enrolled ≥ 50 adult patients with PNP receiving ≥ 1 treatment recommended in the 2025 NeuPSIG guidance and with duration of ≥ 13 weeks. A supplemental search for long-term studies on high-concentration capsaicin 8% patch (HCCP) was performed. Results were synthesized narratively due to heterogeneity in study designs and outcomes.

Results

We included 34 articles in our analysis (31 studies) varying in design, including interventional (RCTs and open-label studies) and observational studies. Most studies had a duration of 26–52 weeks. Four treatments (HCCP, duloxetine, gabapentin, pregabalin) were evaluated in ≥ 5 studies. HCCP was evaluated in the greatest number of long-term studies (n = 13) and in various PNP conditions, followed by duloxetine (n = 7) in painful diabetic peripheral neuropathy (pDPN) only, gabapentin (n = 6) in various PNP conditions and pregabalin (n = 5) mostly in pDPN. Collected data support long-term efficacy of all products. The lowest rates of discontinuations due to adverse events were reported with HCCP.

Conclusions

The findings from this literature review emphasize the importance of expanding long-term evidence for chronic pain therapies. The review determined that the largest body of evidence is available for HCCP. Finally, it highlights the need for more standardized long-term evidence studies for chronic pain therapies.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1007/s40122-026-00884-1.

Key Summary Points

Peripheral neuropathic pain (PNP) is a chronic condition that often requires long-term treatment, yet data supporting long-term treatment are scarce.

A systematic literature review (SLR), including clinical trials and observational studies of any design, was conducted to explore the long-term efficacy/safety of treatments for PNP recommended by the Neuropathic Pain Special Interest Group (NeuPSIG).

The identified studies varied in design, including interventional (randomized controlled and open-label studies) and observational studies, with most having a duration of 26–52 weeks.

Four treatments (high-concentration capsaicin 8% patch [HCCP], duloxetine, gabapentin, pregabalin) were evaluated in ≥ 5 long-term studies.

It is important to expand long-term evidence for chronic pain therapies, and this evidence should be taken into consideration when making treatment decisions.

Introduction

Peripheral neuropathic pain (PNP) is a chronic, debilitating condition often requiring long-term therapy over many years [1, 2]. It has a complex and heterogeneous underlying pathophysiology, and although understanding of its disease mechanisms has increased, it remains challenging to treat effectively and often requires a multidisciplinary approach [3–5]. Management of PNP often focuses on treatment of symptoms rather than the underlying cause, and patients frequently require combination therapy, including pharmacological and non-pharmacological interventions [4, 6]. Commonly used pharmacological treatments include oral analgesic drugs, such as α2δ ligands (e.g., pregabalin), serotonin and norepinephrine reuptake inhibitors, and tricyclic antidepressants [7]. These treatments do not always offer complete pain relief, with pain reductions ranging between 30 and 40%, and may be associated with adverse events (AEs) [8]. Any of these factors—alone or in combination—can lead to treatment discontinuation, suboptimal adherence, cycling through different oral treatments or a combination of treatments and, in some cases, potential for abuse [5, 8–11]. Topical treatment options are available, including lidocaine 5% plaster, low-concentration capsaicin creams, and high-concentration capsaicin 179 mg (8% w/w) patch (HCCP)—a topical transient receptor potential vanilloid 1 (TRPV1) agonist providing neuropathic pain symptom relief for up to three months with each treatment through defunctionalization of nociceptors and nerve fiber regeneration [12, 13]. However, topical treatment options remain under-utilized in clinical practice [8, 9, 14]. As a result, PNP is often persistent and undertreated, with patients experiencing severe disability, poor quality of life and high healthcare utilization [9].

Recently, the Neuropathic Pain Special Interest Group (NeuPSIG) updated their treatment recommendations for neuropathic pain (i.e., central and peripheral neuropathic pain) to include new evidence from double-blind, randomized, placebo-controlled trials with a parallel or crossover design and trial duration of 3–24 weeks [7]. It included both pharmacological and neuromodulation interventions for neuropathic pain (NeuP) that were administered for at least 3 weeks, or in which patients had at least 3 weeks of follow-up after a single administration. The group of experts conducted a systematic literature review (SLR), calculated the risk difference for all comparisons and performed a random-effects meta-analysis. Recommendations were based on evidence of efficacy, AEs, accessibility, cost and feedback from patients.

NeuPSIG provided a strong recommendation for the use of α2δ ligands (e.g., pregabalin), serotonin and norepinephrine reuptake inhibitors, and tricyclic antidepressants as first-line treatments; a weak recommendation for using lidocaine 5% medicated plaster, HCCP and capsaicin cream as second-line treatments (first-line for susceptible patients such as elderly, those with many comorbidities or receiving multiple medications with high risk of drug interactions); and a weak recommendation for using botulinum toxin type A, repetitive transcranial magnetic stimulation and opioids as third-line treatments [7]. The authors stated that the findings and recommendations highlight the modest treatment efficacy of pharmacological interventions and the need for longer-term, large-scale, placebo- or sham-controlled trials performed over clinically relevant timeframes.

PNP is often a chronic condition, and patients are likely to receive treatment for PNP for long periods—sometimes lifelong. Therefore, it is important to evaluate the benefit-risk profile of PNP treatments beyond the 24-week timeframe assessed in the NeuPSIG recommendations [15, 16].

We sought to complement the existing body of evidence reviewed by NeuPSIG [7] and therefore reviewed the available long-term (≥ 13 week) efficacy and safety data of those PNP treatments. We conducted an SLR that included interventional and observational studies, and descriptively summarized efficacy and safety/tolerability findings for recommended treatments with at least five identified long-term studies. To our knowledge, no prior review has synthesized long-term evidence across all NeuPSIG-recommended treatments.

Methods

Inclusion and Exclusion Criteria

The SLR followed the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) methodology [17]. We considered peer-reviewed articles reporting studies that evaluated ≥ 1 treatment recommended in the 2025 NeuPSIG guidance [7], with durations ≥ 13 weeks (excluding titration) and that enrolled ≥ 50 adult patients with PNP. The inclusion and exclusion criteria used to select studies (Table 1) were designed to be as inclusive as possible to ensure the SLR was comprehensive. The SLR was conducted in accordance with a prespecified protocol that is available from the authors upon request.

Table 1.

Study inclusion and exclusion criteria

	Included	Excluded
Population	Adults (≥ 18 years old) diagnosed with peripheral neuropathic pain caused by nerve injury or disease (IASP definition) [16] Important causes include diabetic neuropathy and chronic postsurgical pain	Central neuropathic pain (i.e., neuropathic pain mediated by central causes) Trigeminal neuralgia Acute pain (≤ 3 months) Children (< 18 years)
Interventions of interest	Pregabalin Gabapentin Duloxetine Venlafaxine Amitriptyline Imipramine Clomipramine Desipraminea Mirogabalina Nortriptylinea Maprotilinea High-concentration (8%/179 mg) capsaicin patch Low-concentration capsaicina 5% lidocaine plaster Tramadol Other opioids: tapentadol, fentanyl, oxycodone ± naloxone, morphine, buprenorphine and hydromorphone Botulinum toxin A Combination therapy involving any of above treatments	Non-drug therapies (e.g., TENS) Lidocaine cream
Outcomes	Primary efficacy outcomes: • Change from baseline in pain intensity in the long term (≥ 42 weeks) • Responder rates in the long term (≥ 42 weeks): Response to treatment (≥ 30% decrease from baseline) • Short (≤ 13 weeks) and intermediate efficacy outcomes (e.g., at 36 weeks) will also be recorded Safety outcomes: • Treatment-related adverse AEs and discontinuations	• Studies not reporting change in pain or treatment-related AEs • Studies only reporting short-term (≤ 13 weeks) data
Study design	Clinical trials (phases II, III, and IV), any design Observational studies, any design Note: Any systematic literature reviews and meta-analyses that meet other inclusion/exclusion criteria were checked so all relevant studies were included	Case reports
Duration of treatment	> 13 weeks or > 3 months	≤ 13 weeks
Treatment setting	All settings where PNP is treated
Report characteristics	Peer-reviewed articles	Conference abstracts
Timeframe	Past 20 years	Studies published before December 2004
Language	Any, provided title and abstract are available in English
Other		Studies with < 50 participants in total Review articles Non-human studies

AE adverse event, IASP International Association for the Study of Pain, NeuPSIG Neuropathic Pain Special Interest Group, PNP peripheral neuropathic pain, TENS transcutaneous electrical nerve stimulation

^aInterventions included in the addendum, following publication of the 2025 NeuPSIG guidance on treatment of neuropathic pain

Search Strategy

Two database searches were performed in Embase (www.embase.com) and PubMed (https://pubmed.ncbi.nlm.nih.gov); an original search in December 2024 and a second search (addendum) in May 2025. The original search was performed using the recommended treatments outlined in the 2015 NeuPSIG guidance [18], and the addendum (May 2025) considered studies of five treatments added in the 2025 guidance [7]—mirogabalin, nortriptyline, maprotiline, desipramine and low-concentration capsaicin (Table 1 includes inclusion/exclusion criteria for both searches). Reference checking was also performed to identify additional studies. All search strategies are shown in the supplementary material (Tables S1–S4).

Embase and PubMed results were combined in Rayyan [19] and checked for duplicates, which were removed. Titles and abstracts were screened by two independent reviewers (MA, DB) against inclusion/exclusion criteria, with each reviewer being blinded to the other during the screening. SLRs or meta-analyses that were identified were retrieved and the reference lists were scanned for additional studies. After completion of title/abstract screening by both reviewers, the reviewers’ selections were unblinded, and discrepancies were resolved by consensus.

Articles that were not excluded at title/abstract screening were reviewed as full-text documents. This was done by both MA and DB, both blinded to the other’s review. Once full-text review was complete, the decisions were unblinded, and any discrepancies were resolved through consensus with a third reviewer (RF).

To complement the systematic literature search, we conducted targeted identification of additional publications reporting long-term effectiveness of HCCP from existing reference records, applying the same pre-defined inclusion and exclusion criteria. This search was conducted as the initial search did not pick up large studies that the authors were aware of (e.g., Lantéri-Minet et al. [20]).

Data Extraction

Efficacy and safety data on PNP treatments were extracted from the selected articles. Each reviewer extracted approximately half of the articles and then checked the data that had been extracted by the other reviewer for potential errors. We extracted data on the following efficacy/effectiveness outcomes for baseline, major reported timepoints (e.g., 3 months) and study end: change from baseline in average pain intensity (over the previous 24 h) measured by the numerical pain rating scale (NPRS) score; proportion of patients achieving a ≥ 30%, ≥ 50% and ≥ 80% reduction/decrease from baseline in 24-h average pain intensity (response to treatment); and pain reduction by any scale if NPRS scores were not included (absolute and relative reductions). The safety/tolerability outcomes included percentage of patients who discontinued treatment and percentage of patients who discontinued treatment due to AEs. If studies included treatments not listed in the protocol, only data from the treatments of interest were included.

Due to the heterogeneity of study designs, including both observational and interventional studies, a formal risk-of-bias assessment was not performed, and the review focused on descriptive synthesis of the available evidence.

Results

Literature Searches on PNP Treatments

Figure 1 shows the flowchart of study selection based on the three searches that were performed.

Fig. 1.

PRISMA diagram of included articles. Two searches were performed in Embase and PubMed; an original search in December 2024 and a second search (addendum) in May 2025. HCCP high-concentration capsaicin patch, PRISMA Preferred Reporting Items for Systematic reviews and Meta-Analyses, SLR systematic literature review

We screened the title and abstract of 716 articles and reviewed the full text of 76 (Fig. 1). Following full-text review, 31 articles were included from database searches; a further three HCCP articles were identified separately through targeted searching. Table S5 includes details of 39 articles that appeared to meet the inclusion criteria but were excluded upon full-text review. Among the 34 included articles, some publications reported overlapping datasets: Vinik et al. and Vinik et al. reported the same HCCP study [21, 22], Bienfait et al. [23] reported a subset of patients from the larger Dupoiron et al. study [24], and Eardley et al. and Bestard et al. reported the same gabapentin dataset with different comparators (Table 2) [30, 31].

Table 2.

Summary of efficacy and safety outcomes from selected studies per PNP treatment based on SLR and additional search (treatments with ≥ 2 studies)

Reference	PNP type	Study design	Number of patients (dose)	Study duration	Key efficacy outcome	Efficacy data	Discontinuations (including reasons)
Pregabalin
Happich, 2014 [25]	pDPN	Observational, prospective, three-arm (D, G, P)	248 Mean dose (P): 173.5 mg/day	≤ 6 months	Change in mean daily pain intensity (SD) from BL to week 24 as measured on BPI	– 1.9 (2.22)	17.7%c
Patel, 2014 [26]	pDPN	Observational, prospective, single-arm	Total: 101 P (dose NR): 32	≤ 6 months	Number of patients with VAS score ≥ 5 (scale 0–10) at BL and month 6	BL: 4 Month 6: 0	NR
Satoh, 2011 [27]	pDPN	Interventional, prospective, open-label, single-arm, extension	123 Dose: 150–600 mg/day BID	> 12 months	Change in mean daily pain intensity from BL to week 52 as measured on VAS (100-mm scale)	– 25.4 mm	n = 26 (21.1%) AEs: n = 19 (15.4%); lack of efficacy: n = 3 (2.4%); other: n = 4 (3.3%)
Stacey, 2008 [28]	pDPN/PHN	Interventional, open-label, single-arm, extension	81 Dose: 150–600 mg/day BID	> 12 months	Change in mean daily pain intensity from BL to month 15 as measured on VAS (100-mm scale)	DPN: – 25.4 mm, p < 0.001 PHN: – 23.2 mm, p < 0.001	n = 25 (30.9%) AEs: n = 10 (12.3%); lack of efficacy: n = 5 (6.2%); LTFU: n = 6 (7.4%)
Toth, 2010 [29]	pDPN: 38% Idiopathic: 34% Monoclonal gammopathy: 11% Vit B12 deficiency: 6% Autoimmune: 10% Multiple myeloma: 1%	Observational, prospective, open-label, three-arm (G continued; G responders switch to P; G non-responders switch to P)	Total: 146 Switch G respondersg (mean dose of pregabalin: 389 mg/day): n = 33 Switch G non-responders:h (mean dose of pregabalin: 405 mg/day): n = 36	> 12 months	Change in mean daily pain intensity (SD) from BL to month 12 as measured on VAS	Switch G responders: 27% (13%) Switch G non-responders: 24% (11%)	Switch G responders: n = 1 (3%) (LTFU) Switch G non-responders: n = 7 (19.4%) (LTFU)
Gabapentin
Bestard, 2011 [30]	Idiopathic: n = 41 pDPN: n = 70 Vit B12: n = 39 Other: n = 99	Observational, prospective, open-label, four-arm (N or G MonoTx; N or G as adjuvant)	Total: 249 G adjuvant: 64; mean dose: ~ 2300 mg/day MonoTx: 52; mean dose: ~ 2400 mg/day (No Tx: 29)	≤ 6 months	Change in mean daily pain intensity (SD) from BL to month 6 as measured on VAS	Adjuvant: – 16.9 (9.6) MonoTx: – 16.4 (9.1)	Adjuvant: n = 18 (28%) AEs: n = 10 (16%); lack of efficacy: n = 8 (13%) MonoTx: n = 20 (38%) AEs: n = 12 (23%); lack of efficacy: n = 8 (15%)
Eardley, 2010 [31]	Idiopathic: n = 20 pDPN: n = 33 Vit B12: n = 18 Other: n = 45	Observational, prospective, non-randomized, open-label, four-arm (V or G MonoTx; V or G as adjuvant therapy)	Adjuvant G: 64; mean dose: ~ 1900 mg/day MonoTx G: 52; mean dose: ~ 2400 mg/day (No Tx: 29)	≤ 6 months	Change in mean daily pain intensity (SD) from BL to month 6 as measured on VAS (scale 0–10)	Adjuvant: – 16.9 (9.6) (33.9%) MonoTx: – 16.4 (9.1) (32.7%)	Adjuvant: AEs: n = 10 (16%); lack of efficacy: n = 8 (13%) MonoTx: AEs: n = 11 (22%); lack of efficacy: n = 8 (15%)
Giannopoulos, 2007 [32]	pDPN	Interventional, randomized-controlled (SSRI vs. G)	Total: 101 G: 51 Dose: 400–1600 mg/day	≤ 6 months	How did the pain feel with treatment: better, no effect or worse?	51% felt better; 41% no effect; 9% worse	Four patients (7.8%) discontinued due to intolerable dizziness or ataxia
Happich, 2014 [25]	pDPN	Observational, prospective, three-arm (D, G, P)	351 Mean dose (G): 727.8 mg/day	≤ 6 months	Change in mean daily pain intensity (SD) from BL to week 24 as measured on BPI	– 1.1 (2.15)	12.5%c
Ruessmann, 2009 [33]	pDPN	Observational, retrospective, two-arm (G, untreated)	Total: 443 G: 293 Dose: 600–2400 mg/day (No Tx: 150)	≤ 6 months	Responders with at least moderate relief from pain at month 6	45% (n = 132)	n = 131 (45%) (all due to AEs)
Toth, 2010 [29]	pDPN: 38% Idiopathic: 34% Monoclonal gammopathy: 11% Vit B12 deficiency: 6% Autoimmune: 10% Multiple myeloma: 1.9%	Observational, prospective, open-label, three-arm (G continued; G responders switch to P; G non-responders switch to P)	Total: 146 Continuous G (no switch): 47	> 12 months	Mean daily pain intensity (SD) at month 12 as measured on VAS	BL G: 7.7 (0.7) Month 12 G: 5.8 (1.0)	n = 7 (15%) (all due to LTFU)
Duloxetine
Happich, 2014 [25]	pDPN	Observational, prospective, three-arm (D, G, P)	931 Mean dose (D): 53.9 mg/day	≤ 6 months	Change in mean daily pain intensity (SD) from BL to week 24 as measured on BPI	–2.3 (2.30)	14.8%c
Raskin, 2006a [34]	pDPN	Interventional, open-label, randomized, (duloxetine, RCd)	Total: 449 60 mg BID: 334 120 mg QD: 115	≤ 6 months	Change in mean daily pain intensity (SD) from BL to week 24 as measured on BPI	– 2.7 (approximate, from chart)	60 mg: n = 121 (36%) 120 mg: n = 43 (37%) NUMBER (%) of total D: AEs: n = 98 (21.8%); protocol violation/protocol entry criteria not met: n = 18 (4%); personal conflict/patient decision/consent withdrawal: n = 18 (4%); lack of efficacy: n = 10 (2.2%); physician decision: n = 9 (2%); LTFU: n = 5 (1.1%); death: n = 3 (0.7%); other clinically significant laboratory value: n = 2 (0.4%); sponsor’s decision: n = 1 (0.2%)
Raskin, 2006b [35]	pDPN	Interventional, open-label, randomized-controlled	Total: 237 D 60 mg BID: 161 RC: 76	> 12 months	NR	NR	n = 45 (28%) AEs: 9.3%; death: 1.2%; LTFU: 3.7%; patient decision: 8.7%; physician decision: 0.6%; protocol violation: 1.2%; lack of efficacy: 3.1%
Skljarevski, 2009 [36]	pDPN	Interventional, prospective, open-label, two-arm (after 8 weeks of tx with 60 mg QD, responders continued on 60 mg QD; non-responders on 120 mg QD)	Total: 216f 60 mg QD (maintenance): 115 120 mg QD (rescue): 69	6–12 months	Change in mean daily pain intensity from BL as measured on BPI	Acute phase (weeks 0–8): – 2.49, p ≤ 0.001 60 mg QD: + 0.35 120 mg QD: – 1.39, p ≤ 0.001	n = 97 (44.9%) AEs: n = 39 (18.1%); lack of efficacy: n = 29 (13.4%); other: n = 29 (13.4%)
Wernicke, 2006 [37]	pDPN	Interventional, randomized controlled OLE to an RCT with two arms (D or RC)	Total: 337 D 60 mg BID: 222 RC: 115	12 months	NR	NR	D 60 mg BID: n = 63 (28.4%) AEs: n = 31 (14%); lack of efficacy: n = 4 (1.8%); LTFU: n = 4 (1.8%); patient decision: n = 18 (8.1%); sponsor’s decision: n = 1 (0.5%); protocol violation: n = 1 (0.5%); death: n = 2 (0.9%); physician’s decision: n = 2 (0.9%) RC: n = 22 (19.1%)
Wernicke, 2007 [38]	pDPN	Interventional, randomized OLE to an RCT with two arms (D or RC)	Total: 293 D 60 mg BID or QD: 197 (RC: 96)	12 months	NR	NR	D 60 mg BID or QD: n = 22 (11.2%) AEs (including death): n = 11 (5.6%); patient decision: n = 7 (3.6%); physician decision: n = 1 (0.5%); protocol violation: n = 1 (0.5%); lack of efficacy: n = 2 (1%)
Yasuda, 2016 [39]	pDPN	Interventional, randomized OLE to an RCT with 2 arms (D 40 mg or 60 mg)	Total: 258 D 40 mg/day: 129 D 60 mg/day: 129	12 months	Change in mean daily pain intensity (SD) from BL to week 52 as measured on BPI (scale 0–10)	D 40 mg/day: – 2.1 (1.7), p < 0.0001 D 60 mg day: – 2.1 (1.6), p < 0.0001	40 mg/day: n = 30 (23.3%) AEs: n = 24 (18.6%); lack of efficacy: n = 1 (0.8%); consent withdrawal: n = 4 (3.1%); unspecified: n = 1 (0.8%) 60 mg/day: 29.5% (n = 38) AEs: n = 30 (23.3%); lack of efficacy: n = 2 (1.6%); consent withdrawal: n = 4 (3.1%); unspecified: n = 2 (1.6%)
HCCP
Bienfait, 2023 [23]	CIPN	Observational, retrospective, single-arm	57	> 12 months	Complete or important impact on pain as measured on CGIC at EoS	Found in 61 applications (33.2%) concerning 30 patients (52.6%)	No pain: n = 8 (14%); patient decision: n = 6 (10.5%); LTFU: n = 10 (17.5%); death: n = 12 (21.1%)
Dupoiron, 2022 [24]	PSNP: 62% CIPN/RIPN: 18% Other: 20%	Observational, retrospective, single-arm	279	> 12 months	Complete or important impact on pain as measured on CGIC at EoS	82.3% of patients	No pain: n = 63 (22.6%) Patient decision: n = 22 (7.9%); LTFU: n = 39 (14%); lack of efficacy: n = 7 (2.5%); death: n = 39 (14%); other: n = 19 (6.8%)
Gálvez, 2017 [40]	PHN: 107 PNI: 99 HIV-DSP: 80 Other PNP: 20	Observational, prospective, single-arm	306	6–12 months	Change in mean daily pain intensity (SD) from BL to month 12 as measured on BPI Q5	–1.9 (1.89)	n = 130 (42.5%) Lack of efficacy: n = 54 (17.6%); consent withdrawal: n = 33 (10.8%); LTFU: n = 17 (5.6%)
Kern, 2024 [41]	NBP: 33% (n = 32) PSNP/PTNP: 25% (n = 24) Others (incl. PHN, pDPN, CIPN, CRPS): 42%	Observational, retrospective, single-arm	97	> 12 months	Important impact on pain as measured on BOS-3 (and medical records) at EoS	65% of patients	NR
Mankowski, 2017 [42]	PHN: 89 HIV-AN: 5 NBP: 50 CRNP: 22 PSNP/PTNP: 198 Other: 56	Observational, prospective, single-arm	420	6–12 months	Change in mean pain intensity (95% CI) from BL to week 52 as measured on NPRS (scale 0–10)	– 37% (95% CI 31.3–42.7; n = 176)	Not available
Schweitzer, 2015 [43]	PSNP/PTNP: 69.1% PHN: 9.9% pDPN: 4.4% Other: 16.6%	Observational, retrospective, single-arm	1st Tx: 91 2nd Tx: 61 3rd Tx: 35	> 12 months	Change in mean pain intensity from BL to month 12 as measured on NPRS (scale 0–10)	– 1.9 (27.1%; p < 0.05)	After 1st Tx: n = 21/91 (23.1%) After 2nd Tx: n = 17/61 (27.9%)
Simpson, 2010 [44]	PHN: n = 54 HIV-DSP: n = 52	Observational, prospective, single-arm	106	6–12 months	Change in mean pain intensity from BL to week 48 as measured on NPRS (scale 0–10)	PHN: – 35.6% HIV-DSP: – 12.4%	n = 27 (25.5%) PHN: n = 11 (20%) HIV-DSP: n = 16 (31%) LTFU: n = 10 (9%); AEs: n = 5 (5%); lack of efficacy: n = 5 (5%); other: n = 5 (5%); non-compliance: n = 2 (2%)
Treillet, 2023 [45]	PSNP: 49.1% (n = 115) PTNP: 24.8% (n = 58) PHN: 8.5% (n = 20) pDPN: 3.8% (n = 9) Iatrogenic: 4.3% (n = 10) Idiopathic: 7.3% (n = 17) Sarcoidosis: 1.7% (n = 4) Small-fiber neuropathy: 0.4% (n = 1)	Observational, retrospective, single-arm	333	> 12 months	≥ 30% pain relief (NPRS); ≥ 50% pain relief (NPRS)	81.3% of patients; 66.9% of patients	NR
Vinik, 2016 [22]/ Vinik, 2019 [21]	pDPN	Interventional, open-label, randomized, active-controlled, three- arm (HCCP + SoC vs. SoC)	Total: 468 30-min HCCP + SoC: 156, 60-min HCCP + SoC: 157, SoC: 155	6–12 months	Change in mean pain intensity (SD) from BL to week 52 as measured on BPI-DPN Pain Severity Index (scale 0–10)	30-min HCCP + SoC: – 1.9 (1.8) 60-min HCCP + SoC: – 2.2 (1.9) SoC: – 0.9 (1.7) Mean difference (90% CI): 30-min HCCP + SoC vs. SoC: – 0.9 (– 1.3, – 0.6) 60-min HCCP + SoC vs. SoC: – 1.2 (– 1.6, – 0.9)	n = 80 (17.1%) 30-min HCCP + SoC: n = 24 (15.4%) LTFU: n = 2 (1.3%); consent withdrawal: n = 10 (6.4%); AEs: n = 7 (4.5%); protocol violation: n = 1 (0.6%); lack of efficacy: n = 3 (1.9%); other: n = 1 (0.6%) 60-min HCCP + SoC: n = 29 (18.5%) LTFU: n = 1 (0.6%); consent withdrawal: n = 15 (9.6%); AEs: n = 8 (5.1%); protocol violation: n = 3 (1.9%); lack of efficacy: n = 1 (0.6%); other: n = 1 (0.6%) SoC: n = 27 (17.4%) LTFU: n = 1 (0.6%); consent withdrawal: n = 19 (12.3%); AEs: n = 3 (1.9%); protocol violation: n = 3 (1.9%); other: n = 1 (0.6%)
Hussain, 2021 [46]	pDPN	Interventional, double-blind, randomized, placebo-controlled, three-arm (LL/LP/LC)	Total: 273 L + HCCP: 89 L + placebo: 90	≤ 6 months	Change in mean daily pain intensity (95% CI) from BL to week 24 as measured on BPI-DPN Q4 (scale 0–10)	LC: – 2.2 ± 1.5 (– 2.45, – 1.5), p < 0.01 LP: – 0.4 ± 0.3 (– 0.48, – 0.35), p = 0.89	NR
Lanteri-Minet, 2019a [20]	PSNP: 76.3% PHN: 9.1% Polyneuropathy: 8% Other: 6.6%	Observational, prospective, single-arm	684	> 12 months	Success: Reduction in pain intensity of ≥ 30% and improved PGIC (6 months after last application) Failure: reduction in pain intensity < 30% and no improvement in PGIC (6 months after last application)	Success: 21.8% (95% CI, 17.5–26.7) Moderate success: 40.3% (95% CI, 35.0–45.8) Failure: 37.9% (95% CI, 32.6–43.4)	Not available
Hansson, 2018a [47]	PNI: 70% PHN: 13% PNP: 5% Other: 12% (1st Tx)	Observational, prospective, single-arm (data from three studies)	1st Tx: 412 2nd Tx: 184	≤ 6 months	Change in mean pain intensity (SD) from week 2 to 12 as measured on NPRS ‘usual’ score (scale 0–10)	1st Tx HCCP: – 0.97 (2.04), p < 0.001 2nd Tx HCCP: – 0.54 (1.87),b p < 0.001	1st Tx HCCP: n = 30 (7.3%) LTFU: n = 11 (2.7%); lack of efficacy: n = 7 (1.7%); consent withdrawal: n = 4 (1%); progression of concurrent diagnoses: n = 2 (0.5%); surgery/injury of treated area: n = 2 (0.5%); incorrect enrolment: n = 1 (0.2%); death: n = 1 (0.2%); sensitivity to HCCP: n = 1 (0.2%); hospitalization due to pneumonia: n = 1 (0.2%) After 2nd Tx HCCP: n = 3 (1.63%)
Lux, 2024a [48]	CIPN	Observational, prospective, single-arm	73	6–12 months	Change in mean pain intensity (SD) over the last 4 weeks from BL as measured on NPRS (scale 0–10)	1st Tx (V1) HCCP: – 0.88 (2.29), n = 57, p = 0.006 2nd Tx (V2) HCCP: – 1.64 (2.28), n = 45, p < 0.0001 3rd Tx (V3) HCCP: – 1.18 (2.67), n = 39, p = 0.009	Total: n = 35 (48%) Logistical issues: n = 3 (4.1%); medical conditions: n = 4 (5.5%); lack of efficacy: n = 8 (11%); pain during application: n = 4 (5.5%); LTFU: n = 7 (9.6%); other: n = 9 (12.3%) After 1st Tx HCCP: n = 15 (20%) After 2nd Tx HCCP: n = 12 (21%) After 3rd Tx HCCP: n = 8 (17%)
Lidocaine 5% medicated plaster
Hussain, 2021 [46]	pDPN	Interventional, double-blind, randomized, placebo-controlled, three-arm (LL/LP/LC)	Total: 273 L pre-treatment then L: 94 L pre-treatment then placebo: 90	≤ 6 months	Change in mean daily pain intensity (95% CI) from BL to week 24 as measured on BPI-DPN Q4 (scale 0–10)	LLL: – 1.0 ± 1.1 (– 1.25, – 0.85), p < 0.05 LP: – 0.4 ± 0.3 (– 0.48, – 0.35), p = 0.89	NR
Hans, 2009 [49]	PHN	Interventional, open-label, single-arm extension of a phase III trial	Safety: 249 Efficacy: 247	> 12 months	Mean daily pain intensity (SD) at month 12 as measured on NPRS	New patients: BL: 5.9 (1.4); Month 12: 3.9 (1.6) Final visit (week 52): 3.9 (2.3) Pre-treated patients: BL: 3.9 (1.9); Final visit: 3.4 (2.0)	n = 106 (42.6%) AEs: n = 30 (12%); lack of efficacy: n = 51 (20.5%); consent withdrawal: n = 21 (8.4%); protocol violation: n = 7 (2.8%); death: n = 3 (1.2%); other: n = 25 (10%)
Mirogabalin
Baba, 2020e [50]	pDPN	Interventional, open-label, single-arm, extension of a phase III trial	214 Dose: 5 mg BID increased to 10 or 15 mg BID	> 12 months	Change in mean daily pain intensity (SD) from BL to week 52 as measured on VAS (100-mm scale)	– 9.8 mm (14.1)	n = 42 (19.6%) AEs: n = 21 (9.8%); consent withdrawal: n = 19 (8.9%)
Yamanaka, 2022e [51]	PNP (spine diseases) LCS: 65.1% CSR: 28.3% LDH: 6.6%	Observational, retrospective (patients who switched from P to M)	106 Mean dose: 13.8 mg/day	> 12 months	Patient-determined response to tx (yes or no)	66% of patients	n = 23 (21.7%) AEs: n = 11 (10.4%); lack of efficacy: n = 12 (11.3%)

AE adverse event, BID twice daily, BL baseline, BOS-3 Behavioural Observation Scale-3, BPI Brief Pain Inventory, BPI-DPN Brief Pain Inventory for Diabetic Peripheral Neuropathy, CGIC clinician global impression of change, CI confidence interval, CIPN chemotherapy-induced peripheral neuropathy, CRNP chronic radicular neuropathic pain, CRPS complex regional pain syndrome, CSR cervical spondylotic radiculopathy, D duloxetine, DPN diabetic peripheral neuropathy, EoS end of study, G gabapentin, HCCP high-concentration capsaicin patch, HIV-AN human immunodeficiency virus–associated autonomic neuropathy, HIV-DSP human immunodeficiency virus–associated distal sensory polyneuropathy, L lidocaine, LC lidocaine + capsaicin, LCS lumbar canal stenosis, LDH lumbar disc herniation, LL lidocaine + lidocaine, LP lidocaine + placebo, LTFU lost to follow-up, M mirogabalin, MonoTx monotherapy, N nabilone, NBP neuropathic back pain, NPRS numerical pain rating scale, NR not reported, OLE open-label extension, P pregabalin, pDPN painful diabetic peripheral neuropathy, PGIC patient global impression of change, PHN post-herpetic neuralgia, PNI peripheral nerve injury, PNP peripheral neuropathic pain, PSNP post-surgical neuropathic pain, PTNP post-traumatic neuropathic pain, Q# question #, QD once daily, RC routine care, RCT randomized controlled trial, RIPN radiation-induced peripheral neuropathy, SD standard deviation, SLR systematic literature review, SoC standard of care, SSRI selective serotonin reuptake inhibitor, Tx treatment, V venlafaxine, V# visit #, VAS visual analogue scale, Vit vitamin

^aIncluded following HCCP manual search in May 2025

^bFrom pre-treatment BL at week 12 after first tx

^cReasons for discontinuations not reported by study tx

^dConsisted primarily of gabapentin, amitriptyline, and venlafaxine

^eIncluded following the May 2025 addendum to the SLR

^fPatients were initially treated with duloxetine 60 mg QD for 8 weeks (acute phase); responders continued on the same dose, while non-responders had their duloxetine dose increased to 120 mg QD (maintenance and rescue phase). ^gPregabalin in patients who were gabapentin responders

^hPregabalin in patients who were gabapentin non-responders

Characteristics of Selected Studies on PNP Treatments: Patient Population, Study Design and Duration

Of the 34 articles included, 16 articles were predominantly in patients with painful diabetic peripheral neuropathy (pDPN) and 18 articles addressed other PNP etiologies (Table 2).

There was a mixture of study designs, including observational studies (e.g. retrospective chart reviews or prospective pragmatic studies), randomized controlled trials (RCTs) and single-arm open-label interventional studies. Most studies had a duration of 26–52 weeks (Table 2). One study investigated pregabalin, gabapentin, and duloxetine [25]; one study investigated gabapentin and pregabalin [29]; one study investigated venlafaxine and gabapentin [31]; and one study investigated lidocaine 5% medicated plaster and HCCP [46]. All other studies had a single-arm design or compared the intervention with routine/standard of care (SoC) (Table 2).

Four treatments (HCCP, duloxetine, gabapentin, pregabalin) were evaluated in ≥ 5 studies with duration ≥ 6 months (all identified in the SLR) (Fig. 2, Table 2). HCCP was investigated in the greatest number of long-term studies (n = 13) in PNP of a range of etiologies, including pDPN. Most studies (n = 11) had a duration of > 6 months. Duloxetine was investigated in seven studies (only in pDPN); five with follow-up > 6 months. Gabapentin was investigated in six studies (in pDPN and other PNP etiologies), one with follow-up of 12 months. Pregabalin was investigated in five studies (in pDPN and other PNP etiologies), three with follow-up > 6 months. In these long-term studies, there were no restrictions on use of concomitant medications.

Fig. 2.

Number of studies per PNP treatment identified in two database searches plus an additional search (treatments with ≥ 2 studies). HCCP high-concentration capsaicin patch, PNP peripheral neuropathic pain

Of the remaining PNP treatments, mirogabalin and lidocaine 5% medicated plaster were investigated in two long-term studies each, while botulinum toxin type A, amitriptyline and venlafaxine were evaluated in only one long-term study each. There were no published long-term data for any of the other treatments (imipramine, clomipramine, tramadol, opioids, low-concentration capsaicin) included in the 2025 NeuPSIG recommendations.

Summary of Efficacy and Safety Data of PNP Treatments Investigated in the Greatest Number of Long-Term Studies

Owing to design differences across studies, efficacy and safety outcomes were reported heterogeneously. Where available, efficacy was assessed using measures such as the patient global impression of change (PGIC) scale, clinician global impression of change (CGIC) scale, absolute or relative change from baseline in average pain score with either the NPRS or a visual analogue scale (VAS), and responder rates. In line with the analysis presented by NeuPSIG in 2025, safety was assessed using the proportion of patients discontinuing treatment due to AEs. When available, the total proportion of patients discontinuing treatment was also included. Outcomes are presented in Table 2.

Pregabalin

The identified studies were generally consistent with maintained pain relief with pregabalin over follow-up periods from 6 to 15 months; most patients studied had pDPN (Table 2). Two of the five publications showed data over the whole study period versus baseline, suggesting that, after the initial decrease in pain scores, pain scores remained stable over the following 12–15 months [27, 28]. Discontinuation rates varied across studies ranging between 3% and ~ 31% [28, 29], while AE-related discontinuations were reported in two studies, with rates of ~ 12% and ~ 15%, respectively (Table 2) [27, 28].

Gabapentin

Most long-term studies had a duration of ≤ 6 months and investigated a population of patients with pDPN. These studies suggested sustained analgesic effect in patients receiving adjuvant or gabapentin monotherapy (Table 2). In two studies, improved pain relief was observed when comparing 3- and 6-month timepoints (Table 2) [30, 31]. Four of five studies reported reasons for discontinuations due to AEs, with rates ranging from ~ 8 to 45% (Table 2) [32, 33].

Duloxetine

The studies were consistent, with maintained pain relief at 6–12 months with duloxetine in pDPN (Table 2). Two studies showed data over the whole study period versus baseline [36, 39]. Both suggested that, after the initial decrease in pain scores, pain scores remained stable over the following 6–12 months. Discontinuation rates ranged from ~ 11 to ~ 45% [36, 38]. Of the six studies that reported reasons for discontinuations, discontinuation due to AEs ranged from ~ 6 to ~ 18% (Table 2) [36, 38].

HCCP

The identified HCCP studies were broadly consistent with sustained pain relief over 6–12 months (Table 2) in PNP with a range of etiologies. Across studies and irrespective of number of treatments used, ~ 30 to 80% of patients achieved clinically meaningful pain relief [20, 23, 41, 42, 45]. In the two identified RCTs, reductions in average pain intensity were numerically greater with HCCP than with SoC alone [21] or placebo [46]. Across multiple studies, after initial decreases in pain scores, improvements were either maintained or continued with repeated HCCP treatments, including between later timepoints such as 3 and 12 months [40, 42–44, 47, 48]. In an RCT, a greater proportion of patients showed a 30%/50% response with HCCP + SoC compared with SoC alone at month 12 (67.3–67.5%/44.8–48.4% vs. 40.6%/23.8%) [21]. Total discontinuation rates ranged from ~ 17 to 48% [22, 48]. Reporting of discontinuation rates varied across studies; some reported discontinuation in the total population, others reported it per treatment group, number of HCCP applications, or by reason (e.g. lack of efficacy, AEs, lost to follow-up), and some studies did not report any discontinuations (Table 2) Of the seven studies that reported reasons for discontinuations, two studies reported the rate of discontinuation due to AEs (~ 5%) (Table 2) [22, 44]. In the RCT comparing HCCP + SoC with SoC alone, discontinuation rates due to AEs were ~ 4.5–5% vs. 1.9%, respectively [22].

A meta-analysis of efficacy (e.g. at 6 or 12 months) was not feasible owing to the heterogeneity of the data. Nevertheless, for individual treatments (pregabalin, gabapentin, duloxetine and HCCP), the collected evidence suggests maintenance of effect over ≥ 6 months. Across the studies that reported AE-related discontinuations, rates ranged from approximately 12–15% with pregabalin [27, 28], 8–45% with gabapentin [32, 33], 6–18% with duloxetine [36, 38], and around 5% with HCCP [22].

Discussion

Summary and Interpretation of Results

Our systematic review aimed to explore the long-term efficacy and safety of pharmacological treatments for PNP that were included in the 2025 NeuPSIG recommendations. Our definition of ‘long-term’ was a duration of ≥ 13 weeks and we included a broad range of designs, including interventional and observational studies. Our review identified a limited number of long-term studies, with only four of the NeuPSIG-recommended first- and second-line therapies (HCCP, duloxetine, gabapentin, pregabalin) being evaluated in ≥ 5 published studies assessing long-term effects. These studies varied extensively in design, patient population and outcome reporting, highlighting a lack of standardization in long-term studies of pain medication [52]. Additionally, most studies used an open-label design without restrictions on concomitant medication use, making it difficult to isolate the effect of the treatment under investigation. Of the four therapies evaluated in ≥ 5 published studies, HCCP was investigated in the greatest number of studies and across various PNP conditions including post-herpetic neuralgia, peripheral nerve injury, chemotherapy-induced peripheral neuropathy, HIV-associated peripheral neuropathy, and pDPN, whereas the other therapies were evaluated in fewer conditions (e.g. duloxetine was studied only in pDPN). Oral treatments (pregabalin, gabapentin) as well as topical treatments (HCCP, lidocaine) showed a sustained analgesic effect up to 12 months.

To interpret the magnitude of changes in pain intensity compared with baseline, reductions of ≥ 10–20% are generally considered minimally important, reductions of ≥ 30% reflect at least moderate clinically important differences, and reductions of ≥ 50% represent substantial improvements [53]. Only one comparative study was identified that reported data on responder rates. In this study evaluating HCCP in pDPN, ~ 45–48% of patients treated with HCCP + SoC achieved substantial pain relief compared with ~ 24% with SoC alone [21]. In this and several other studies, successive HCCP treatments were associated with increasing benefit, based on measures such as evaluation of pain relief over time and PGIC [21, 40, 42–44, 47, 48]. In general, total discontinuation rates reported were higher in observational studies compared with the RCTs across treatments (Table 2).

Long-term data are scarce, despite the clear clinical need for information on treatment persistence in chronic PNP conditions. This need is driven by the limitations of current therapeutic approaches when used over longer periods, including tolerability issues (such as drowsiness, dizziness and the potential for addiction with pharmaceutical treatments; risks of infection, bleeding and device-related complications with invasive treatments), as well as varying efficacy and diminishing effect over time [54]. Potential reasons for the limited number of published long-term studies include regulatory approvals only requiring short-term placebo-controlled trials, with the recommended duration of controlled treatment being ~ 12 weeks [55, 56]; high discontinuation rates, often within 6 months, in chronic NeuP populations [57]; ethical challenges associated with the conduct of long-term RCTs [58]; and the heterogeneity of NeuP conditions [59].

This lack of long-term data has not only been acknowledged as a limitation by NeuPSIG but also by the Scottish Intercollegiate Guidelines Network (SIGN) for chronic pain (updated in 2025). They found significant limitations in the design of studies evaluating pain, as well as the quality and certainty of the data derived from these studies, with one of the chief limitations being short study duration [60]. SIGN highlights a lack of long-term efficacy and safety data for opioids (> 12 months) and antidepressants (> 3 months) for the treatment of chronic pain and calls for optimization of clinical trials assessing chronic pain. The results of our research are in line with these findings. They confirm that most recommended first-line treatments for PNP have only limited data on long-term efficacy and tolerability. In contrast, HCCP (recommended in the second-line setting and first-line in specific situations) was associated with the largest number of publications with data on long-term efficacy and safety. Additionally, our research highlighted a lack of consistency in data collection across long-term trials, which limits cross-study comparisons. Greater standardization in long-term study design and outcome reporting would improve the interpretability of future evidence.

Strengths and Limitations

Strengths of our analysis include the breadth of the literature searches, which encompassed a wide range of studies, capturing real-world effectiveness, long-term outcomes, AEs and patient populations often excluded from RCTs. However, meta-analysis and direct comparisons of treatments were not feasible owing to the limited number and the heterogeneity of the selected studies and patient populations. Another limitation is that three HCCP articles were identified outside the main database searches, suggesting that the search strategy may not have captured all potentially relevant long-term studies across treatments.

Implications

PNP is a chronic condition that requires effective and well-tolerated long-term treatments [15, 16]. The 2025 NeuPSIG recommendations represent one of the most comprehensive and quantitative evidence syntheses for the management of PNP so far and is used by clinicians globally [7]. They were built on a large and rigorous evidence base with a systematic review and meta-analysis of 313 RCTs, including 284 pharmacological and 29 neuromodulation studies, providing pooled estimates of treatment efficacies and evaluating potential biases. They also take into account safety and the patient perspective by considering accessibility and cost. However, the 2025 NeuPSIG recommendations highlight that efficacy outcomes of treatments are modest and there is a need for longer-term RCTs assessing treatments for PNP, as duration of selected studies was ≤ 24 weeks. Limiting eligibility to double-blind, randomized, placebo-controlled trials meant that longer-duration studies, typically conducted in open-label or real-world formats, were not captured, resulting in an evidence base focused on short-term outcomes.

Our findings highlight that a substantial evidence gap remains for most treatments recommended for PNP in the 2025 NeuPSIG guidance, underscoring the need for long-term research in chronic PNP management. Addressing the lack of long-term data in chronic PNP management will require optimization of post-launch observational research and evidence synthesis. Key considerations for future clinical studies may include broadening patient selection to ensure that a real-world population is represented; ensuring the sample size is large enough to account for long-term discontinuations; and incorporating flexible dosing or combination therapy with appropriate dose adjustments. Results from traditional RCTs often cannot be generalized to real-world populations, and real-world evidence (RWE) from observational studies is increasingly important [61]. RWE can complement RCT data by exploring the effectiveness/safety of a medication and in identifying patient subgroups that may respond better to treatment or be exposed to heightened risk. Furthermore, treatment selection decisions must take into consideration factors beyond clinical effectiveness and safety, such as patient preference, economic and healthcare system integration considerations [62].

Further research will be needed to investigate how meta-analyses and guideline development can incorporate RCT and RWE data to capture long-term outcomes, including durability of response and long-term safety [60, 61, 63]. Lessons can be drawn from other chronic conditions in the cardiovascular and oncology fields, where guidelines are underpinned by long-term studies [64, 65].

Conclusions

This review provides an overview of the published long-term evidence for PNP treatments recommended by the 2025 NeuPSIG guidance. We found a limited number of long-term studies of PNP treatments, with HCCP identified as the most studied treatment, followed by duloxetine, gabapentin and pregabalin. Many of the first- and second-line treatments recommended by NeuPSIG for PNP lack published long-term data. The largest body of published evidence regarding long-term decrease in pain intensity and tolerability was found for HCCP. The studies included varied widely in design, patient population and outcome reporting, limiting direct comparisons and highlighting a lack of standardization in studies of pain medication. These findings emphasize the need for more standardized long-term evidence studies for chronic pain therapies.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contributions

Mariëlle H. Eerdekens, Charlotte Both, and Rita L. Freitas contributed to the study conception and design, and drafting and/or critically revising the manuscript. All authors met ICMJE criteria and all those who fulfilled those criteria are listed as authors.

Funding

The study was funded by Grünenthal GmbH. The Journal’s Rapid Service Fee was funded by Grünenthal GmbH.

Data Availability

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Conflict of Interest

Mariëlle Eerdekens, Charlotte Both, and Rita Freitas are employees of Grünenthal GmbH, or its affiliates. Grünenthal holds the global rights to HCCP (Qutenza), one of the NeuPSIG-recommended treatments for PNP included in the scope of the SLR reported in this manuscript.

Ethical Approval

This article is based on previously conducted studies and does not contain any new studies with human participants or animals performed by any of the authors.