Targeting neurotrophic factors for low back pain and sciatica: a systematic review and meta-analysis

Abstract

Objectives

This meta-analysis aims to investigate the efficacy and safety of medicines that target neurotrophic factors for low back pain (LBP) or sciatica.

Methods

We searched published and trial registry reports of randomized controlled trials evaluating the effect of medicines that target neurotrophic factors to LBP or sciatica in seven databases from inception to December 2020. Two reviewers independently identified studies, extracted data, and assessed the risk of bias and certainty in the evidence.

Results

Nine studies (3370 participants) were included in the meta-analyses. Low certainty evidence showed that anti-nerve growth factor (NGF) may reduce pain at 4 weeks (mean difference [MD] −6.75, 95% CI: −8.61, −4.90) and 12 weeks (MD −6.16, 95% CI: −8.38, −3.94), and may increase adverse effects for chronic LBP (odds ratio [OR] 1.18, 95% CI: 1.01, 1.38). Higher doses of anti-NGF may offer a clinically important reduction in pain at the cost of increased adverse effects for chronic LBP. Very low certainty evidence showed that anti-NGF and glial cell line-derived neurotrophic factor (pro-GDNF) may not reduce pain for sciatica at 4 weeks (MD −1.40, 95% CI: −8.26, 5.46), at 12 weeks (MD −2.91, 95% CI: −13.69, 7.67) and may increase adverse effects for sciatica (OR 3.27, 95% CI: 1.78, 6.00).

Conclusion

Anti-NGF may offer small reductions in pain intensity for chronic LBP. The effect may depend on the dose and types of medicines. For sciatica, anti-NGF or pro-GDNF may not reduce pain. Medicines that target neurotrophic factors for LBP or sciatica are associated with different adverse effects compared to those observed in commonly prescribed medicines for these conditions.

back pain, analgesics, drug therapy, monoclonal antibodies, nerve growth factor review, meta-analysis

Rheumatology key messages

• The overall effects of anti-NGF medicines are small and not clinically meaningful for chronic low back pain (LBP).

• Higher doses of anti-NGF may offer a clinically important reduction for chronic LBP.

• Medicines targeting neurotrophic factors are associated with adverse effects for chronic LBP and sciatica.

Introduction

Low back pain (LBP) has been the leading cause of disability worldwide for at least 30 years [1, 2]. LBP is a common reason for presentations to emergency departments, general practice and rehabilitation services worldwide [3–5]. Even after treatment, a substantial number of people report severe disability (28%) [6, 7] and persistent pain 3 months after the first episode (65%) [8, 9].

Clinical guidelines for LBP now recommend pharmacological management only for those who fail to respond to non-pharmacological interventions, mainly due to evidence of limited effectiveness and concerns over safety [10–12]. To improve the pharmacological management of musculoskeletal and neuropathic pain, novel analgesic agents with different mechanisms compared with commonly prescribed medicines have been studied in recent years [13].

Biologic medicines targeting neurotrophic factors (e.g. nerve growth factor [NGF], brain-derived neurotrophic factor, glial cell-derived neurotrophic factor [GDNF], neurotrophin-3 and neurotrophin-4 [14, 15]) have been developed and tested in patients with musculoskeletal and neuropathic pain [16–18]. NGF molecules, for example, are overexpressed in nociceptive and neuropathic pain conditions [14, 19], and the inhibition of NGF attenuates hyperalgesia in a variety of pain models, including OA, bone fracture pain and autoimmune arthritis [19]. NGF inhibitors may have highly precise pharmacological mechanisms of action and different adverse effect profiles when compared with conventional medicines [20, 21]. NGF inhibitors are commonly administered to people with chronic pain via subcutaneous and intravenous injections, delivered at least 1 week apart [22, 23]. Results of randomized controlled trials (RCTs) of biologics for OA have shown promising analgesic effects [22], leading pharmaceutical companies to apply for regulatory approval with the Food and Drug Administration (FDA) for commercial use of tanezumab, an anti-NGF medicine for this condition [24, 25]. However, the efficacy and safety of medicines targeting neurotrophic factors for LBP or sciatica remain uncertain due to a small number of published studies in previous systematic reviews and the lack of a detailed evaluation of the included studies in the earlier reviews [19, 23, 26]. This systematic review aims to evaluate the efficacy and safety of medicines that target neurotrophic factors in patients with LBP or sciatica.

Methods

We prospectively registered the protocol on the Open Science Framework (OSF) on 19 May 2020 (osf.io/b8adn) and published the protocol in January 2021 [27]. We reported the findings of this study according to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guideline [28].

Data sources and searches

We searched MEDLINE (Ovid), Embase (Ovid), CINAHL (EBSCO), Cochrane Central Register of Controlled Trials (CENTRAL), ClinicalTrials.gov, EU Clinical Trials Register and the World Health Organization’s (WHO) International Clinical Trial Registry Platform from inception to 2 December 2020. We used terms for randomized controlled trials, LBP and spinal disorders, and neurotrophic factors (Supplementary Data S1, available at Rheumatology online). We reviewed the reference list from retrieved full-texts and previous reviews to identify additional eligible studies [14, 19, 23, 26, 29].

Study selection

We included published and unpublished records of parallel-group RCTs that allocated adult participants with LBP or sciatica (as defined by authors) with any pain duration to receive any medicine that targets neurotrophic factors; and either (i) a placebo medicine or (ii) another treatment or (iii) continuation of usual care or (iv) placement on a waiting list or (v) no treatment. We included records written in any language that we could read or translate. We excluded trials that sampled participants with specific spinal pathology other than sciatica, such as fracture, infection, neoplasm, metastasis or inflammatory disease (e.g. ankylosing spondylitis). Enriched designs were excluded because there is evidence that they may underestimate adverse effects [30]. Investigators independently screened the titles, abstracts and full texts in duplicate. We resolved disagreements through discussion or consultation with a third independent reviewer.

Outcomes

The primary outcomes were back pain intensity (measured by any self-reported scale) and safety. The secondary outcomes were leg pain intensity, low back-specific function, measured by any self-reported scale, and harm. Safety and harm were defined as the number of participants who reported any adverse effect and serious adverse effect (defined by each study) during the treatment period, respectively [31]. For sciatica, leg pain was defined as pain intensity in the leg [32]. Back pain intensity, leg pain intensity and low back-specific function were measured at the time point closest to 4, 12, 24 and 48 weeks after the first injection, regardless of the number of injections administered in the study.

Data extraction and quality assessment

Two reviewers independently extracted data (characteristics of the trial, participants, interventions, comparisons and outcomes), appraised bias and rated the confidence in the evidence from the included studies using a standardized, piloted Excel spreadsheet. When further information was required, we contacted authors or funding agencies three times within 6 weeks. Where data (s.d., s.e., or P-values) for conducting meta-analyses were absent, we estimated measures of variance using the recommendations in the Cochrane Handbook for Systematic Reviews [33].

We used the Cochrane ‘Risk of bias’ tool (version 5.1.0) [33] and additional Cochrane recommendations for back pain studies to appraise bias at the study level, prioritizing information regarding the primary outcomes for judgements [34]. We determined the overall risk of bias for each trial by adapting the criteria reported in a previous study [35] that rate the overall risk of bias as ‘low’ when three or fewer domains are rated unclear risk, and no domains were rated high; ‘moderate’ overall risk of bias if a single domain was rated as high risk, but four or more were rated unclear; and ‘high’ overall risk of bias in all other instances.

We rated the confidence in the evidence for each analysis using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) system [36, 37]. We took a conservative approach to GRADE assessments, considering studies with a ‘moderate’ overall risk of bias at a ‘high’ risk of bias.

Data synthesis and analysis

We conducted meta-analyses of trials that reported sufficient data for each outcome and compared the effect of medicines with placebo. We stratified all analyses by follow-up time point (4, 12, 24 and 48 weeks) and the clinical condition of the participants (LBP or sciatica). We conducted subgroup analyses by medicine target (type of neurotrophic factor such as anti-NGF and pro-GDNF) and anti-NGF type (fasinumab, fulranumab and tanezumab). We incorporated trials with multiple comparisons when different doses were compared with placebo by dividing the number of participants in the placebo group by the number of arms included in the study analysis [33]. To facilitate the clinical interpretation of our results, we converted all outcome data for pain and function to a common 0–100 scale [38]. We used the mean difference between groups and accompanying 95% CI to report the effects of continuous outcomes and considered 10 points as the minimal clinically important effect for pain and function [38, 39]. We used odds ratio (OR) and accompanying 95% CI to report the effects of dichotomous outcomes [40]. We also presented the absolute effects of dichotomous outcomes using risk difference to aid interpretability [41].

We synthesized data using random-effects meta-analysis models. We fit the models using the restricted maximum likelihood (REML) approach with the metafor package [42] in R (version 4.0.1) [43]. For each analysis, we computed Cochran’s heterogeneity statistic (Q), the between-study variance (τ²), and the percentage of variance across studies not due to sampling error (I²-values). We assessed heterogeneity and publication bias following recommended guidelines [44–46] as described in detail in our protocol [27]. We communicated the findings in accordance with GRADE guidelines for informative statements [47]. Data from studies that met the inclusion criteria but were not included in the meta-analyses were reported narratively (e.g. studies that did not compare medicines of interest with placebo or did not provide sufficient data for quantitative analyses).

Sensitivity analyses

In the main analysis, we removed studies where measures of variance were imputed, studies that were classified as high risk of bias and studies with <10 participants per group to assess the influence of these factors on the results. We constructed extended funnel plots to explore the potential impact of a new trial on the effect estimate for pain intensity by estimating the parameters of a hypothetical trial needed to reach a minimal clinically important difference of 10 mm (100 mm visual analogue scale) favouring the medicines of interest [48].

Results

The search identified 1932 records. After removing duplicates, we screened 1654 titles and abstracts for inclusion. We excluded 1628 records and retrieved full texts of 26 potentially eligible records. Nine clinical trial registrations were linked to their journal articles, and five journal articles were excluded after full-text screening (Supplementary Table S1, available at Rheumatology online). We included 12 unique trials. From all included trials, we extracted data from nine studies for quantitative analyses (Fig.  1).

Fig. 1

PRISMA flow diagram of the record selection process

PRISMA: Preferred Reporting Items for Systematic Reviews and Meta-Analysis; RCT: randomized controlled trial.

Study characteristics

All included studies (n = 12) used a parallel-group design. Ten trials (3791 participants and 30 unique comparisons) were included in meta-analyses. Of these, six analysed data from 3412 participants with chronic LBP and four analysed data from 379 participants with sciatica. The characteristics of the studies are described in Table  1. All studies, except one, used an 11-point numerical pain rating scale.

Table 1

Characteristics of included studies in the systematic review

Record	Year	Intervention, daily dose, n	Regimen	Target of the medicine	Comparison, daily dose, n	Condition	Publication status	Overall risk of bias
Katz et al. [52]	2011	Tanezumab 200 μg/kg, 88	Single dose, i.v.	Anti-NGF	Naproxen 1000 mg, 89 Placebo, 43	CLBP	Journal article published	High
2011-002143-95 [49]a	2012	ABT-110 5 mg, 0 10 mg, 0 20 mg, 0 30 mg, 0	—	Anti-NGF	Naproxen 500 mg, 0 Placebo, 1	CLBP	Terminated based on sponsor decision	—
Kivitz et al. [51]	2013	Tanezumab: 5 mg, 233 10 mg, 298 20 mg, 297	2 doses (8 weeks apart), i.v.	Anti-NGF	Naproxen 1000 mg, 298 Placebo, 233	CLBP	Journal article published	High
Tiseo et al. [57]	2014	Fasinumab: 0.1 mg/kg, 54 0.3 mg/kg, 54	Single-dose, s.c.	Anti-NGF	Placebo, 51	Sciatica (subacute)	Journal article published	High
Rolan et al. [60]	2015	BG00010: 50 μg/kg, 3 100 μg/kg, 3 200 μg/kg, 6 400 μg/kg, 3 800 μg/kg, 3	Single dose, s.c. (50 μg/kg), i.v. (all others)	Pro-GDNF	Placebo, 12	Sciatica (no duration specified)	Journal article published	Moderate
Okkerse et al. [50]	2016	BG00010: 50 μg/kg, 3 150 μg/kg, 3 400 μg/kg, 3 800 μg/kg, 3	3 doses (1 week apart), i.v.	Pro-GDNF	Placebo, 4	Sciatica (chronic)	Journal article published	High
Sanga et al. [59]	2016	Fulranumab: 1 mg, 77 3 mg, 77 6 mg, 78, 10 mg, 79	3 doses (4 weeks apart), s.c.	Anti-NGF	Placebo, 78	CLBP	Journal article published	Moderate
Backonja et al. [58]	2017	BG00010: 50 μg/kg, 38 150 μg/kg, 13 400 μg/kg, 16 800 μg/kg, 20 1200 μg/kg, 41	3 dosesover 1 week, i.v.	Pro-GDNF	Placebo, 48	Sciatica (chronic)	Journal article published	High
NCT03285646 [55]	2019	Fasinumab 3 mg, 31	4 doses (4 weeks apart), s.c.	Anti-NGF	Placebo, 32	CLBP	Completed with results posted in the trial registry	High
NCT02725411 [53]a	2019	Tanezumab: 5 mg, 92 10 mg, 93	7 doses (8 weeks apart), s.c.	Anti-NGF	Celecoxib 200 mg, 92	CLBP	Completed with results posted in the trial registry	High
Markman et al. [54]	2020	Tanezumab: 5 mg, 407 10 mg, 408	3 doses (8 weeks apart), s.c.	Anti-NGF	Tramadol 100–300 mg, 610 Placebo, 407	CLBP	Journal article published	High
Dakin et al. [56]	2021	Fasinumab: 6 mg, 141 9 mg, 140 9 mg, 141	4 doses (4 weeks apart), s.c. (6 mg and 9 mg), i.v. (9 mg)	Anti-NGF	Placebo, 141	CLBP	Journal article published	High

a

Studies not included in the meta-analyses. CLBP: chronic low back pain; GDNF: glial cell line-derived neurotrophic factor; n: number of participants randomized to the intervention or control; NGF: nerve growth factor.

Risk of bias and certainty of evidence

We assessed 11 trials for overall risk of bias: nine trials were classified as high risk of bias and two trials as moderate risk of bias (Table  1). One trial did not provide sufficient data for the risk of bias assessment [49]. In the risk-of-bias domains, one trial [50] did not blind care-providers or outcome assessors to interventions, six trials [51–56] had high losses to follow-up, and two trials [51, 57] had selective reporting. There was ‘unclear’ risk of bias of random sequence generation (n = 4) [51, 52, 55, 57], allocation concealment (n = 6) [50–52, 55, 57, 58], care-provider blinding (n = 1) [59], missing data (n = 1) [50], intention-to-treat analysis (n = 7) [50, 53–57, 60], similarity at baseline (n = 1) [50], use of co-interventions (n = 3) [50, 55, 58] and treatment compliance (n = 6) [50–52, 54, 55, 58]. We rated all studies as ‘high’ risk of bias in the ‘other bias’ domain due to pharmaceutical company sponsorship (Supplementary Table S2, available at Rheumatology online).

The GRADE assessment of confidence in the evidence for each analysis is summarized in Supplementary Table S3, available at Rheumatology online.

Efficacy

Chronic low back pain

Eight trials evaluated the efficacy of medicines that target neurotrophic factors for chronic LBP. Of these, six trials [51, 52, 54–56, 59] were included in the meta-analyses comparing the effect of anti-NGF to placebo. Low certainty evidence showed that anti-NGF medicines (fasinumab, fulranumab and tanezumab) may have a small and unimportant effect on pain intensity for chronic LBP at 4 weeks (mean difference [MD] −6.75, 95% CI: −8.61, −4.90; six trials, 14 comparisons, 3370 participants) and 12 weeks (MD −6.16, 95% CI: −8.38, −3.94; six trials, 14 comparisons, 3098 participants). Very low and low certainty evidence shows that fasinumab (ranging from 3 to 9 mg, four doses, and 4 weeks apart) and tanezumab (ranging from 5 to 20 mg, one to seven doses, and 1–8 weeks apart) may reduce pain intensity, respectively, at 4 and 12 weeks (Fig.  2). The results for disability are reported in Supplementary Table S4, available at Rheumatology online.

Fig. 2

Effect of medicines that target neurotrophic factors compared with placebo on pain intensity (0–100 scale) for chronic LBP

RE: random effects

Sciatica

Fours trials [50, 57, 58, 60] evaluated the efficacy of medicines that target neurotrophic factors (anti-NGF and pro-GDNF) for sciatica (two chronic sciatica, one acute sciatica, one did not specify the pain duration for sciatica). The overall effect of these medicines may have little to no effect on pain intensity at 4 weeks (very low evidence, MD −1.40, 95% CI: −8.26, 5.46; three trials, 11 comparisons, 339 participants) and 12 weeks (very low evidence, MD −2.91, 95% CI: −13.69, 7.67; two trials, nine comparisons, 44 participants) (Fig.  3). The results for leg pain and disability are reported in Supplementary Table S4, available at Rheumatology online.

Fig. 3

Effect of medicines that target neurotrophic factors compared with placebo on pain intensity (0–100 scale) for sciatica

GDNF: glial cell line-derived neurotrophic factor. RE: random effects. Negative values for mean difference indicate effect favors drug compared to placebo. NA means data not available.

Safety and harm

Low back pain

Evidence from six trials (14 comparisons and 3412 participants) [51, 52, 54–56, 59] showed that anti-NGF medicines may increase the odds of adverse effects (OR 1.18, 95% CI: 1.01, 1.38).

The subgroup analysis by anti-NGF medicine type suggests that fasinumab may not increase adverse effects at 4 weeks (low certainty evidence, OR 0.86, 95% CI: 0.58, 1.29; two trials, four comparisons, 620 participants). Low certainty evidence showed that fulranumab and tanezumab may increase the odds of adverse effects (fulranumab: low certainty evidence, OR 1.92, 95% CI: 1.06, 3.46; one trial, three comparisons, 388 participants; tanezumab: OR 1.20, 95% CI: 1.00, 1.44; three trials, six comparisons, 2406 participants) (Supplementary Fig. S1, available at Rheumatology online). Across all trials for chronic LBP, the most commonly reported adverse effects were arthralgia, headache and sensory alterations (paresthesia and hypoesthesia).

Low certainty evidence showed that anti-NGF medicines do not increase the odds of serious adverse effects (harm) for chronic LBP (OR 0.76, 95% CI: 0.51, 1.14; six trials [51, 52, 54, 56, 59] and 14 comparisons, 3412 participants).

Sciatica

Evidence from four trials (16 comparisons and 379 participants) [50, 57, 58, 60] showed that medicines that target neurotrophic factors may increase the odds of adverse effects (low certainty evidence, OR 3.27, 95% CI: 1.78, 6.00). The subgroup analyses suggest that anti-NGF (fasinumab) may not increase adverse effects (very low certainty evidence, OR 1.72, 95% CI: 0.74, 3.99; one trial, two comparisons, 157 participants). In contrast, pro-GDNF increased the odds of adverse effects for sciatica (very low certainty evidence, OR 5.32, 95% CI: 2.63, 10.77; three trials, 14 comparisons, 222 participants) (Supplementary Fig. S2, available at Rheumatology online). The most commonly reported adverse effects for pro-GDNF were headache, feeling hot and pruritis, reported in all trials.

Overall, medicines that target neurotrophic factors may not increase the odds of serious adverse effects (harm) for sciatica (very low certainty evidence, OR 0.44, 95% CI: 0.18, 1.06; four trials [50, 57, 58, 60], 16 comparisons, 379 participants), but the evidence is very uncertain.

Sensitivity analyses

For sciatica, anti-NGF (fasinumab) still had no effect on pain at 4 weeks for sciatica (MD 1.39, 95% CI: −7.86, 10.64; one trial, two comparisons, 157 participants) after removing three small sample studies (⁠$\le$10 participants in the intervention or placebo group of the study arm included in the analysis) [50, 58, 60], and medicines targeting neurotrophic factors were no longer associated with increased odds for adverse effects (OR 1.72, 95% CI: 0.74, 3.99; one trial, two comparisons, 157 participants). For chronic LBP, there was no difference in the interpretation of the findings for pain intensity at 12 weeks (MD −6.10, 95% CI: −8.34 to −3.87; five studies, 13 comparisons, 3077 participants), but increased the odds for adverse effects (OR 1.20, 95% CI: 1.02, 1.40; five studies, 13 comparisons, 3349 participants) after removing a small sample trial [55]. We reported the potential impact of a new trial on the effect estimate for pain intensity in Supplementary Fig. S5, available at Rheumatology online.

Exploratory analyses: dose–response effect

Since there is no clear definition of dose levels for anti-NGF medicines for low back pain, we arbitrarily defined ‘high’ (≥9 mg) and ‘low’ (<9 mg) doses based on the median doses used in the included studies. The 95% CI for higher doses of anti-NGF crossed the threshold for clinical importance (low certainty evidence, MD −8.56, 95% CI: −11.09 to −6.03; five trials, seven comparisons, 1930 participants) (Supplementary Fig. S3, available at Rheumatology online). Higher doses of anti-NGF may increase the odds of adverse effects (low certainty evidence, OR 1.23, 95% CI: 1.00, 1.52; five trials, seven comparisons, 1953 participants), but lower doses may not increase the odds of adverse effects (low certainty evidence, OR 1.11, 95% CI: 0.87, 1.41; five trials, seven comparisons, 1459 participants) (Supplementary Fig. S4, available at Rheumatology online).

Discussion

We conducted a meta-analysis to evaluate the efficacy and safety of medicines that target neurotrophic factors for low back pain or sciatica. The overall analgesic effect for chronic LBP is small, and not clinically meaningful. However, low certainty evidence suggested higher doses of anti-NGF may produce clinically meaningful reductions in pain intensity, and very low certainty evidence showed that fasinumab may offer a clinically meaningful reduction in pain for up to 3 months. The evidence for sciatica is very uncertain; anti-NGF and pro-GDNF may have little to no effect on pain intensity.

In terms of safety for chronic LBP, the risk difference was 1.03 (95% CI: 1.00, 1.07) against anti-NGF medicines, representing approximately three more people per 100 experiencing at least one adverse effect. Adverse effects may occur less often with fasinumab and lower doses of anti-NGF medicines. For sciatica, the risk difference for pro-GDNF medicines was 1.27 (95% CI: 1.13, 1.43) against the medicine, representing around 27 more people per 100 experiencing at least one adverse effect.

Strengths and weaknesses of the study

We registered this review prospectively and reported the study in line with PRISMA recommendations [28]. We conducted a comprehensive literature search including data from both published and unpublished trials. We used the Cochrane risk of bias tool [33] and evaluated confidence in the evidence using the GRADE system [36]. We had a high-rate response to data requests (five out of seven data requests), and we did not need to impute data in our analyses. Our review included a large number of participants with chronic LBP (>3000 participants). Finally, this is the first review that included separate doses of medicines designed to target neurotrophic factors for LBP or sciatica. This review also has some limitations. We included trials with small samples and different pain durations for sciatica, which increases the uncertainty of our results in this population. Also, we relied on the definition of adverse and serious adverse effects as reported from the included trials, which could have varied between trials.

Comparing the findings with previous systematic reviews

The most recent systematic review [23] evaluating the effects of anti-NGF medicines on chronic LBP (three trials, 2109 participants) found very low certainty evidence that anti-NGF medicines may have clinically important analgesic effects on chronic LBP with a standardized mean difference (SMD) of −0.29 (95% CI: −0.58, 0.00) (SMD >0.5 is equivalent to >10 points on a 0–100 scale [39]). The subgroup analysis of that review showed that tanezumab may have a moderate analgesic effect with a SMD of −0.44 (95% CI: −0.81, −0.06) [23] (CI spanning 20 points on a 0–100 scale [39]). By including three recent studies (contributing an additional 1303 participants to the analyses), our review demonstrated that anti-NGF medicines may have a small but not clinically important reduction in pain intensity for chronic LBP, and that tanezumab does not differ substantially from the overall effect of anti-NGF medicines.

Comparing the findings with other medicines for LBP

Anti-NGF medicines may provide comparable analgesic effects to medicines recommended in clinical practice guidelines for LBP and sciatica [11, 61]. Duloxetine, a commonly prescribed antidepressant, provides small, clinically unimportant reductions in pain intensity (low certainty evidence, MD −5.87 points on a 0–100 scale, 95% CI: −7.88, 3.86) for chronic LBP [62]. Opioids may have small, clinically important analgesic effects for acute and chronic LBP (very low certainty evidence, MD −8.98 points, 95% CI: −11.71, −6.25) [63]. For sciatica, anti-inflammatory drugs, certain types of antidepressants, muscle relaxants and opioid analgesics do not reduce pain intensity (very low certainty evidence) [64–66]. Only epidural corticosteroids seem to offer a small, short-term leg pain reduction for sciatica (moderate certainty evidence, MD −4.93 points, 95% CI: −8.77, −1.09) [67].

All medicines for LBP or sciatica are associated with different non-serious adverse effects in people with LBP. Opioids may be associated with nausea, constipation and sedation [68], and anti-NGF medicines seem to be associated with joint and peripheral sensory events in patients with OA [25]. In our review, the overall incidence of people with joint abnormalities (i.e. arthralgia, OA, periarthritis, rapidly progressive OA) was 6.6% in the placebo group and 9.9% in the anti-NGF group across all seven studies [51, 52, 54–57, 59] included in the meta-analyses. Four trials [51, 54–56, 59] investigated the presence of rapid progression of OA (RPO). We estimated that 0.10% of participants in the placebo group and 1.35% of participants in the anti-NGF group were diagnosed with RPO during the follow-up of the trials. The incidence of sensory abnormalities (i.e. hyperesthesia, allodynia, burning sensation, dysesthesia, hyperpathia, neuralgia, neuritis, pallanesthesia, paraesthesia, peripheral sensory neuropathy, sensory disturbance, sensory loss, polyneuropathy) was 2.8% in the placebo group and 8.1% in the anti-NGF group.

Implications for clinicians, researchers and policymakers

Medicines that target neurotrophic factors for LBP and sciatica are in clinical development and have not yet received regulatory approval for this indication; therefore, clinicians are not permitted to use these medicines in their clinical practices. This review provides evidence for the FDA and other regulatory agencies to consider whether anti-NGF should be licensed for LBP.

In general, the FDA approves medicines that demonstrate sufficient evidence of efficacy and safety (benefit–risk relationship) from at least two large and well-designed RCTs (phase 3 trials) [69], but it is unclear how regulatory agencies would interpret the findings for LBP. We showed that most of the analgesic effects for chronic pain are small and unimportant for chronic LBP, and there are still concerns of potential exacerbation of joint degeneration with the use of anti-NGF medicines [25]. However, we showed that there are avenues to explore the analgesic effects of anti-NGF medicines for LBP and sciatica in future trials. Higher doses and the anti-NGF type (fasinumab) are associated with greater reductions in chronic LBP. Anti-NGF medicines do not seem to be associated with the common adverse effects of opioids, such as addiction, misuse and dependence [25]. Future studies, independent from pharmaceutical companies should be conducted to reduce the risk of bias of anti-NGF studies. We encourage researchers to conduct high-quality systematic reviews of medicines that are soon to be submitted for approval to increase transparency in the regulatory process of medicines for a particular condition.

Acknowledgements

R.R.N.R., M.A.W. and J.H.M. conceived the idea for the project. R.R.N.R., M.C.F., M.A.W., A.G.C., H.B.L., E.T.O., M.D.J., S.M.G., A.J.M., R.C. and J.H.M. contributed to the project design and protocol development. R.R.N.R. conducted the search. R.R.N.R., M.C.F., M.A.W., A.G.C., H.B.L., E.T.O. and M.D.J. conducted the study selection, data extraction and quality appraisal. R.R.N.R., M.C.F. and M.A.W. analysed the data. R.R.N.R. and J.H.M. had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. R.R.N.R. wrote the first draft of the manuscript. All authors provided substantive feedback on the manuscript and have read and approved the final version. The corresponding author (the manuscript’s guarantor) attests that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted. R.R.N.R. is supported by the University of New South Wales School of Medical Sciences Postgraduate Research Scholarship and a NeuRA Ph.D. Candidature Supplementary Scholarship. M.C.F. is supported by an Australian Medical Research Future Fund Grant GNTID1170205. M.A.W. is supported by a Postgraduate Scholarship from the National Health and Medical Research Council of Australia, a School of Medical Sciences Top-Up Scholarship from the University of New South Wales, and a Ph.D. Supplementary Top-Up Scholarship from Neuroscience Research Australia. A.G.C. is supported by the University of New South Wales Prince of Wales Clinical School Postgraduate Research Scholarship and a NeuRA Ph.D. Candidature Supplementary Scholarship. H.B.L. is supported by Australian Government post-graduate award. E.T.O. is supported by an Australian government research training program scholarship and a NeuRA Ph.D. Candidature Supplementary Scholarship. M.D.J. receives a salary from the University of New South Wales. S.M.G. is supported by the Rebecca L. Cooper Medical Research Foundation. J.H.M. receives project funding support from the National Health and Medical Research Council and the Medical Research Future Fund of Australia.

Funding: No specific funding was received from any bodies in the public, commercial or not-for-profit sectors to carry out the work described in this article.

Disclosure statement: The authors have declared no conflicts of interest.

Data availability statement

The dataset used and analysed during this study and the accompanying code are available from the corresponding author upon reasonable request.

Supplementary data

Supplementary data are available at Rheumatology online.