The efficacy and safety of platelet-rich fibrin for rotator cuff tears: a meta-analysis

Two reviewers searched the Pubmed, Embase, Web of Science, and Cochrane library independently (Xiu-hua Mao and Ye-jun Zhan) from inception to April 2018. The following keywords and Mesh terms were used for searching: “rotator cuff,” “rotator cuff tears,” “Rotator Cuff Injuries”[Mesh] “platelet rich fibrin,” “platelet rich,” “PRF,” “platelet rich fibril matrix,” “PRFM,” and “Platelet-Rich Fibrin”[Mesh]. Publication language was restricted to English. Reference list in systematic review or meta-analysis was also manually searched to avoid omitting any relevant studies.

The inclusion criteria were as follows:

Participant (P): arthroscopic rotator cuff repair as regards the age and sex.

Intervention (I): administration with PRF as the intervention group.

Comparison (C): placebo or saline as the control group.

Outcomes (O): re-tear rate, American Shoulder and Elbow Surgeons scale (ASES), University of California at Los Angeles scale (UCLA), Constant score, and adverse effect.

Study (S): only RCTs were included in this meta-analysis.

We used Review Stata 12.0 to perform statistical analysis. For continuous variables, we used the weighted mean difference, whereas for those categorical dichotomous, we used relative risks (RR) to analyze, and 95% confidence intervals (CI) were reported in analysis of both continuous and dichotomous variables. P value beneath 0.05 was considered to be statistically significant. Homogeneity was tested by the Q statistic (significance level at P = 0.10) and the I² statistic (significance level at I² = 50%). A random-effects model was used if the Q or I² statistic was significant; otherwise, a fixed-effects model was used. Subgroup analysis were performed in the analysis of re-tear rate according to the operative technique (single row or double row), risk of bias (low or unclear/high), volume of PRP (< 5 or ≥ 5 ml), follow-up duration (< 15 or ≥ 15 months), and size of rotator cuff tears (small-medium or large-massive). Sensitivity analysis was performed based on omitting one study in turn to investigate the influence of a single study on the overall RR estimates. Publication bias was not performed because the included studies were less than ten.

Details of study identification, screening, and selection are given in Fig. 1. Firstly, we retrieved 320 relevant reports from electronic databases. And 114 papers were removed by Endnote software for duplications. Thus, 206 papers were screened for the next step. Then, according to the inclusion criteria, 198 records were excluded. Finally, eight RCTs [15‐22] involving 364 patients (PRF = 177, control = 187) finally met the predetermined inclusion criteria and were included for final analysis.

We summarized the general characteristics of all the included studies and listed in Table 1. All of the included studies were published from the year 2011. Three studies were originated from the USA, two from Spain, two from Italy, and one from France. Only one study did not report the tear size, and the rest of the studies all reported the tear size. Mean age of the included patients ranged from 55.2 to 66. Sample size ranged from 14 to 43 with a total of 364 patients. Follow-up duration ranged from 12 to 27.2 months.

Figures 2 and 3 present the risk of bias summary and risk of graph respectively. Six studies reported the random sequence generation and one with high risk of bias. Five studies were with low risk of bias, and two were with unclear risk of bias.

Our meta-analysis comprehensively and systematically reviewed the current available literature and found that (1) PRF compared with placebo did not significantly reduced re-tear rate for rotator cuff tear patients; the evidence of the re-tear rate was consistent in most subgroup analyses and was confirmed by TSA; (2) PRF has no benefit on the shoulder function at the final follow-up when compared with placebo; (3) PRF was not associated with an increase of the complications than the control group.

Only one relevant meta-analysis on the topic has been published [23]. Several differences between ours and the previous ones should be noted. First, the previous ones mixed PRP and PRF in the same intervention group and thus cause large heterogeneity across the studies. Second, two studies were not included in the previous meta-analysis and the publication bias was inevitable. Andia et al. [24] conducted a review about the PRP therapy for tendinopathy, plantar fasciopathy, and muscle injuries. Results showed that PRP therapies were useless. Meanwhile, Andia et al. [25] revealed that PRP has no effects on muscle injury and tendinopathy.

The current meta-analysis systematically scanned all of the available studies and has given a relative credible evidence for the clinical effects of PRF on rotator cuff tear patients. In this meta-analysis, we identified re-tear rate as the primary outcome. Results showed that PRF has a negative effect on the overall incidence of re-tear at the final follow-up. Previous meta-analysis did not pool this important outcome [23]. Re-tear could make the patients dissatisfied and increase additional costs. Subgroup analysis indicated that PRF has a positive role in reducing the incidence of re-tear rate than the control group. However, long-term effects of PRF were extremely important for clinical administration.

Hueley et al. [23] conducted a meta-analysis, and the pooled result was similar with our meta-analysis. PRF is considered as one kind of platelet concentrates, and its molecular structure with low thrombin concentration is an optimal matrix for migration of endothelial cells and fibroblasts, which can progressively release several cytokines to help fibrin matrix remodeling.

In an animal experiment, we found that PRF has a beneficial role in tissue regeneration whereas there was a negative role in a clinical experiment [19]. Randelli et al. [26] reported that autologous PRP reduced pain in the first postoperative months and affected cuff rotator healing for both grade 1 and 2 tears. Furthermore, Andia et al. [27] revealed that PRP, as an autologous biotechnology product, has a positive effect on experimental tendon healing.

The reason for the failure of PRF to fulfill its promise remains unclear. There are some possible interpretations for this phenomenon. On the one hand, patients all received autologous source PRF and the growth factors contained in PRF vary from person to person, for which there were much more difficulty for experimentally bias control. To be specific, there is a chance that the patient’s blood plasma contains excessive TGF-β, and its potential effect on exuberant fibrosis may affect the therapeutic effect of PRF. Nevertheless, some patients’ plasma may contain abundant inflammatory mediators, which could adversely affect healing process. More importantly, none of us has enough data to determine the best clinical usage of PRF products. And there were some prior articles that noticed this problem [27]. On the other hand, platelet-rich products may also influence the effect.

For example, recent studies showed that not all separation systems yield a similar product, because there are many factors that can influence the separation, including the volume of blood, single- versus double-spin cycles, centrifuge rates, the need for an activator, white blood cell concentrations, and the final platelet and growth factor concentrations. In other words, different products can have varied platelet concentrations, and therefore, platelet-derived growth factor concentrations may differ between various systems [28]. Additionally, it is also possible that the clot may occupy the space between the tendon and bone, resulting in a gap. Once the material dissolves, they may inhibit the healing process.

Moreover, although patients all received autologous source PRF, these procedures are not absolutely safe. Some postoperative complications seem to be related with PRF. The most common one is infection. Even though it is performed with aseptic techniques, the PRF group has a higher infection rate than the control group [29]. The cause of infection is unclear, but multiple steps obliged to prepare PRFM require additional interactions between sterile and non-sterile fields and introduce variables, increasing infection risk. However, we did not find a significant difference in postoperative complication between the two groups in our meta-analysis.

Several limitations also existed in this meta-analysis: (1) initial tear size was not compared between the PRF and control group; (2) PRF volume, platelet concentration, and activating agent were different in the included studies, and thus, clinical heterogeneity was large in the outcomes; (3) the follow-up period varied among included studies, and thus, clinical effects of PRF in the same follow-up period need to be further confirmed; (4) sample size was relatively small in the included studies, and thus, high quality with large-scale sample RCTs were needed.