Regenerative Injections Including 5% Dextrose and Platelet-Rich Plasma for the Treatment of Carpal Tunnel Syndrome: A Systematic Review and Network Meta-Analysis

This network meta-analysis aimed to integrate the available direct and indirect evidence on regenerative injections—including 5% dextrose (D5W) and platelet-rich plasma (PRP)—for the treatment of carpal tunnel syndrome (CTS). Literature reports comparing D5W and PRP injections with non-surgical managements of CTS were systematically reviewed. The main outcome was the standardized mean difference (SMD) of the symptom severity and functional status scales of the Boston Carpal Tunnel Syndrome Questionnaire at three months after injections. Ranking probabilities of the SMD of each treatment were acquired by using simulation. Ten studies with 497 patients and comparing five treatments (D5W, PRP, splinting, corticosteroid, and normal saline) were included. The results of the simulation of rank probabilities showed that D5W injection was likely to be the best treatment, followed by PRP injection, in terms of clinical effectiveness in providing symptom relief. With respect to functional improvement, splinting ranked higher than PRP and D5W injections. Lastly, corticosteroid and saline injections were consistently ranked fourth and fifth in terms of therapeutic effects on symptom severity and functional status. D5W and PRP injections are more effective than splinting and corticosteroid or saline injection for relieving the symptoms of CTS. Compared with splinting, D5W and PRP injections do not provide better functional recovery. More studies investigating the long-term effectiveness of regenerative injections in CTS are needed in the future.


Introduction
As the most common peripheral entrapment neuropathy, carpal tunnel syndrome (CTS) affects approximately 3.8% of the general population [1], with an incidence of around 300 cases per 100,000 person-years [2]. CTS leads to clinical symptoms including pain, numbness, tingling sensation, and weakness in the hand innervated by the median nerve. A recent observational study showed that painful symptoms were more likely to develop near the carpal tunnel, whereas non-painful sensory disturbances were identified to distribute more peripherally in patients with CTS [3]. The risk We summarized the characteristics of the included studies in Tables 1 and 2. This meta-analysis included 497 patients with 518 affected wrists. The majority (> 75%) of the overall participants were women. The average age in the patient group across the included trials ranged from 36.6 to 60.4 years. The average symptom duration ranged from 13.7 to 72.0 months.

Assessment of the Quality of the Included Studies
We summarized the quality assessment in Figures 1 and 2. In most of the included studies, the item found to have a high risk of bias was blinding of participants and personnel. The aforementioned procedure was not applicable in studies in which splinting was used in the comparative arms. Likewise, for the PRP trials, the patients were aware of which treatment they received if no sham blood withdrawal was used in the control injection groups.   On the other hand, selective reporting was revealed to have a low risk of bias in the majority of the enrolled trials. Almost all of the studies reported the outcome by using acceptable clinical assessment tools, together with electrophysiological examinations.

Comparison of Standardized Mean Differences (SMDs) of SSS among Different Treatments
A forest plot of pairwise meta-analysis for the SMDs between the different therapeutic regimens is presented in Figure 2. For the network meta-analysis, the network graph is shown in Figure S3, and its forest plot is demonstrated in Figure 3. The league tables for pairwise and network meta-analyses of SSS and functional status scale (FSS) are summarized in Tables S1 and S2, respectively. In the pairwise meta-analysis (Figure 2), PRP injection was superior to corticosteroid and saline injections, with SMDs of 0.71 (95% confidence interval [CI], 0.62 to 0.80; I 2 = 93.0%) and 1.05 (95% CI, 0.92 to 1.18), respectively. The difference of SMD between PRP injection and splinting was 0.01, with no significance (95% CI, −0.10 to 0.11; I 2 = 98.8%). D5W was better than corticosteroid and saline injections, with SMDs of 0.67 (95% CI, 0.51 to 0.82) and 1.11 (95% CI, 0.94 to 1.29), respectively. There was a significant difference of PRP injection vs. D5W injection, with a SMD of −0.48 (95% CI, −0.63 to −0.34).
Ranking of the treatment effect for SSS is presented in Figure 4. The results of the simulation of rank probabilities showed that D5W injection had the highest probability (99.8%) to be the best treatment, whereas PRP injection (52.8%) had the largest chance to be the second best treatment. Likewise, splinting, corticosteroid injection, and saline injection were likely to be the third best (with a probability of 52.8%), fourth best (with a probability of 100%), and the least effective (with a probability of 100%) treatments, respectively.
In the network meta-analysis, the SMDs comparing PRP injection with splinting, corticosteroid injection, and D5W injection were The ranking of the treatment effect according to FSS is presented in Figure 4. The results of the simulation of rank probabilities showed that splinting had the highest probability (99.5%) to be the best treatment. PRP, D5W, corticosteroid, and saline injections were likely to be the second best (with a probability of 99.5%), third best (with a probability of 100.0%), fourth best (with a probability of 100.0%), and least effective (with a probability of 100.0%) treatments, respectively.

Publication Bias
The funnel plots for SSS and FSS ( Figure 5) revealed a symmetric distribution of the inter-group comparisons of the SMDs. The Egger's test did not reveal a significantly small study bias (p = 0.386 for SSS and 0.692 for FSS, respectively).

Discussion
This meta-analysis incorporated direct and indirect evidence from relevant clinical trials, yielding several important findings on regenerative injections in CTS. First, considering the effectiveness of symptom relief, D5W injection was likely to be the best alternative, followed by PRP injection. Second, splinting ranked higher than PRP and D5W injections in terms of functional recovery. Third, corticosteroid and saline injections ranked fourth and fifth, respectively, with respect to clinical effectiveness in providing symptom and function improvement.
Several meta-analyses have investigated the effects of injection therapies for CTS. Most of them focused on comparisons of various local injection techniques instead of on different regimens. In 2007, a systematic review [14] including 671 participants from 12 studies concluded that corticosteroid injection provided short-term symptom relief in contrast to placebo injection. In 2015, a network meta-analysis [35] with 633 patients from 10 studies compared the effectiveness of corticosteroid injections, using different injection approaches. Their analysis indicated that the ultrasound-guided in-plane injection method through the ulnar aspect of the wrist was the best approach in CTS treatment. In 2018, another meta-analysis [36] including 181 hands from three randomized trials attempted to determine whether ultrasound guidance provided better clinical outcomes. The authors reported that ultrasound-guided corticosteroid injection was superior to the landmark-guided approach for the improvement of symptom severity but not functional status and electrophysiological findings. While regenerative medicine is an emerging field in the treatment of musculoskeletal diseases, only one systematic review [37] proposed PRP injection to be a potential alternative treatment for mild to moderate CTS. Nevertheless, there is lack of quantitative analysis incorporating evidence from multiple studies to demonstrate the effectiveness of regenerative injection for CTS.
In our network meta-analysis, the simulation of therapeutic effectiveness in terms of symptom relief revealed D5W and PRP injections to be the best and second best treatment, respectively. Although little statistical significance was observed in the network comparisons of SMDs between regenerative injections and other treatments, the trends of effect sizes were in accordance with those in the pairwise comparisons (Table S1). Low-concentration dextrose had been first studied as a solution for the hydro-location of the needle tip during regional anesthesia, revealing no decrease of block effect with additional use of D5W [38]. Later, several case reports were published on the potential utility of D5W injection in entrapment neuropathies [39,40]. Although there is insufficient evidence from animal and histopathological studies for the exact beneficial mechanism of D5W, a commonly accepted pathway is through the inhibition of transient receptor potential vanilloid receptor-1 (TRPV1) [41]. The TRPV1 receptor can trigger nociceptive sensation by evoking an action potential from the peripheral nerve endings, and is susceptible to inflammatory stimuli [41]. Our results might also reflect the crucial influence of neurogenic inflammation on CTS symptoms.
On the basis of our review, the evidence on PRP injection for CTS was more abundant than that on D5W injection. PRP, which contains high concentrations of autologous growth factors, is commonly used to accelerate tissue repair in various kinds of musculoskeletal disorders [42,43]. Several in vitro and animal studies have revealed the potential of PRP in remyelination, axonal regeneration, and angiogenesis of injured peripheral nerves [44]. Our quantitative analysis further demonstrated a favorable outcome of PRP for symptom relief in mild to moderate CTS. The main concern about PRP application in CTS treatments lies on how it is prepared and which portion of plasma has been extracted for injection. Unlike D5W, which contains a fixed concentration of dextrose, the final platelet concentration in PRP varies across patients and different preparation systems. Further, because increased expression of inflammatory cytokines (e.g., transforming growth factor-β) is commonly identified in endothelial cells and subsynovial connective tissues of the median nerves in patients with CTS [45], injection of leukocyte-rich PRP may potentiate neurogenic inflammation and worsen the symptoms. However, among our retrieved trials, the obtained information was not detailed enough to resolve the aforementioned concern.
There was one finding worth mentioning concerning the comparative effectiveness of symptom relief between D5W and PRP injections. The volume of D5W (3-5 mL) was higher than that of PRP (1-3 mL) in our included studies. Additionally, ultrasound guidance was used in the two trials of D5W, but not in all studies using PRP. A cadaveric study has demonstrated that a bolus saline injection could effectively reduce the peak gliding resistance of the median nerve [46]. Another randomized controlled trial pointed out that precise hydro-dissection of the median nerve using saline under ultrasound guidance yielded better clinical outcomes than subcutaneous saline injection [47]. Therefore, the observed superiority of D5W over PRP (with respect to CTS symptoms) in this meta-analysis may be partly derived from a higher injection volume and the mechanical effect of nerve hydro-dissection guided by ultrasound imaging.
Concerning the functional status of patients with CTS, our analysis revealed that splinting is the best treatment, followed by PRP injection. The mechanism of the effect of splinting lies in reducing the intra-tunnel pressure through the neutral positioning of the wrist. Although splinting is not likely to provide short-term (e.g., one month) benefits compared with no treatment, its long-term effect has rarely been studied [8]. Regenerative injections may cause a rapid decrease of symptom severity; however, functional improvement still depends on the recovery of muscle endurance and strength, which requires adequate rest and proper wrist positioning, with the use of a splint. Further, although PRP injection ranked second in effectiveness in terms of functional improvement, the patients in the two included PRP studies [23,29] had additional splinting after the injection. An earlier randomized controlled trial has revealed that splinting combined with corticosteroid injection was superior to corticosteroid injection alone, in terms of functional recovery and symptom relief in CTS [48]. Therefore, it is reasonable to speculate that post-injection splinting contributed to the functional outcome in the PRP group.
The present meta-analysis has several clinical implications. First, D5W and PRP injections can be considered useful regimens for the treatment of CTS of mild to moderate severity because both methods yield better effectiveness in terms of symptom relief and functional improvement than corticosteroid and saline injections. Second, the mechanical effect of hydro-dissection may substantially contribute to the benefits derived from D5W injection. Hence, injection of D5W with an enough volume (e.g., 5 mL) and under ultrasound guidance is recommended. Third, splinting provides proper positioning and adequate rest of the wrist in patients with CTS, and seems crucial for functional recovery. Post-injection splinting should be considered in combination with regenerative injections.
This meta-analysis has several limitations. First, because not every included study documented electrophysiological findings and ultrasonographic nerve cross-sectional area measurements, we were not able to compare the changes of those parameters after treatments. Second, the timing of outcome assessment varied, and the third month (or close to it) after injection was the most suitable point because of the availability of data for this time point in almost all included trials. However, the long-term (> 6 months) effectiveness of regenerative injection could not be analyzed in the present meta-analysis. Third, some of the enrolled studies, especially those investigating D5W injection, were from the same research group. The generalization of results might be limited by some systemic factors, such as ethnicity and disease severity. Therefore, we strongly believe that more studies are needed to examine whether similar efficacy can be observed in various patient populations. Fourth, all the patients enrolled in the present meta-analysis only had CTS of mild to moderate severity. Considering the regenerative potential of both proposed regimens (especially for PRP), it would be of great interest to apply them on severe cases or those with an unsatisfactory outcome after CTR in future studies.

Conclusions
D5W and PRP injections are more effective than splinting, corticosteroid injection, or saline injection for relieving the symptoms of CTS. D5W and PRP injections might not be superior to splinting in terms of functional recovery. More studies are needed in the future, to explore the long-term effectiveness of regenerative injections in the treatment of CTS.

Literature Search
This review was conducted in accordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines [49]. PubMed and Embase were used as the two main electronic databases for the literature search, which covered records from the earliest time to July 2019, without language limitation. The references of relevant articles and the registry of the clinical trials, such as ClinicalTrials.gov, were also manually searched for eligible studies. The following keywords were used in different combinations: "carpal tunnel syndrome", "platelet-rich plasma", and "dextrose". Two authors independently read the titles and abstracts of the retrieved literature and reviewed the full text of articles, potentially fulfilling our inclusion criteria. The search strategy is provided in Appendix A.

Inclusion and Exclusion Criteria
The inclusion criterion was being a clinical study comparing regenerative injections (D5W and PRP) with non-surgical treatments for CTS. The exclusion criteria were (1) being case reports/series, reviews, and non-human trials; (2) lacking quantitative outcome assessment; (3) lacking a control group; and (4) being studies on other peripheral nerve entrapment syndromes (other than CTS).

Data Extraction and Quality Assessment
Two authors independently evaluated the quality of the included trials by using the Cochrane risk-of-bias tool for randomized controlled trials [50]. Six aspects were evaluated, including (1) random sequence generation, (2) allocation concealment, (3) blinding of participants and personnel, (4) incomplete outcome data, (5) selective reporting, and (6) other biases. Each item is appraised as having a low, high, or unclear risk of bias. Any discrepancy of opinions between the two authors was resolved by discussion or through the decision of the corresponding author.

Data Synthesis and Analysis
The following data were retrieved from the included studies: first author, publication year, study design, sample size, disease severity, follow-up timing and duration, regenerative injection regimen, guiding technique, and details of the comparative treatments. The main outcome was the SMD of the scores in the BCTS before and after treatments [51]. The SSS and FSS were analyzed separately. If the study only used the VAS for pain for symptom assessment, we considered the VAS as a surrogate for the SSS. Because the follow-up timing and duration were different among the studies, we chose the assessments at baseline and at or close to the 3rd month post-injection, for analysis.
In the pairwise analysis, the random-effect model was chosen because the regenerative injection regimens and their counterpart treatments varied across the trials. The heterogeneity across the studies was expressed by using the I2 statistics and analyzed by using the Cochrane Q test [52]. A value of I2 > 50% was considered statistically significant heterogeneity [52]. For the network meta-analysis, a generalized linear mixed model was used to combine the direct and indirect comparisons. The loop inconsistency model was used to examine whether inconsistency existed between direct and indirect evidence [53]. Ranking probabilities of the SMD of each treatment were acquired by using simulation and presented by using ranking probability curves [53]. The funnel plot method and Egger's test were used to investigate the potential existence of publication bias [53], indicating that studies with larger sample sizes and significant results are more likely to be published than those with small participant numbers and insignificant outcomes. The statistical software package Stata (StataCorp. 2015. Stata Statistical Software: Release 14. StataCorp LP, College Station, TX, USA) was employed for all analyses, and p < 0.05 was considered statistically significant.
Supplementary Materials: The following are available online at http://www.mdpi.com/1424-8247/13/3/49/s1, Figure S1. Summary of the quality assessment of the included studies: green indicates low risk of bias; red, high risk of bias; and yellow, unclear risk of bias. Figure S2. Quality assessment graph: green indicates low risk of bias; red, high risk of bias; and yellow, unclear risk of bias. Figure S3. Network graphs for the comparison of the standardized mean difference between different subgroups in terms of the (A) symptom severity scale and (B) functional status scale of the Boston Carpal Tunnel Questionnaire. PRP, platelet-rich plasma; D5W, 5% dextrose.