FormalPara Key Points

Low to moderate quality randomized controlled trials show that PRP injections provide no superior, clinically relevant, effect on return to sport, recurrences, function, and pain for athletes with acute muscle injuries.

PRP treatment is a safe procedure with negligible adverse effects, but the large and indiscriminate use of PRP injections for the treatment of acute muscle injuries in clinical practice is not justified and PRP use should be limited to controlled trials until more positive evidence emerges and suggests otherwise.

1 Introduction

Muscle lesions account for one third of sport-related injuries and 92% of them affect the four major muscle groups of the lower limb: first and foremost hamstrings, followed by adductors, quadriceps, and calf muscles [1]. Taken as a whole, they are responsible for prolonged absence from sport, they may require long rehabilitation, and re-injury rates have been reported to be as high as 39% within the same season, thus representing a substantial problem for both players and their teams [1, 2].

Treatment goals are therefore to achieve the same functional level as prior to injury and to allow for the return to sports practice quickly and with minimal recurrence risk [3]. While PRICE (protection, rest, ice, compression, and elevation) is a commonly accepted overall approach to control the early inflammatory process, followed by rehabilitative exercises and gradual training therapy to recondition the injured structure, many other interventions are used to promote healing as well [4]. Among these, anti-inflammatory medications, electrotherapeutic modalities, hyperbaric oxygen therapies, photothermal therapy or injection strategies such as prolotherapy have been proposed. However, these treatments have no firm scientific basis and clinical evidence to support them is sparse, so they are mainly applied as empirical medicine due to the lack of indications from high-level trials [4,5,6].

The increasing knowledge of basic science related to muscle healing set the stage for the introduction of a novel biological approach to accelerate the healing of injured muscles. The early inflammatory phase is followed by repair and remodeling processes, where undifferentiated satellite cells proliferate and differentiate to restore the muscle structure, in response to several growth factors (GFs) [7]. Many of these bioactive molecules are stored in the platelet alpha granules, such as insulin-like GF (IGF-I), hepatocyte GF (HGF), fibroblast GF (FGF-2) and transforming GF (TGF-β1), which may be key regulators of muscle regeneration and myogenesis [8]. Therefore, platelet-rich plasma (PRP), which can be defined as a blood derivative with a higher platelet concentrate than whole blood, gained increasing interest to deliver a high concentration of autologous GFs and bioactive molecules in physiologic proportions, with low costs and in a minimally invasive way [9]. Under the acronym PRP, countless procedures have been developed, leading to heterogeneous concentrates in terms of absolute number of platelets, activation methods, presence or absence of white blood cells, etc., all factors that might influence the effect on the tissue target [10,11,12,13]. While some authors have attempted to introduce some order into this multifaceted field, the biological nature of this autologous approach makes standardization difficult. Classifications that endeavor to summarize procedures according to macro-categories (PRP with or without leukocytes, with or without fibrin, etc.) are useful whereas other more detailed classifications have proved more problematic in terms of practical applicability [12, 14]. Nonetheless, regardless of the heterogeneity and classifications limitations, preclinical evaluations suggested that the local delivery of PRP might reduce recovery time [8]. Thus, the application in clinical practice increased rapidly, prompted by an unmet clinical need with a large commercial market [15]. However, after early reports of positive preliminary experience, higher quality studies have recently questioned the real benefit provided by PRP injections to promote the healing process and return to sport.

The purpose of this meta-analysis was to evaluate whether PRP injections improve the outcomes of acute muscle injuries compared with standard rehabilitation, in terms of time of return to sport, as well as re-injuries, complications, pain, strength and flexibility recovery, functional scores, and imaging.

2 Methods

2.1 Search Strategy

A meta-analysis was conducted in accordance with the Preferred Reporting Items for Systematic reviews and Meta-Analysis (PRISMA) guidelines [16]. A systematic electronic search was performed on 1 November 2017 in the following databases: PubMed (MEDLINE), Cochrane Central Register of Controlled Trials (CENTRAL) and Web of Science electronic databases. The search period was 1987–2017; articles included in the meta-analysis were published between 2013 and 2016. The grey literature databases clinicaltrials.gov, who.int, greylit.org, opengrey.eu and isrctn.com, were searched as well for unpublished studies.

The search string was built as follows: “(PRP OR platelet-rich plasma OR plasma rich in growth factors OR platelet derived growth factor OR platelet derived OR platelet gel OR platelet concentrate OR PRF OR platelet rich fibrin OR ACP OR autologous conditioned plasma OR PRGF OR platelet lysate) AND (muscle injury)”. The electronic database search was supplemented by a manual search of the reference lists of included articles.

2.2 Article Selection

The eligibility of searched articles was assessed on the basis of prespecified inclusion criteria.

  • Articles published in peer-reviewed journals, unpublished studies.

  • Articles written in any language.

  • Randomized, controlled trials (RCTs) and quasi-RCTs investigating the effect of PRP in the treatment of acute muscle injuries.

  • At least one control group including patients treated with placebo injection or physical therapy.

  • The treatment of acute muscle injuries including upper and lower limbs.

All the criteria had to be fulfilled for the article to be included. Biomechanical, in vitro studies, review articles, surgical techniques, case reports, letters to the editor, and editorials were excluded. When two or more papers evaluated the same patient cohorts, the relevant data were extracted from each study, but only the study with a longer follow-up was considered in order to avoid data duplication. Two authors independently reviewed the title and abstract of each article from the systematic literature search. The full text of the article was obtained and evaluated if eligibility could not be assessed from the first screening. Any disagreements were resolved via a consensus discussion between the two reviewers and the senior author was consulted if the disagreement could not be resolved.

2.3 Outcome Measurements

The outcomes extracted from the included studies were the time to return to sport, the re-injury and complication rate, pain, strength, flexibility, and range of motion (ROM) and, finally, the healing process evaluated with ultrasound or magnetic resonance imaging (MRI).

2.4 Data Extraction and Synthesis

An electronic piloted form for data extraction was created prior to the study. Patient demographic details, including sex, age, type of muscle injury, and level of activity, were extracted. Details of study design, such as level of evidence, inclusion and exclusion criteria, type of randomization, blinding of patients or outcome assessors, injection protocol, PRP characteristics, and rehabilitation details, were collected. For the outcome measurements, the time needed to return to unrestricted sport activity, according to the criteria reported in each study, and the number of patients who experienced a complication or a re-injury, were extracted. Information regarding pain measured with subjective scales, strength measured with an isokinetic machine, handheld dynamometry or manual testing, and ROM/flexibility measured through clinical evaluation were extracted as well.

2.5 Assessment of Risk of Bias and Quality of Evidence

The risk of bias was evaluated according to the standardized Cochrane Database questionnaire [17]. Selection bias (random sequence generation and allocation concealment), performance bias (blinding of participants and personnel), detection bias (blinding of outcome assessment), attrition bias (incomplete outcome data), reporting bias (selective reporting) and other bias were rated as “high risk,” “low risk,” or “unclear risk.” Each item for each study was reported in a table and a summary table reporting the percentage of studies with a specific bias risk was produced. The performance and detection bias were evaluated twice: both for outcomes that could have been affected by the patient’s knowledge of the type of treatment (time to return to sport, pain and functional scores) and those that were considered independent from treatment knowledge (re-injuries, complications, strength, flexibility, and radiologic features). Reporting bias was assessed by checking the respective trial registrations, when available, to determine whether the outcomes reported in the trial protocol corresponded to the outcomes finally reported in the published paper. If trial registration protocols were not available, reporting bias was considered to be low risk if the final outcomes included key outcomes such as time to return to sport, complications and functional or pain evaluation; when this was not the case, a high risk of reporting bias was considered to be present. The purpose of the assessment of risk of bias was to provide a descriptive summary of the main sources of potential bias in the included studies. Articles were not excluded on the basis of the assessment.

The overall quality of evidence for each outcome was graded as “high,” “moderate,” “low,” or “very low,” according to the Grading of Recommendations Assessment, Development and Evaluation (GRADE) guidelines [18], based on study design, risk of bias, inconsistency, indirectness, imprecision, and publication bias. Based on the GRADE guidelines [18], the quality of evidence of each outcome was downgraded from the highest level in the case of: high risk of bias, high statistical heterogeneity of the outcome or conflicting results among the various studies, use of surrogate measurements (e.g., score for muscle function) or heterogeneous definitions for the same outcome (e.g., muscle healing with MRI or US), and CIs overlapping the null value or a total sample size lower than the calculated optimal information size (obtained with post hoc sample analysis using https://www.stat.ubc.ca/~rollin/stats/ssize/b2.html) [18].

2.6 Statistical Analysis

The meta-analysis was performed using RevMan V.5.0.18.33 (the Cochrane Collaboration, Copenhagen, Denmark). Continuous variables were extracted and analyzed as the mean ± standard deviation (SD) [19]. If the mean or SD were not reported, authors were contacted via e-mail in order to obtain raw data for the calculations. The mean difference (MD) and 95% confidence interval (CI) were calculated for the continuous variable of “time (days) to return to sport.” The risk difference (RD) and 95% CI were calculated for the dichotomous variables of re-injuries and complications. We tested for heterogeneity using the χ2 and Higgins’ I2 tests. Data were pooled using a Mantel–Haenszel random-effects model if the statistical heterogeneity was moderate to substantial (> 40% at I2 test) [19]; a fixed-effects model was used if the statistical heterogeneity was below 50% [20]. Publication bias was planned to be investigated with a funnel plot if more than ten studies were available for a specific outcome. A sensitivity analysis was performed separately analyzing studies with a single-blind or double-blind design and studies evaluating hamstring injuries or different injury location. Also, a meta regression evaluating sample size was performed in the case of more than 10 studies available. Due to extreme heterogeneity in the presentation of pain, strength, ROM/flexibility and imaging outcomes, a pure meta-analysis was not possible and the results of these outcomes were reported in a narrative manner. A p value of < 0.05 was considered statistically significant in all analyses.

3 Results

3.1 Article Selection

The initial search resulted in a total of 2181 articles. After the removal of papers not relevant to the purpose of this study, 104 papers were considered eligible for inclusion in this meta-analysis. Another 93 papers were removed because they did not meet the inclusion criteria and 11 RCTs were therefore identified. Three of these RCTs, two found in clinicaltrials.gov and one study abstract were excluded because the final results had already been published in peer-reviewed journals. No further papers were found in the other grey literature databases who.int, greylit.org, opengrey.eu, isrctn.com.

Eight articles were therefore included for qualitative synthesis [6, 15, 21,22,23,24,25,26] (Fig. 1). However, two studies [6, 21] involved the same patient population: one paper was the study protocol and the other contained the study results, but both were included since they presented non-mutually exclusive information. Another two studies [15, 22] evaluated the same patient cohort: one study reported the time to return to sport outcome, while the other reported the re-injury rate and functional outcomes and both these studies were therefore included, but they were regarded as a single study in the data presentation and statistical analysis. Finally, six studies were included in the quantitative synthesis and meta-analysis.

Fig. 1
figure 1

PRISMA flow chart for study inclusion. RCTs randomized controlled trials

Full size image

3.2 Study Characteristics

All six included studies were RCTs. The exclusion criteria were similar, especially regarding injury onset, the presence of non-steroidal anti-inflammatory drugs (NSAIDs) and the assumption or use of other injected therapies. All but one of the studies had two arms comparing PRP injections with isolated conventional physical therapy (n = 3), physical therapy and hematoma aspiration (n = 1) or physical therapy and isotonic saline injection (n = 1). The study by Hamilton et al. [23] had three arms, comparing PRP, platelet-poor plasma (PPP), and no injections. In the latter case, only the no injection arm was used as control group and accounted in the analysis. All studies reported to blind outcome assessors, but only two studies were considered to be double-blind (Table 1).

Table 1 Methodological characteristics of the included studies
Full size table

3.3 Patient Characteristics

Overall, 374 patients were enrolled in the six studies, ranging from 28 to 90 patients per study. However, only 145 who received a PRP injection and 191 who were included in the control group were finally included in the meta-analysis, due to a 0–14% rate of lost to follow-up. Patients were predominantly professional male athletes, with a mean age of under 30 years, except for Martinez-Zapata et al. [25] (45 years). Three studies exclusively comprised hamstring injuries [6, 15, 23], while the remaining three included several injury locations such as hamstrings, rectus femoris, quadriceps, gastrocnemius, thigh, foot and ankle, and shoulder. For the four studies that measured the injury dimension with ultrasound or MRI, no differences were reported between those treated with PRP and those in the control group at the baseline evaluation (Table 2).

Table 2 Patient characteristics of the included studies
Full size table

3.4 PRP Injection and Rehabilitation Protocols

The description of PRP type used was presented in detail in five of the six studies, while in the work by Bubnov et al. [24], no information was given to enable the possible replication of their PRP preparation. The injection was usually intralesional, under ultrasound guidance and within 5–7 days of injuries in four studies; one study reported a mean time interval of 14 days, while two studies did not report this detail. Only two studies performed hematoma aspiration, one of them used this practice to blind the patients to PRP or no injection treatment. The rehabilitation protocol was described or referenced satisfactorily in all studies, apart from the one by Bubnov et al. [24] (Table 3).

Table 3 Characteristics of the PRP preparations and injection protocols of the included studies
Full size table

3.5 Outcomes of PRP versus Control

3.5.1 Return to Sport

A clear definition of return to sport was reported in all six studies. The mean time to return to sport ranged from 10 to 42 days in the PRP group and from 22 to 42 days in the control group. Three studies reported a significantly shorter time to return to sport in patients treated with PRP and three studies reported no differences. A random-effect meta-analysis (I2 = 95%; p < 0.00001) revealed a significant mean difference of − 7.17 (95% CI − 12.26 to − 2.08; p = 0.006) in favor of PRP (Fig. 2).

Fig. 2
figure 2

Pooled mean difference for return to sport for PRP and control interventions. Square size indicates the size of the population investigated in each study; diamond is the estimated pooled effect: width indicates the 95% CI. Data using PPP as control were excluded from the analysis. PRP platelet rich plasma, CI confidence intervals, PPP platelet poor plasma, IV inverse variance, SD standard deviation

Full size image

However, considering only the two double-blind studies, the mean difference between PRP and the control group obtained with a fixed-effect meta-analysis (I2 = 0%; p = 0.71) was not significant (− 5.65, 95% CI − 12.14 to 0.84; p = 0.09) (Fig. 3a). A similar finding was also reported in the random-effect meta-analysis (I2 = 44%; p = 0.17) of the three studies evaluating only hamstring injuries (− 5.95, 95% CI − 12.48 to 0.57, p = 0.07) (Fig. 3b). On the other hand, a significant mean difference in favor of PRP was reported when considering the four single-blind studies or the three studies with heterogeneous muscle involvement (Table 4).

Fig. 3
figure 3

Sensitivity analysis, the mean difference for return to sport for PRP and control based on double-blind studies only (a) and studies including only hamstring injuries (b). Square size indicates the size of the population investigated in each study; diamond is the estimated pooled effect: width indicates the 95% CI. Data using PPP as control were excluded from the analysis. Risk of bias: (A) Random sequence generation (selection bias); (B) Allocation concealment (selection bias); (C) Blinding of participants and personnel for subjective outcomes or (D) objective outcomes (performance bias); (E) blinding of outcome assessment for subjective outcomes or (F) objective outcomes (detection bias); (G) Incomplete outcome data (attrition bias); (H) Selective reporting (reporting bias) and (I) Other bias. SD standard deviation, IV inverse variance, CI confidence intervals, PRP platelet rich plasma, PPP platelet poor plasma, RCT randomized controlled trials

Full size image
Table 4 Sensitivity analysis according to study patient blinding, muscle involved and control groups
Full size table

3.5.2 Re-injuries

A clear definition of re-injury was reported in four studies. The re-injury rate ranged from 0 to 27% in the PRP group and from 0 to 30% in the control group. The fixed-effect meta-analysis (I2 = 0%; p = 0.87) revealed a non-significant risk difference of − 0.03 (95% CI − 0.10 to 0.05; p = 0.50) (Fig. 4a).

Fig. 4
figure 4

Pooled risk difference for re-injury (a) and complications (b) for PRP and control interventions. Square size indicates the size of the population investigated in each study; diamond is the estimated pooled effect: width indicates the 95% CI. Data using PPP as control were excluded from the analysis. PRP platelet rich plasma, PPP platelet poor plasma, M-H Mantel–Haenszel, CI confidence intervals

Full size image

3.5.3 Complications

Only two studies reported at least one complication due to treatment (discomfort at the injection site, hematoma or hyperesthesia in the posterior thigh). The fixed-effect meta-analysis (I2 = 0%; p = 1.00) revealed a non-significant risk difference of 0.01 (95% CI − 0.05 to 0.06; p = 0.84) (Fig. 4b).

3.5.4 Pain

Five studies evaluated pain during rehabilitation. Two studies reported no differences, while three studies reported significantly better outcomes (p < 0.05) in patients treated with PRP, especially in the first 3 weeks (Table 5).

Table 5 Results of the different outcomes of each included study
Full size table

3.5.5 Strength

All four studies that evaluated muscle strength reported no differences between PRP and the control group at the final follow-up. Only Bubnov et al. [24] reported higher strength in patients treated with PRP but only during the first 2 weeks.

3.5.6 ROM/Flexibility

Only two studies evaluated ROM or flexibility. Reurink et al. [15] reported no differences, while Bubnov et al. [24] reported a higher ROM in patients treated with PRP at all the considered time points (1–4 weeks).

3.5.7 Functional Scores

Only Reurink et al. [15] used a subjective score (the Hamstring Outcome Score and Satisfaction) to assess muscle function, reporting no significant differences between patients treated with PRP or isotonic saline injection.

3.5.8 Imaging

No differences in muscle healing were reported both in the two studies that evaluated muscle edema reduction at MRI and in the two studies that evaluated ultrasonographic appearance (Table 5).

3.6 Study Quality and Risk of Bias

Since most of the studies described randomization and allocation concealment correctly, “selection bias” was considered as low risk. “Performance bias” was high risk for the time to return to sport, pain and functional scores, since three studies did not blind the patients to treatment. Conversely, performance bias was considered low risk when evaluating re-injury, complications, strength, flexibility, and radiologic features, since these outcomes were not likely affected by the knowledge of the treatment by the patients. Blinding of outcome assessors was reported in all except one study (in which blinding was unclear), resulting in an overall low risk of “detection bias” for all the outcomes. Both “attrition bias” and “reporting bias” were considered low risk due to the limited number of patients lost to follow-up and to the adherence of outcome reporting to study protocols (Fig. 5). Only Hamid et al. [6] reported a drop-out rate of 14% in both groups. Finally, a high risk of other bias was present in the study by Bubnov et al. [24], due to the inadequate comparison of patient populations and imprecise descriptions of PRP preparation, while the risk was unclear in the study by Martinez-Zapata et al. [25], due to the late PRP injection (mean 14 days) (Fig. 6).

Fig. 5
figure 5

Risk of bias summary table, with red as high risk of bias, green as low risk of bias and yellow as unclear risk of bias. aBias referred to outcomes that could have been affected by the patient’s knowledge of the type of treatment (time to return to sport, pain and functional scores). bBias referred to outcomes that were reputed independent from treatment knowledge of the type of treatment (time to return to sport, pain and functional scores outcomes)

Full size image
Fig. 6
figure 6

Risk of bias in each study, considered as high (red), low (green) or unclear (yellow). aBias referred to outcomes that could have been affected by the patient’s knowledge of the type of treatment (time to return to sport, pain and functional scores). bBias referred to outcomes that were reputed independent from treatment knowledge of the type of treatment (time to return to sport, pain and functional scores outcomes)

Full size image

Due to the presence of these biases and the limited number of studies and patients for most outcomes, the overall quality of evidence according to the GRADE guidelines for time to return to sport, pain, strength, flexibility, muscle function and muscle healing (through radiologic outcomes) was rated as low or very low (Fig. 7). Other issues that contributed to the lack of high evidence quality were the high statistical heterogeneity or the conflicting results in the studies and the indirect evaluation of muscle healing. However, if considering only double-blind studies, the evidence for the lack of significant difference in the time to return to sport could be considered of moderate quality, due to the lack of performance bias, and limited heterogeneity. Finally, due to the low risk of performance bias for re-injury and complication outcomes, the evidence of similar outcomes between PRP and control could be considered moderate as well.

Fig. 7
figure 7

Summary of findings table, with quality of evidence rated according to the GRADE guidelines. RCT randomized controlled trial, PRP platelet rich plasma, ROM range of motion, GRADE Grading of Recommendations Assessment, Development and Evaluation

Full size image

4 Discussion

The main finding of this meta-analysis is that the current evidence from the available studies with the highest quality does not support the hypothesis that the use of PRP injections promotes muscle healing and return to sport, due to the moderate quality of evidence of no significant differences between PRP and control group. Conversely, considering all the six RCTs included in the review, the time to return to sport was significantly shorter after PRP application. However, due to the bias in these studies analyzed overall, the heterogeneity of the findings and the limited sample size, the evidence of this finding should be regarded as very low quality according to the GRADE guidelines.

No statistically significant superiority for PRP in any of the other evaluated outcomes could be proven by this meta-analysis. Nonetheless, several aspects, which may influence the study results, still deserve to be discussed. The first is the quality of the studies themselves. In fact, among the selected RCTs, high quality variability could be detected. This is not of secondary importance, since a direct correlation between low meta-analysis quality and positive results has been demonstrated in the literature [27], as confirmed by this study. In fact, when considering all the RCTs, return to sport was statistically faster with PRP, but this result was not confirmed when the analysis excluded unblinded studies (Fig. 3). These findings do not currently justify the use of PRP, but they should still be interpreted as a demonstration of the need for a larger number of high-quality trials, rather than a definitive demonstration of lack of potential for the PRP strategy. In fact, despite the lack of significance for both blinded studies and studies evaluating only hamstring muscles, in both cases the confidence intervals of the mean difference between the groups contained both the null value and a value of 7 days which could be considered of important clinical relevance, especially for athletes. Therefore, this allows the possibility of different results in the future, when more high-quality studies enable the evaluation of a larger number of patients. In addition to the small number of patients documented to date, the current literature also presents other important limitations, such as the heterogeneity of the studies in terms of patients enrolled, lesion degree and location, type of PRP and administration protocol, as well as treatment timing and associated rehabilitation protocol.

Patients were included according to different criteria, with the mean age ranging from 20 to 46 years and their activity level ranging from professional athletes to non-competitive sport. Only three studies selected a specific muscle group, while the others included different muscle groups. Differences can also be found in lesion degree and chronicity, since the mean time from injury to PRP injection was up to 14 days. The latter is not a secondary consideration, since the timing of administration may dramatically change the effect of the applied GFs on the healing process. The muscle repair process involves two competing processes: the production of a connective tissue scar and the regeneration of disrupted muscle fibers [28]. Injury causes the disruption of muscle fibers and hemorrhage, followed by the extravasation of inflammatory cells leading to some degree of inflammation, which is required to remove necrotic muscle fibers and allow scar tissue to bridge the defect [29]. Activation of platelets also occurs early in the lesion area, where they help to orchestrate the healing process before the fibrosis phase begins. This occurs from 2 to 3 weeks after injury, when PRP could even be theoretically contraindicated, due to the presence of TGF-B, which may promote further fibrosis over regeneration [30]. Injections should therefore preferably be administered before this phase, as occurred in most but not all of the studies, although evidence that allows more accurate targeting of the window of benefit for PRP is still lacking. Analogously, the presence and amount of hematoma, its evacuation, the volume of PRP injected with respect to the damaged area due to its physical interaction with the disrupted fibers, as well as the containment of PRP in the injected area, are all factors that may jeopardize the local healing effect. In this context, activation of PRP is not always performed and some products rely on in situ, slow collagen-induced activation, which may influence molecular release and the biological effect on the one hand and the timing of gel formation on the other, which in turn may affect the physical interaction with the issue and the likelihood of confining GF secretion to the treated site [10].

One key element of heterogeneity, which is pivotal in PRP research in all fields, is the nature of the platelet concentrate itself. In fact, the abbreviation PRP is an umbrella for countless products, all differing in terms of a large number of variables, first and foremost cellularity and the number of platelets. A larger number of platelets may offer a higher concentration of GFs, which might help to recruit stem cells promoting a stronger healing response. On the other hand, the most locally effective dose of each GF and their combination is a long way from being understood and the scientific discussion relating to the optimal platelet concentration remains merely speculative. The same applies to leukocytes, monocytes, macrophages, and other cells that may be present in variable proportions according to the procedure used to obtain PRP. In particular, a great deal of attention is being focused on the possible negative effects of leukocytes, with in vitro studies showing deleterious effects in terms of inflammation and matrix molecule formation/degradation [11]. Nonetheless, in vitro studies may offer only a partial understanding of the role of these cells, especially in terms of muscle lesions, where some degree of inflammation is necessary in the early phase and where these cells may contribute to GF release and chemotaxis and, in the end, to the overall healing process [11, 29]. While the discussion relating to the best PRP formulation is still open, research efforts are focusing on a more targeted PRP use for muscle, such as the combined use of antifibrotic agents, where the rationale is to limit fibrosis in favor of the regeneration process, with preliminary promising findings [31].

Regardless of the PRP applied, the analysis of the available studies also revealed significant heterogeneity in terms of the mean time to return to sport between studies, which ranged from 10 to 42 days. This may be explained by the different patient populations that were treated, the different rehabilitation protocols applied, the involvement of a physical therapist and different criteria relating to the decision to return to unrestricted activities. Specific protocols have been shown to be able to influence the recovery time [32] and intensive rehabilitation remains a critical element for a successful RTS after muscle injury [23].

This study has some limitations. First of all, the limited number of studies and patients included in the meta-analysis clearly reduces the overall quality of its evidence and could possibly cause several outcomes not to reach statistical significance in the event of a true positive effect. However, the complex study design, with mandatory randomization and multiple blinding, only enables the performance of studies of this kind in high-volume selected centers in a multicentric fashion. Another important limitation that should be underlined, in addition to the heterogeneity of PRPs and populations, is the inhomogeneous reporting of the results. Even if we had been able to obtain missing mean and SD data from the authors through correspondence, use of a parametric test could be questioned due to the not-normal distribution of the time to return to sport. A meta-analysis of hazard ratios could have been more appropriate; however, it was not possible to obtain this outcome for all the studies or to extrapolate it from surrogate data. This could have introduced a bias in the statistical analysis that could have changed the significance of the results. Therefore, extreme care should be taken when interpreting the results of this meta-analysis, taking account of this statistical limitation. Moreover, the different methods that were used to investigate pain, strength, flexibility and muscle imaging did not enable statistical pooling of these outcomes. Finally, a limitation is represented by the lack of pre-registration to systematic review databases such as PROSPERO. However, while we acknowledge that this practice should be encouraged in future works (especially for the adherence of results to the pre-determined study protocol), the sub-group and sensitivity analysis in the present work were planned a priori.

Despite the aforementioned limitations, this meta-analysis still makes it possible to identify important findings. PRP treatment is a safe procedure, as the studies reported only negligible adverse effects, but, in terms of efficacy, the current literature does not support its use for muscle injuries. Its large and indiscriminate use in clinical practice is not justified and PRP injections should be reserved for controlled trials, until positive evidence emerges and suggests otherwise. Further studies are needed, both to strengthen the evaluation of the different PRPs and to investigate more appropriate delivery methods, as well as identifying the more responsive clinical targets. As shown for other PRP applications in the musculoskeletal system [33], PRP has a large placebo effect and future studies should ensure correct blinding procedures to understand its real potential. Homogeneous populations, satisfactory randomization, the precise description of PRP characteristics and the precise reporting of the results, possibly with several measurements of dispersion (standard deviation, standard error and confidence intervals), should be other mandatory features of future studies. Moreover, imaging evaluations of the repair tissue and clinical evaluations of patient symptoms and function should be documented both in the early stages, which are crucial for athletes aiming for a quick return to sport, and also in the long-term, to document whether tissue modifications may lead to a reduction in the recurrence of muscle injury.

5 Conclusion

Recent high-level trials with a moderate GRADE quality level of evidence do not confirm the promising biological rationale, the positive preclinical findings, and the successful early clinical experience with PRP injection. In fact, while the time to return to sport was significantly shorter after PRP application considering all the six RCTs included in the review with overall low-quality evidence, when considering only double-blind studies the evidence for the lack of significant difference could be considered of moderate quality. Therefore, any benefit in terms of clinical outcomes, return to sport, and recurrence using PRP injections for the treatment of acute muscle injuries is not supported by the available literature. PRP treatment showed negligible adverse effects, but its application in clinical practice is not justified, and PRP use should be limited to controlled trials until more positive evidence emerges and suggests otherwise.