Rotator cuff rupture is a common condition that may compromise shoulder function, health status1, and patient-assessed quality of life2,3. When conservative treatment fails to relieve the symptoms associated with a torn rotator cuff, operative intervention is considered. Rotator cuff repair typically provides satisfactory results including decreased shoulder pain and improved shoulder motion4. Unfortunately, imaging studies have demonstrated a retear rate of 30% to 94% after arthroscopic single-row repair5,6, with a higher failure rate among patients with a massive tear. Three prospective series involving double-row repair indicated retear rates of 11%7, 15%8, and 17%9 at a mean of two years postoperatively. Furthermore, it was observed that the retear rate was greater in patients with larger tears9. Recent biomechanical studies demonstrated that, although double-row repair had a high ultimate load to failure, an increased number of failures occurred at the musculotendinous junction or where the medial suture anchors had been placed10,11. Rotator cuff failure is attributed to age and intrinsic overloading rather than to extrinsic impingement12,13.
The use of an autologous platelet-leukocyte gel or membrane is a new method that has been proposed for stimulating and accelerating healing of soft tissue and bone14-18. Platelets can be retrieved and isolated from freshly drawn autologous whole blood. Point-of-care devices fractionate the blood into platelet-poor plasma, platelet-rich plasma, and red blood cells19. The platelet-rich plasma can then be activated by autologous thrombin to create a viscous solution. This may be applied exogenously to form a mass, termed platelet-leukocyte membrane, that delivers platelet growth factors that may stimulate and/or accelerate physiological wound-healing and reparative tissue processes20,21.
The purpose of this study was to evaluate the clinical and magnetic resonance imaging (MRI) results of single-row arthroscopic rotator cuff repair with and without platelet-leukocyte membrane in patients with a large posterosuperior rotator cuff tear.
The study was registered in the ISRCTN public trials registry (number ISRCTN93082180). Eighty consecutive patients with a large full-thickness posterosuperior rotator cuff tear who met the inclusion criteria were enrolled between June 2009 and December 2009. All patients agreed to participate in the study and signed an informed consent form in accordance with the Declaration of Helsinki. The ethics committee of our institution approved all procedures described in this study.
The rotator cuff tear in each patient was confirmed preoperatively with noncontrast MRI scans (repetition time, 3200 ms; echo time, 85 ms). Oblique sagittal, oblique coronal, and axial MRI scans were obtained. We utilized T2-weighted gradient-echo spectral presaturation inversion recovery (SPIR) sequences in the true axial scans and T1-weighted gradient-echo SPIR sequences in the oblique coronal plane (parallel to the course of the supraspinatus muscle) and the oblique sagittal plane (parallel to the glenoid articular surface). The MRI findings represented the principal criteria that were used prior to surgery to determine that patients met the inclusion criteria for the study. A series of radiographs (true anteroposterior, axillary, and outlet views) was used to document signs of degenerative arthritis.
The criteria for inclusion in the study group were the presence of a reparable large full-thickness posterosuperior rotator cuff tear and the ability of the patient to complete the preoperative and postoperative MRI examinations. We excluded patients with a partial-thickness tear, a small or massive full-thickness tear, a subscapularis tear, a traumatic tear, biceps instability, labral pathology amenable to surgical repair, os acromiale, degenerative arthritis of the glenohumeral joint, autoimmune or rheumatologic disease, previous surgery in the same shoulder, and Workers’ Compensation claims. Fatty degeneration of the rotator cuff muscles was documented on MRI and classified as described by Goutallier et al.22 and Fuchs et al.23 (grade 0, no fatty infiltration; grade 1, some fatty streaks; grade 2, more muscle than fat; grade 3, as much muscle as fat; grade 4, less muscle than fat).
Each patient’s suitability for inclusion was confirmed once the dimension of the rotator cuff tear had been verified at the time of arthroscopy. We consider a large full-thickness tear to encompass the entire supraspinatus, have minimal retraction of the tendon ends, and measure 2 to 4 cm (equivalent to type CII or CIII as described by Snyder24). Such crescent-shaped tears are relatively short in relation to their width (having a medial-to-lateral length that is less than the anterior-to-posterior width). The mean time from the onset of symptoms to the arthroscopic repair was eight months (range, two to fourteen months).
Clinical evaluations were performed by two of the authors who had expertise in shoulder surgery but were not directly involved in the repair surgery. The evaluations were performed preoperatively and at a mean of thirteen months (range, twelve to fourteen months) postoperatively. The Constant score25 and the Simple Shoulder Test (SST) score26 were also assessed. The maximum strength of the affected arm (in pounds) after five seconds of contraction was measured (microFET2 Digital Handheld Muscle Tester; Hoggan Health Industries, West Jordan, Utah). Three strength measurements were made at 90° of elevation in the scapular plane and averaged.
MRI scans were acquired preoperatively and twelve months after repair. All scans were evaluated independently by two radiologists who were musculoskeletal specialists and who had no knowledge of the patient’s clinical information or surgical history. Disagreements were discussed in a consensus meeting at which the scans were reevaluated and a final decision was reached27. The images were used for structural and qualitative assessment of the rotator cuff, and repair integrity was determined according to the classification described by Sugaya et al.9,28. This classification distinguishes five repair categories with the use of oblique coronal, oblique sagittal, and transverse T2-weighted images. Type I indicates a repaired rotator cuff that has sufficient thickness with homogeneously low intensity on each image; type II, sufficient thickness with a partial high-intensity area; type III, insufficient thickness without discontinuity; type IV, the presence of a minor discontinuity in more than one slice of each image, suggestive of a small tear; and type V, the presence of a major discontinuity on each image, suggestive of a large tear.
All surgical procedures were performed with the patient in the beach-chair position under general anesthesia and an interscalene block. A standard arthroscopic pump was used and standard posterior, lateral, anterolateral, and midglenoid portals were established to perform a thorough diagnostic examination. After the intra-articular evaluation, the arthroscope was placed in the subacromial space. The subacromial bursa was removed to gain a clear view of the rotator cuff tear.
After the tear had been debrided, the patient was treated either with or without platelet-leukocyte membrane. A randomization list was generated for the treatment assignment, and a sealed envelope contained forty cards on which “with—Group I” was printed and another forty cards on which “without—Group II” was printed. Three days prior to surgery, patients were randomized and assigned to one of the two groups of forty patients each. Seven patients were excluded from the study during the arthroscopic evaluation; five were found to have a massive rotator cuff tear (two tendons involved, type CIV) and two had a small tear (<1 cm, type CI). These patients’ cards were put back in the sealed envelope and another seven consecutive patients were recruited for the study.
All patients in group I underwent blood tests (including platelet count). Platelet-rich plasma was obtained from 10 cc of peripheral blood after slow centrifugation at 120 ×g for ten minutes (RegenKit; Regen Lab, Le Mont-Sur-Lausanne, Switzerland). This platelet-rich plasma was added to calcium gluconate and batroxobin, and the mixture was centrifuged at high speed (≥1500 ×g) for twenty to thirty minutes to obtain a round membrane with a diameter of 13 mm and a thickness of 3 to 4 mm (Fig. 1). This membrane is thin, elastic, malleable, and deformable because of its fibrin matrix. Platelets and leucocytes are trapped in the fibrin matrix, and for this reason growth factors are gradually released in situ to provide a presumably greater therapeutic effect. The concentration of platelets was considered to be similar in each membrane because the platelet concentration in each patient in group I was within the normal range. In an ultrastructural investigation conducted prior to the trial, we observed that a membrane with dimensions equal to those of the membranes in the trial contained high concentrations of white blood cells (7 × 103/mm3, primarily lymphocytes and monocytes) and platelets (>400 × 103/mm3, 1.7 times greater than the normal level in whole blood).
All of the surgical procedures were performed in a standardized manner by two surgeons using the same randomization list. The footprint area on the greater tuberosity was prepared with use of a full-radius resector until bleeding in this area was observed. TWINFIX Ti 5.0 mm Suture Anchors with number-2 preloaded ULTRABRAID sutures (Smith & Nephew, Andover, Massachusetts) were placed within the footprint of the rotator cuff. Each suture was passed through the tendon approximately 15 mm medial to the tear margin. After the sutures had been placed, they were sequentially tied in a simple configuration with a sliding SMC (Samsung Medical Center) knot29 followed by three alternating half-hitches.
If the membrane was being used, the post suture was introduced inside the membrane before passage of the suture throughout the lateral edge of the tendon. In all cases, the rubber diaphragm of the 8-mm cannula was momentarily removed and the pump was stopped to facilitate the passage of the membrane inside the cannula (Fig. 2). In this manner, the membrane was positioned between the abraded footprint area and the lateral edge of the rotator cuff so that it could not be displaced or moved during the successive phases of the rotator cuff suturing (Fig. 3). One membrane was utilized for each anchor. The total number of anchors used was two or three (mean, 2.63). The mean extra cost due to the use of the membrane was approximately $90 per operation.
The mean operative time was eleven minutes longer in group I. Tenotomy of the long head of the biceps tendon was performed in sixteen patients (seven in group I and nine in group II); all of these patients had biceps tendinitis with tendon fraying. Subacromial decompression was performed in all patients.
Postoperatively, all patients used a sling with the shoulder internally rotated. Passive shoulder motion was initiated under supervision during the first postoperative week. The sling was removed at four weeks postoperatively and active-assisted motion was started at that time (except in two patients who started at six weeks because they missed a follow-up visit). Full active shoulder motion was allowed at six to eight weeks postoperatively, and strengthening exercises were initiated at fifteen weeks.
Statistical Analysis
The primary outcomes were the difference between the preoperative and postoperative Constant scores and the repair integrity assessed by MRI according to the Sugaya classification. Comparisons between the two groups were performed with use of the Student t test for quantitative variables and the chi-square test for qualitative variables. For the chi-square test, classes with a frequency of less than five were grouped with other classes. The variables analyzed were age, sex, side dominance, fatty degeneration, tenotomy of the long head of the biceps tendon, Constant score, and SST score. An accompanying confidence interval (CI) was calculated for the difference in the change in the Constant score.
Two-way analysis of covariance (ANCOVA) was used to investigate whether the differences in preoperative and postoperative Constant pain subscores between the two groups could be explained by differences in age. The pain subscore was the dependent variable, the group (treatment with or without membrane) was the independent variable, and age was the covariate. The chi-square test was used to test the hypothesis that the use of platelet-leukocyte membrane provided superior clinical results with regard to repair integrity. The Student t test was used to compare patients with each type of repair integrity in the two groups to evaluate whether differences in repair integrity were influenced by age.
A p value of 0.05 was considered significant. Statistical analyses were performed with use of SAS software (SAS Institute, Cary, North Carolina). The Cohen kappa coefficient was used to assess interobserver agreement with regard to rotator cuff repair integrity on MRI; a kappa value of >0.8 indicates almost perfect agreement.
Source of Funding
The authors did not receive any outside funding or grants in support of their research or preparation of this work.
Patient flow through the stages of the study is depicted in Figure 4. Four patients (one in group I and three in group II) were lost to follow-up at one year for reasons that are not known. The remaining seventy-six patients, forty-one men and thirty-five women, underwent the final evaluation. The mean patient age was sixty-one years (range, forty-six to seventy-one years). The mean age was sixty years (range, fifty-three to seventy years) in group I and sixty-three years (range, forty-six to seventy-one years) in group II.
The improvement in the Constant score after surgery did not differ significantly between the two groups (difference = 0.4 [95% CI, −2.6 to 3.4], p = 0.73). The only variables that differed significantly between the two groups were the preoperative Constant score (difference = −4.2 [95% CI, −6.0 to −2.4], p < 0.01), postoperative Constant score (difference = −3.7 [95% CI, −6.4 to −1.0], p = 0.01), and age (p = 0.02) (Table I).
The differences in the preoperative and the postoperative Constant score can be explained by the shoulder pain level, which differed significantly between the two groups at both time points (Table I). In fact, preoperative and postoperative Constant scores recalculated without inclusion of the pain domain did not differ significantly between the two groups. The decrease in pain level after surgery did not differ significantly between the two groups (difference = −0.01 [95% CI, −0.8 to 0.7], p = 0.9). The ANCOVA results showed that the interaction of age and treatment was not significant (p = 0.8 preoperatively and p = 0.7 postoperatively), and we therefore assessed the effect of treatment on the pain score controlling for age. The main effect of treatment was significant (p < 0.01 preoperatively and postoperatively), but the main effect of age was not significant (p = 0.23 preoperatively and p = 0.26 postoperatively).
MRI scans acquired at a mean of thirteen months postoperatively demonstrated that the repairs in group I were type I in twenty-three patients (59%), type II in eleven (28%), type III in five (13%), and type IV or V in none. The repairs in group II were type I in thirteen patients (35%), type II in eleven (30%), type III in ten (27%), type IV in one (3%), and type V in two (5%) (Table II). The distribution of repair types differed significantly between the two groups (p = 0.04, chi-square = 6.29 [degrees of freedom = 2]).
As noted above, the only patients with a postoperative retear were in group II. The ages of these three patients were fifty-six, fifty-eight, and sixty-seven years (mean, 60.3 years) compared with 63.4 years for the remaining patients in group II. The mean Constant score of these three patients (64) was lower than the mean value in group I (77.9) and in the remaining patients in group II (75.1).
The mean age of the patients with type-I and type-III repair integrity did not differ significantly between the two groups (p = 0.92 and 0.33, respectively), but a significant difference was observed for the patients with type-II repairs (p < 0.0001) (Table III). (The test was not applied to type-IV and V repairs since no patient in group I had a rotator cuff retear.) The interobserver agreement for rotator cuff repair integrity assessed on MRI was almost perfect (Cohen kappa = 0.92).
Rotator cuff retears may occur after open or arthroscopic repair. Galatz et al.5 reported a retear rate as high as 94%. Failures have been attributed to a variety of causes, including poor quality of the repaired tendon, pullout of the suture anchors, suture breakage, and rehabilitation that was too early or inappropriate30-33. Liem et al.34 observed that a higher degree of muscle atrophy and fatty infiltration preoperatively were associated with recurrence of the tear as well as with progression of fatty infiltration and an inferior clinical result. Biomechanical studies have demonstrated that double-row tendon-to-bone fixation provided superior initial fixation strength and footprint coverage compared with the single-row technique35-37; however, the retear rate remains high even with the double-row technique, particularly for large and massive tears9. Rotator cuff retears were not eliminated with the introduction of new arthroscopic techniques. The “transosseous-equivalent” suture-bridge technique, which led to improved contact pressures compared with the double-row technique38,39, has not eliminated rotator cuff retears. In fact, Cho et al.40 evaluated the clinical results and MRI evaluations after arthroscopic single-row and suture-bridge repair of rotator cuff tears and observed that, compared with the single-row technique, the suture-bridge technique tended to yield better preservation of the rotator cuff tissue repaired at its insertion site but also increased the percentage of retears that were located at the musculotendinous junction.
The use of autologously prepared platelet-leukocyte membrane is a new method that may stimulate and/or accelerate healing of soft tissue and bone. The inside of the platelet cytoplasm contains α granules and dense granules. The α granules contain platelet growth factors such as transforming growth factor β (TGF-β), basic fibroblast growth factor (bFGF), platelet-derived growth factor (PDGF-AB), epidermal growth factor (EGF), vascular endothelial growth factor (VEGF), and connective tissue growth factors (CTGF). All of these growth factors except bFGF and PDGF stimulate and increase angiogenesis; furthermore, almost all stimulate proliferation of undifferentiated mesenchymal cells and stimulate or regulate mitosis in fibroblasts37. Therefore, we used the platelet-leukocyte membrane to facilitate tendon-healing to the footprint area on the greater tuberosity, where bleeding had previously been induced.
Platelet-leukocyte gel has already been utilized for treatment of tendon disease. Mishra and Pavelko16 obtained excellent results, including both pain relief and functional improvement, in the treatment of chronic elbow tendinitis using this gel. Aspenberg and Virchenko15 used the gel in a rat model of Achilles tendon injury and observed an approximately 30% increase in tensile strength and stiffness of the repair at one week compared with control animals. Everts et al.41 hypothesized that enhancement of healing by platelet-leukocyte gel might be explained by elevation of the concentration of VEGF, which is released from platelets and stimulates and promotes angiogenesis. This hypothesis agrees with Anitua et al.42, who affirmed that blood supply to the injured tendon had improved.
In our study, the use of platelet-leukocyte membrane during arthroscopic single-row repair improved repair integrity, as determined by MRI, compared with single-row repair without the use of membrane. Although the duration of follow-up in the present study was not long, no cases of retear were noted in the group in which the membrane had been used. We were surprised to observe such a low rate of rotator cuff retears. However, we believe that this could be attributed to the short duration of follow-up and the fact that our study did not enroll patients with massive tears. Nevertheless, the improved integrity of the repaired tendon in the group treated with membrane, as indicated by the MRI evaluations, was not associated with an improved functional outcome. In fact, the two groups would have had similar preoperative and postoperative Constant scores if the pain component had been excluded. (Exclusion of the pain component from the comparison is justifiable because the intensity of the pain was significantly higher in group II even before the arthroscopic treatment.)
Although the patient ages in the two groups differed significantly, we demonstrated that the difference in pain was not explained by age. The difference in patient age between the two groups was significant only for patients with type-II repair integrity. However, the number of these patients was the same in both groups; therefore, since the total number of participants in both groups was almost the same, this age difference would also not account for the overall difference in repair integrity between the two groups. Consequently, the effect of the treatment on repair integrity in the present study does not appear to have been influenced by the age difference between the two groups.
Retears occurred only in three patients in group II whose mean age was younger that that of the remaining patients in the same group. Therefore, although no statistical analysis could be carried out because of the small number of cases, the difference in retear rate between the groups also does not appear to be attributable to age differences.
Castricini et al.43 recently performed a randomized study on the use of platelet-rich plasma augmentation for arthroscopic rotator cuff repair and observed no significant difference in the overall Constant score between the groups treated with and without the membrane. The only difference that was noted involved alterations in MRI signal intensity in the tendon; such alterations were more common in the group in which the membrane had not been used. The authors concluded that the data in their study did not support the use of the platelet-rich fibrin matrix during repair of a rotator cuff tear to improve the healing of the rotator cuff. However, the study by Castricini et al. differs from ours in several respects: (1) many patients with a small (<1-cm) tear were included, (2) only one membrane was used for each patient (including a patient who had a medium rotator cuff tear and therefore required more than one anchor), (3) all rotator cuff repairs were performed with use of a double-row technique, and (4) the duration of immobilization after surgery was shorter. We believe that the first two points represent the main differences between the two studies.
Randelli et al.44 obtained results similar to ours in a study involving the use of platelet-rich plasma—in combination with an autologous thrombin component—in patients with a variety of types of rotator cuff tears. The spray applicator kit loaded with syringes of platelet-rich plasma and thrombin was positioned between the bone and the repaired rotator cuff. The study showed that autologous platelet-rich plasma reduced pain in the first postoperative months and suggested that it also positively affected cuff rotator healing when the tear exposed the humeral head but did not retract to the glenoid articular surface. Therefore, Randelli’s study and ours suggest that platelet-rich plasma applied in the form of either a spray or membrane may promote rotator cuff healing.
In conclusion, the use of platelet-leukocyte membrane led to a slight improvement, as assessed by MRI, in the repair integrity of large tears involving the supraspinatus tendon, although this improvement was not associated with a better objective functional outcome. Nevertheless, many aspects of the use of this membrane still have to be elucidated; further research is necessary to determine (1) whether the membrane is absorbed, and how much time this requires, (2) whether the membrane prevents or only slows down the development of rotator cuff retears, and (3) whether passing the post suture through the membrane and covering it with the rotator cuff is sufficient to maintain the membrane in situ. The primary limitation of this study is the relatively short duration of follow-up, and intermediate-term follow-up is needed.