Real‐World Evaluation of Botulinum Toxin A in Focal Peripheral Neuropathic Pain: Longitudinal Outcomes

ABSTRACT

Background

Neuropathic pain (NP) is frequently resistant to conventional treatments. Botulinum toxin type A (BT‐A) is a recommended option for focal peripheral NP, but the dynamics of its effect in real‐life conditions remain poorly characterized.

Objective

To assess BT‐A efficacy in a real‐world study of patients with focal peripheral NP, over a 1‐year follow‐up period.

Methods

In this prospective, observational study, adult patients with chronic focal peripheral NP refractory to standard therapies were treated with BT‐A in two French pain centers. Injections were individualized and administered at approximately 3‐month intervals. The primary outcome was the patient‐reported percentage of improvement since the first injection, assessed at each follow‐up visit (Cycles 2–5). Secondary outcomes included the Patient Global Impression of Change (PGIC), average pain intensity over the previous 8 days, and exploratory analysis of response trajectories distinguishing early (≥ 30% improvement at Cycle 3) from late responders.

Results

Eighty‐two patients received ≥ 2 BT‐A cycles. Mean self‐reported improvement increased from 30.7% at Cycle 2 to 51.0% at Cycle 5 (p < 0.001). PGIC and pain intensity also showed significant improvements. Among 59 patients evaluable at Cycle 3, 66% were early responders; late responders showed significant benefit after the third injection. BT‐A was well tolerated: 63.6% reported end‐of‐dose effects, and 25.8% experienced only mild, transient adverse events.

Conclusion

This real‐life study over 1 year suggests that BT‐A provides progressive and sustained benefit in focal peripheral NP, supporting continued treatment beyond two cycles.

1. Introduction

Neuropathic pain (NP) is a complex and often disabling condition that remains challenging to manage. First‐line and second‐line treatments—such as tricyclic antidepressants, serotonin‐norepinephrine reuptake inhibitors, gabapentinoids, and topical agents like lidocaine or capsaicin—provide partial relief for many patients and are frequently limited by side effects or contraindications [1, 2]. In routine practice, a considerable number of patients remain symptomatic despite appropriate oral or topical therapies, underscoring the need for alternative, locally targeted interventions [3].

Botulinum toxin type A (BT‐A) has emerged as an unexpected candidate in the treatment of peripheral NP and included as a third‐line treatment option for focal peripheral NP [2] and even as second‐line as in France [1]. Its mechanisms of action have not yet been fully elucidated, but are probably peripheral (mainly involving inhibition of neuropeptide release and TRPV1 and P2X3 receptors expression) but also central (secondary to reduced peripheral sensitization and possibly related to retrograde transport to the CNS) [4]. According to the latest meta‐analysis by NeuPSIG, BT‐A used to treat NP has an NNT of 2.7 (95% CI 1.8–5.1) and an NNH of 216.3 (95% CI 23.5–∞) [2]. Among the randomized controlled trials included in this meta‐analysis, the BOTNEP trial was the one with the longest evaluation of the therapeutic effect of BT‐A but its follow‐up was limited to 24 weeks after two cycles [5]. While BOTNEP provided robust data on short‐term efficacy, its fixed protocol and limited follow‐up duration leave open important questions about the longer‐term evolution of response and how such effects manifest in routine clinical care in real life studies.

Such short‐term evaluation in the treatment of NP contrasts with clinical developments of BT‐A in other clinical situations, such as chronic migraine. However, the PREEMPT program in chronic migraine revealed that therapeutic benefit from BT‐A can increase over time, with a substantial proportion of patients achieving meaningful improvement only after the third or fourth cycle [6, 7]. A similar kinetic approach has not yet been explored in NP, although delayed or cumulative responses are likely, particularly in complex and chronic conditions.

This real‐world study aims to assess the kinetic profile of BT‐A efficacy in patients with peripheral NP, with a primary focus on the patient‐reported percentage of improvement across successive treatment visits. In addition, we explored secondary outcomes such as the Patient Global Impression of Change (PGIC) [8], average pain intensity over the previous 8 days, and explored response trajectories, including the distinction between early and late responders, to better characterize patterns of clinical evolution under BT‐A treatment.

2. Methods

2.1. Study Design

This was a prospective, observational, bi‐centric study conducted in two pain centers in southern France (Assistance Publique des Hôpitaux de Marseille and CHU de Nice).

2.2. Patient Population

Adult patients referred for chronic focal peripheral NP refractory to conventional treatments were consecutively included between May 2022 and May 2025. This date corresponded to the predefined censoring point for data extraction.

All patients received botulinum toxin type A (BT‐A) as part of standard care, following a multidisciplinary clinical decision. Eligible patients presented with pain that was localized, persistent for at least 3 months, and compatible with a diagnosis of peripheral NP according to the grading system proposed by the International Association for the Study of Pain (IASP) [9]. Patients had previously failed or not tolerated at least two pharmacological options, including gabapentinoids (gabapentin, pregabalin), tricyclic antidepressants (e.g., amitriptyline), serotonin‐norepinephrine reuptake inhibitors (e.g., duloxetine, venlafaxine), or topical agents (capsaicin, lidocaine).

Patients were excluded if previously treated with BT‐A in the past year, allergic, or pregnant/breastfeeding. All patients provided written consent for the use of their anonymized clinical data for research purposes.

2.3. Treatment Protocol

Patients received subcutaneous injections of onabotulinumtoxinA (Botox), administered according to standard clinical practice. Injection sites and dosages were individualized by the treating physician based on the distribution and intensity of pain, without a standardized injection scheme. In accordance with current prescribing recommendations, the maximum dose per treatment cycle did not exceed 150 units for facial indications and 300 units for other body regions [10, 11]. Treatment cycles were generally spaced at approximately three‐month intervals, although this timing could vary by approximately 15 days depending on clinical or logistical constraints.

On patient request, additional comfort measures could be provided, including topical anesthetic patches (lidocaine/prilocaine), nitrous oxide/oxygen inhalation (N₂O/O₂), relaxation techniques, or hypnosis (delivered by a specialized nurse assisting during the procedure).

2.4. Follow‐Up and Evaluation

Each treatment cycle comprised a clinical evaluation followed immediately by BT‐A injection. The first consultation (Cycle (C) 1) corresponded to the initiation of treatment. For subsequent cycles (C2–C5), the effect of the previous injection was assessed just before the administration of the next. When patients discontinued treatment (most often after C2) follow‐up data could still be collected through routine clinical encounters unrelated to BT‐A or through remote communication (e.g., telephone or email), depending on availability and patient preference.

Only patients who had received at least two treatment cycles were included in the present analysis, as the primary outcome—patient‐reported percentage of improvement—was first assessed during C2. Patients who had only completed C1 by the administrative censoring date (May 2025) were therefore not retained in the analytic sample.

The primary outcome was the percentage of improvement reported by the patient since the first injection. This was collected orally by the physician during each cycle visit, using the standardized question: “Since the first injection, by how much do you feel your pain improved, in percentage?” The answer was recorded in the clinical report, with values ranging from 0% (no improvement) to 100% (complete relief). This outcome was selected as the primary endpoint because it captures the patient's overall perception of treatment benefit across the full inter‐injection interval. Unlike point‐in‐time pain ratings, it reflects the integrated therapeutic effect over the three‐month period, accounting for fluctuations and potential end‐of‐dose phenomena. This patient‐centered approach is consistent with IMMPACT recommendations for chronic pain trials [12]. It was also selected to avoid ceiling effects and to ensure a patient‐personalized primary outcome, as defined by the patient himself [13].

Secondary outcomes were also assessed during each visit. The Patient Global Impression of Change (PGIC) was recorded using a 7‐point Likert scale, from 1 (“very much improved”) to 7 (“very much worse”), as verbally indicated by the patient [8]. Pain intensity over the past 8 days was self‐evaluated on a numerical rating scale (NRS) from 0 to 10 [14]. Pain was assessed based on the mean intensity over the previous 8 days to limit memory bias and better reflect recent clinical status, considering that NP may fluctuate over time and is not consistently perceived at its worst.

Additionally, patients completed standardized self‐administered questionnaires, including the EQ‐5D (heath status) [15], the Hospital Anxiety and Depression Scale (HADS‐A and HADS‐D) [16], and the Neuropathic Pain Symptom Inventory (NPSI) [17] at C1.

Patients were not expected to complete a fixed number of treatment cycles. The duration and completeness of follow‐up varied according to each patient's clinical trajectory and treatment adherence. Patients who had a subsequent injection scheduled beyond the predefined censoring date of May 2025 were considered censored from that timepoint onward.

All patients who received at least two BT‐A injection cycles were eligible for inclusion in the analysis of treatment effectiveness. Reasons for treatment discontinuation or incomplete follow‐up were recorded whenever available.

The number of treatment cycles analyzed was limited to a maximum of five, corresponding to approximately 1 year of follow‐up after BT‐A initiation. The effect of the fifth injection was not evaluated, as C5 served as the final assessment point to capture the longitudinal trajectory of response over 1 year. Although some patients continue treatment well beyond this timeframe in routine care, follow‐up was intentionally restricted to this 1‐year window to allow for standardized comparison across patients and to reflect a meaningful timeframe for clinical decision‐making. In our experience, and in line with clinical reasoning, the absence of meaningful or sustained improvement after 1 year of treatment generally prompts a reassessment of the therapeutic strategy.

Safety/tolerability data and the presence of a patient‐perceived end‐of‐dose effect were collected during follow‐up visits and summarized descriptively.

3. Statistical Analysis

The distribution of continuous variables was assessed using the Shapiro–Wilk test. Results were expressed as mean ± standard error for normally distributed variables or as median with interquartile range (IQR) for non‐normally distributed ones. Categorical variables were reported as frequencies and percentages.

The primary analysis focused on the evolution of the patient‐reported percentage of improvement across treatment cycles (C2–C5). A linear mixed‐effects model was used to account for repeated measures, with time (treatment cycle) as a fixed effect and a random intercept for each patient. This approach accommodates unbalanced follow‐up data and within‐subject variability. Missing data and lost to follow‐up cases were handled within the framework of the mixed‐effects models used for longitudinal analysis, under the assumption of data missing at random (MAR).

Secondary outcomes were modeled using similar linear mixed‐effects approaches. Although PGIC is ordinal, it was treated as a continuous variable to enable longitudinal modeling, in line with prior methodological recommendations [8]. Pairwise comparisons between treatment cycles were performed post hoc using Bonferroni‐adjusted contrasts.

To better characterize the kinetics of treatment response and clarify the source of improvement observed after the third injection (C3), an exploratory analysis was conducted. The objective was to determine whether this effect reflected a continued gain among early responders, or rather the emergence of a delayed therapeutic effect in non‐early responders who had not yet met the threshold for clinical response after two cycles. This distinction was inspired by previous observations in chronic migraine, where both early and late responder profiles to botulinum toxin have been described [6]. As an exploratory analysis, patients were stratified based on their response at C3, that is, after the second injection and before the third. Patients reporting more than 30% improvement at that point were classified as early responders, and those with 30% or less as non‐early responders. A threshold of 30% improvement was used as a clinically meaningful cutoff to distinguish responders, in line with established definitions of minimal clinically important difference (MCID) in chronic pain studies [18]. For each subgroup, the progression of treatment response over time was examined using separate linear mixed‐effects models, allowing for a descriptive evaluation of individual response kinetics. No direct statistical comparison was performed between the two groups.

Statistical significance was defined as a two‐tailed p‐value < 0.05. All statistical analyses were performed using IBM SPSS Statistics (version 20).

4. Results

4.1. Baseline Characteristics

A total of 82 patients who received at least two cycles of botulinum toxin type A were included in the analysis. Baseline clinical and demographic characteristics of the population are summarized in Table 1.

TABLE 1.

Demographic and clinical characteristics.

Characteristic	Mean ± SD (range)
Age (years)	60.1 ± 16.9 (18–96)
Sex (female)	59.8%
Duration of neuropathic pain prior BT‐A initiation (years)	5.07 ± 5.53 (1–45)
Average pain intensity (past 8 days, NRS)	7.44 ± 1.65
HADS—anxiety	8.44 ± 5.51
HADS—depression	10.5 ± 2.83
EQ‐5D Score	0.86 ± 0.05
EQ‐5D VAS (/100)	46.4 ± 19.2
Pain Catastrophizing Score (Sullivan)	27.1 ± 13.2
NPSI—spontaneous burning pain	7.56 ± 2.48
NPSI—paresthesias/dysesthesias	5.00 ± 3.56
NPSI—deep pressing pain	5.50 ± 3.41
NPSI—paroxysmal pain	3.82 ± 3.59
NPSI—evoked pain	3.58 ± 2.94
NPSI—total score (excluding items 4 and 7)	50.8 ± 24.8

The most commonly treated pain localizations were the face (28.0%), trunk (22.0%), and lower limbs (20.7%). Other targeted areas included the upper limbs (12.2%), non‐facial head regions (12.2%), and polyneuropathy‐related diffuse symptoms (11.0%).

The most common causes were post‐traumatic NP (24.4%), post‐herpetic neuralgia (18.3%), post‐surgical pain syndromes (15.9%), and painful trigeminal neuropathy (10.9%). Other etiologies included diabetic neuropathy, tumor‐related pain, trigeminal neuralgia, discogenic pain, ischemic neuropathy, neurosarcoidosis, and amputation‐related pain.

The mean number of injection points per session was 41.7 ± 27.3 (range 3–140), with a mean total dose of 172.9 ± 104.0 units per cycle, adapted to the anatomical distribution and severity of pain.

4.2. Patient Follow‐Up

A total of 82 received at least two injections of BT‐A and were evaluable at C2. At C3, 59 patients remained evaluable, 45 at C4, and 37 at C5. Flowchart and reasons for discontinuation are detailed in Figure 1.

FIGURE 1.

Patient flow and reasons for discontinuation during follow‐up. Flowchart summarizing the number of patients evaluable at each treatment cycle (C2–C5) and the reasons for discontinuation or non‐evaluation between visits. Reasons include insufficient efficacy, adverse events, procedural intolerance (e.g., needle phobia), intercurrent medical conditions, elective discontinuation due to adequate pain control, and visits scheduled beyond the administrative censoring date. Loss to follow‐up was frequently attributed to administrative scheduling issues, missed appointments, or patient relocation. Reasons are reported cumulatively across cycles and are not mutually exclusive.

4.3. Primary Outcome: Patient‐Reported Improvement

The primary outcome was the percentage of improvement reported by the patient at each follow‐up visit, relative to their baseline condition (before the first injection, at C1) (Figure 2).

FIGURE 2.

Primary outcome in the overall population. Mean patient‐reported percentage of improvement from baseline (± standard error of the mean) across treatment cycles C2–C5 in the overall study population (n = 82). The dotted horizontal line represents the threshold for clinically meaningful improvement (30%). Sample sizes at each timepoint are indicated. Blue lines represent statistically significant or non‐significant pairwise comparisons (Bonferroni‐adjusted p‐values) between cycles, based on linear mixed‐effects modeling.

A significant and progressive improvement was observed over time, as assessed by a linear mixed‐effects model (p < 0.001). The mean patient‐reported improvement was 30.7% ± 3.07% at C2, 38.4% ± 3.35% at C3, 44.1% ± 3.72% at C4, and 51.0% ± 4.17% at C5.

Notably, from the first follow‐up visit (C2), the mean improvement exceeded the 30% threshold considered clinically meaningful [18], suggesting a relevant therapeutic effect after a single injection cycle.

Pairwise comparisons showed a significant gain between C2 and C3 (p = 0.017), C2 and C4 (p = 0.001), and C2 and C5 (p < 0.001). A further significant improvement was also observed between C3 and C5 (p = 0.011), supporting the hypothesis that the third injection contributes to continued clinical benefit. While the slope of change was less steep between adjacent later cycles (C3 vs. C4, p = 0.296; C4 vs. C5, p = 0.196), the cumulative effect observed after C3 suggests the importance of evaluating efficacy beyond two treatment cycles.

4.4. Secondary Outcomes

4.4.1. PGIC

The Patient Global Impression of Change (PGIC), scored from 1 (“very much improved”) to 7 (“very much worse”), showed a significant evolution over time according to the linear mixed‐effects model (p < 0.001). Detailed Mean PGIC scores at each cycle and pairwise comparisons within the model are represented in Figure 3a. From the first post‐treatment visit, patients already reported a global impression of improvement ranging between “minimally improved” (score 3) and “much improved” (score 2) on average, reflecting early perceived benefit consistent with the primary outcome. PGIC scores continued to improve numerically between C3 and C5, suggesting a possible ongoing benefit after the third injection, followed by a trend toward clinical stabilization.

FIGURE 3.

Evolution of secondary outcomes over time. N = number of patients evaluable at each timepoint; p‐values from post hoc pairwise comparisons. (a) Patient Global Impression of Change (PGIC) scores across treatment cycles (C2–C5), showing a progressive improvement with a plateau effect after Cycle 3. Mean PGIC scores progressively decreased from 2.85 ± 0.11 at Cycle 2 (C2) to 2.55 ± 0.12, 2.46 ± 0.13, and 2.28 ± 0.15 at C3, C4, and C5 respectively, indicating a gradual shift toward more favorable patient‐reported impressions of change. From the first post‐treatment visit, patients already reported a global impression of improvement ranging between “minimally improved” (score 3) and “much improved” (score 2) on average, reflecting early perceived benefit consistent with the primary outcome. Pairwise comparisons within the model showed that the difference between C2 and all subsequent cycles was statistically significant (C2 vs. C3: p = 0.002; C2 vs. C4: p = 0.006; C2 vs. C5: p = 0.001). No statistically significant differences were observed between later cycles (C3 vs. C4: p = 1.000; C3 vs. C5: p = 0.221; C4 vs. C5: p = 0.426). While statistical significance was not reached, PGIC scores continued to improve numerically between C3 and C5, suggesting a possible ongoing benefit after the third injection, followed by a trend toward clinical stabilization. (b) Pain intensity averaged over the past 8 days (Numerical Rating Scale), showing a significant and gradual decrease across cycles (C1–C5). The mean NRS progressively declined from 6.71 ± 0.19 at Cycle 1 (C1) to 5.95 ± 0.20, 5.56 ± 0.24, 4.97 ± 0.27, and 4.74 ± 0.33 at C2, C3, C4, and C5 respectively, reflecting a gradual reduction in perceived pain intensity over time. Pairwise comparisons indicated significant differences between all successive visits (C1 vs. C2: p < 0.001; C2 vs. C3: p = 0.049; C3 vs. C4: p = 0.009), with the exception of C4 versus C5 (p = 0.329), suggesting a plateauing trend after the fourth cycle.

4.4.2. Pain Intensity Averaged Over the Previous 8 Days (NRS)

Pain intensity, assessed using the numerical rating scale (NRS) averaged over the previous 8 days, showed a significant decrease across cycles according to the linear mixed‐effects model (p < 0.001).

The mean NRS progressively declined from 6.71 ± 0.19 at C1 to 4.74 ± 0.33 at C5, reflecting a gradual reduction in perceived pain intensity over time. Pairwise comparisons are summarized in Figure 3b.

4.4.3. Exploratory Analysis: Response Profiles

Among the 59 patients evaluable at C3, 39 (66.1%) were early responders and 20 (33.9%) were non‐early responders. In the early responder group, 37 patients were still evaluable at C4 and 30 at C5. In the non‐early responder group, 8 patients were evaluable at C4 and 7 at C5. Linear mixed‐effects models were conducted separately in each subgroup to assess the evolution of treatment response (Figure 4c).

FIGURE 4.

Response trajectories in early and non‐early responders. (a) Treatment response over time in early responders. Mean percentage of improvement from baseline (± standard error of the mean) in patients classified as early responders (≥ 30% improvement at Cycle 3), across treatment cycles C2–C5 (n = 39 at C2). The dotted horizontal line marks the clinically meaningful threshold of 30%. Sample sizes at each timepoint are shown above the data points. Blue lines indicate pairwise comparisons between cycles; no statistically significant change was observed beyond Cycle 3 (Bonferroni‐adjusted p‐values). (b) Treatment response over time in non‐early responders. Mean percentage of improvement from baseline (± standard error of the mean) in patients classified as non‐early responders (< 30% improvement at Cycle 3), across treatment cycles C2–C5 (n = 20 at C2–C3). A significant increase in perceived improvement was observed between Cycle 3 and Cycle 4 (p = 0.003), and between Cycle 3 and Cycle 5 (p < 0.001), suggesting a delayed response profile. The dotted horizontal line indicates the clinically meaningful threshold (30%). Blue lines show Bonferroni‐adjusted pairwise comparisons between timepoints. (c) Comparative response kinetics in early and non‐early responders. Mean percentage of improvement from baseline (± standard error of the mean) in early responders (solid line) and non‐early responders (dashed line), from Cycle 2 to Cycle 5. Early responders (≥ 30% improvement at Cycle 3) exhibited a rapid initial benefit that plateaued over time. In contrast, non‐early responders (< 30% improvement at Cycle 3) showed a delayed but progressive improvement, surpassing the clinically meaningful threshold (30%) by Cycle 4. The dotted horizontal line indicates the 30% threshold. These trajectories illustrate the heterogeneous kinetics of response to BT‐A in neuropathic pain.

In early responders, the mean percentage improvement increased from 46.8% ± 3.8% at C2 to 57.6% ± 3.8%, 57.5% ± 3.8%, and 60.9% ± 4.1% at C3, C4, and C5 respectively. A clinically meaningful improvement was already present after the first injection, with the average response exceeding the 30% threshold from C2 onwards. The largest gain was observed between C2 and C3 (p = 0.006), with no statistically significant differences thereafter (Figure 4a).

In contrast, non‐early responders exhibited a different trajectory. Their mean improvement remained low at C2 (10.0% ± 4.2%) and C3 (9.8% ± 4.1%)—below the clinically relevant threshold—but increased significantly at C4 (30.7% ± 5.8%) and C5 (42.2% ± 6.6%). Statistically significant differences were observed between C3 and both C4 (p = 0.003) and C5 (p < 0.001), suggesting a delayed but clinically meaningful response initiated after the third injection (Figure 4b).

4.5. Safety/Tolerability and End‐Of‐Dose

Among the 82 patients included in the study, 62 provided evaluable data regarding treatment safety/tolerability and end‐of‐dose effect. Among them, 63.6% reported experiencing an end‐of‐dose effect, typically described as a progressive return of symptoms before the next scheduled injection. A total of 25.8% of patients reported at least one adverse event, most of which were mild and transient. The most frequently reported adverse effect was transient facial asymmetry (n = 8), followed by transient post‐injection worsening of pain (n = 4), transient paresthesias (n = 1), motor weakness (n = 1), asthenia (n = 1), and transient worsening after the first session (n = 1). One case of facial asymmetry led to treatment discontinuation. No serious adverse events were reported.

Procedural intolerance, such as discomfort or needle phobia, led to treatment discontinuation in seven patients (8.5%), representing the most frequent cause of dropout related directly to the injection procedure. Data regarding procedural intolerance after the first session (C1) were not systematically collected. Transient worsening of pain was reported by five patients (8.0%), typically within the days following injection, and resolved spontaneously without sequelae.

5. Discussion

This prospective real‐world study provides new insights into the kinetic profile of BT‐A in peripheral NP. In line with prior evidence, we observed a clinically meaningful reduction in patient‐perceived global improvement from the first injection, with sustained and cumulative benefit across treatment cycles. Notably, the concordance with PGIC and NRS evolution reinforces the robustness of our findings.

The trajectory of response we observed aligns with, but also extends, the results of the BOTNEP randomized controlled trial [5]. In our study, we confirm similar efficacy trends: a clinically relevant benefit after a single injection, followed by further improvement after the second. Our population was comparable in size and age distribution, although we included a broader spectrum of etiologies, notably facial pain syndromes such as trigeminal neuropathies, which were not represented in BOTNEP. The inclusion of these localizations is supported by previous studies suggesting that BT‐A may provide significant benefit in trigeminal neuralgia or painful trigeminal neuropathy, particularly in refractory cases [19, 20]. In BOTNEP, the mean reduction in daily pain intensity at week 24 was −2.6 points on a 0–10 scale in the BT‐A group, with an effect size of approximately 30%–40% depending on the analysis method [5]. In our cohort, the mean percentage of improvement at the third cycle (6 months) was 38.4%, with similar gains observed in PGIC and clinically meaningful changes in NRS scores. These consistent results across settings reinforce the real‐world reproducibility of BT‐A efficacy in focal peripheral NP. Furthermore, our study extends these findings by showing that treatment benefit continues to increase beyond the second cycle, with a cumulative mean improvement reaching 51.0% at 1 year.

Importantly, the number of patients still evaluable at C5 (n = 37) closely matches that of BOTNEP at 24 weeks (n = 34 per group), providing a meaningful point of comparison despite methodological differences. Our data also highlight the importance of response kinetics, a dimension not addressed in BOTNEP due to its fixed two‐cycle design. Inspired by the PREEMPT program in chronic migraine [6, 7], we explored early and late responder trajectories in an exploratory analysis. While two‐thirds of patients were early responders, a notable proportion of initial non‐responders exhibited delayed but clinically meaningful benefit after the third injection, suggesting a cumulative effect of BT‐A in certain patients. However, this analysis must be interpreted with great caution, particularly beyond Cycle 3. The number of patients in the non‐early responder subgroup decreased markedly at C4 (n = 8) and C5 (n = 7), limiting statistical power and generalizability. These exploratory findings should therefore be considered hypothesis‐generating, underscoring the need for prospective studies specifically designed to characterize kinetic response profiles and predictors of delayed benefit.

Interestingly, while perceived improvement continued to rise, the PGIC scores showed a plateau effect after the third cycle, indicating a possible ceiling in subjective benefit or a stabilization of patient expectations. This phenomenon may also reflect the non‐linear nature of Likert‐type scales such as PGIC, where a one‐point difference does not necessarily correspond to an equal clinical difference across the scale. This limitation, inherent to ordinal ratings, has been highlighted in prior methodological studies [8], and suggests that PGIC should be interpreted in conjunction with other continuous, patient‐centered measures [13].

The absence of a placebo control is a common limitation in real‐world observational studies. However, the longitudinal, intra‐individual design, in which each patient serves as their own control, offers meaningful insight into treatment effect under routine care conditions. Moreover, our primary endpoint—the patient‐reported percentage of improvement—was selected precisely to reflect the patient's pertinent and personalized subjective perception of benefit, which remains the cornerstone of pain management. This choice appears especially appropriate in our study, given that nearly two‐thirds of patients reported an end‐of‐dose effect, indicating a fluctuating but perceptible benefit over the entire injection interval. Unlike point‐in‐time measures such as the NRS or PGIC, the percentage of improvement integrates the patient's overall experience across the full cycle and thus better captures the therapeutic dynamics of BT‐A.

The onset of analgesic effect was not systematically recorded in our study, preventing formal analysis of time to onset after each injection. This aspect should be addressed in future real‐life studies, as it is a key factor for patient information and expectation management.

Regarding safety, BT‐A was generally well tolerated, with 25% of patients reporting only mild and transient adverse events. Procedural intolerance led to dropout in 8.5% of patients, a rate that should be acknowledged in real‐life practice given the chronic nature of treatment. While relatively low, this figure underlines the importance of discussing potential discomfort with patients and considering comfort measures to enhance adherence. Transient post‐injection pain exacerbation occurred in 8% of our cohort, consistent with previous observations in interventional pain procedures [5]. Although mild and self‐limited, such events can temporarily affect comfort and satisfaction and warrant pre‐procedural counseling. Overall, our findings align with previous trials [5] and confirm the favorable tolerability profile of BT‐A in routine care.

The 3‐month interval between cycles was chosen in line with the BOTNEP protocol, which demonstrated robust efficacy over two cycles [5]. However, our finding that nearly two‐thirds of patients reported end‐of‐dose effects indicates that shorter or individualized intervals might be worth considering, particularly for those with consistent early waning of benefit. This high prevalence also underlines the need for clinicians to anticipate symptom re‐emergence before reinjection and to tailor follow‐up accordingly.

As with any observational study, our results must be interpreted in light of several limitations. The attrition rate beyond Cycle 3 is substantial, with only 37 of the initial 82 patients evaluable at C5. While this limits statistical power and may introduce selection bias, it reflects real‐world constraints such as intercurrent illness, logistical barriers, or elective discontinuation after satisfactory response. Notably, most reasons for loss to follow‐up were documented and were not systematically related to treatment inefficacy, supporting the plausibility of the missing‐at‐random (MAR) assumption used in our modeling. In this context, the flexibility of real‐world data may introduce heterogeneity, but also enhances external validity, offering a more realistic view of long‐term BT‐A use in diverse clinical settings.

6. Conclusion

In this real‐world cohort of patients with focal peripheral NP, BT‐A treatment was associated with a clinically meaningful improvement from the first injection, a progressive cumulative benefit over successive cycles, and a favorable tolerability profile. Nearly two‐thirds of patients reported an end‐of‐dose effect, and both early and late responder profiles contributed to the overall therapeutic gain. These findings suggest that BT‐A may offer sustained benefit over a 1‐year period, supporting its use as a long‐term option in focal peripheral NP management. However, the durability of efficacy beyond 1 year, the optimal number of cycles before concluding non‐response, and the identification of predictive factors all remain to be clarified. These results should be confirmed in larger, prospective studies with longer follow‐up and stratified designs, to better inform individualized treatment strategies.

Author Contributions

Eva Sole‐Cruz, Elise Van Obberghen‐Blanc, Chloé Hirtz, Michel Lanteri‐Minet, Anne Donnet: conceptualization. Eva Sole‐Cruz, Michel Lanteri‐Minet: methodology. Eva Sole‐Cruz: software. Eva Sole‐Cruz, Imene Bennour: data curation. Eva Sole‐Cruz, Elise Van Obberghen‐Blanc, Chloé Hirtz, Michel Lanteri‐Minet, Anne Donnet: investigation. Imene Bennour, Michel Lanteri‐Minet, Anne Donnet: validation. Eva Sole‐Cruz: formal analysis. Elise Van Obberghen‐Blanc, Chloé Hirtz, Michel Lanteri‐Minet, Anne Donnet: supervision. Eva Sole‐Cruz: visualization. Imene Bennour, Michel Lanteri‐Minet, Anne Donnet: project administration. Eva Sole‐Cruz: writing – original draft. Elise Van Obberghen‐Blanc, Chloé Hirtz, Imene Bennour, Michel Lanteri‐Minet, Anne Donnet: writing – review and editing.

Conflicts of Interest

AD received personal fees for consultancy activities from: AbbVie/Allergan, Amgen, Eli Lilly, Lundbeck, Novartis, Orion Pharma, Perfood, Pfizer, Teva. MLM received personal fees for consultancy activities from: Abbvie/Allergan, Amgen, Eli Lilly, IPSEN, Lundbeck, Novartis, Orion Pharma, Perfood, Pfizer, Reckitt‐Benckiser, Teva, UPSA. ESC; EVOB, CH and IB declare no conflicts of interest.

Sole‐Cruz E., Van Obberghen‐Blanc E., Hirtz C., Bennour I., Lanteri‐Minet M., and Donnet A., “Real‐World Evaluation of Botulinum Toxin A in Focal Peripheral Neuropathic Pain: Longitudinal Outcomes,” European Journal of Neurology 32, no. 9 (2025): e70362, 10.1111/ene.70362.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.