- Email: [email protected]
Initial fixation and cyclic loading stability of knotless suture anchors for rotator cuff repair
Initial fixation and cyclic loading stability of knotless suture anchors for rotator cuff repair Barrett S. Brown, MD,a Andrew D. Cooper, MD,b Terence E. McIff, PhD,c Vincent H. Key, MD,c and E. Bruce Toby, MD,c New York, NY, Salt Lake City, UT, and Kansas City, KS
This study evaluated the resistance to gapping and the mode of failure for 2 knotless suture anchor systems used for rotator cuff repair compared with the performance of a conventional titanium anchor system. Eight matched pairs of fresh-frozen humeri were dissected free of all soft tissues and scanned to measure bone mineral density (BMD). The suture anchor systems tested were the TwinFix 5.0 Titanium (Smith & Nephew, Andover, MA), Bioknotless RC (DePuy Mitek, Norwood, MA), and Magnum (Opus Medical, San Juan Capistrano, CA), and each was inserted into each humerus. Cyclic, tensile loading was applied through the suture loop for 5000 cycles, or until failure, by using a servohydraulic testing machine. Gapping distances, defined as increasing elongation of the bone/anchor/suture system, were continuously measured. Total cycles to failure and mechanism of failure were documented. Mean initial (first cycle) and final (last cycle) gapping distances were 3.81 mm and 5.36 mm for the TwinFix 5.0, 4.02 mm and 5.34 mm for the Bioknotless RC, and 3.56 mm and 4.98 mm for the Magnum anchors. No significant difference was detected among mean gap openings (P > .05). However, the Bioknotless RC had more early failures (5) than the other 2 implants (1 each), approaching significance (P ¼ .07). Trials of the Bioknotless RC that did not fail early were found to have significantly less gap opening than the other 2 systems for both initial (1.89 mm vs 3.82 mm for the TwinFix 5.0 and 3.56
a
Sports Medicine and Shoulder Service, Hospital for Special Surgery; bComprehensive Orthopedic Specialists; and cDepartment of Orthopedic Surgery, University of Kansas Medical Center. Supported by a grant from the Marc A. and Elinor J. Asher Orthopedic Research Endowment. Reprint requests: E. Bruce Toby, MD, Peltier/Reckling Chairman, Department of Orthopedic Surgery, University of Kansas Medical Center, 3901 Rainbow Blvd, Kansas City, KS 66160 (E-mail: [email protected]). Copyright ª 2008 by Journal of Shoulder and Elbow Surgery Board of Trustees. 1058-2746/2008/$34.00 doi:10.1016/j.jse.2007.05.016
mm for the Magnum) and final (2.00 mm vs 4.68 mm for the TwinFix 5.0 and 4.24 mm for the Magnum) gap opening. BMD was a significant predictor of initial (P ¼ .029) and final (P ¼ .008) gap opening, whereas the site of anchor insertion was a significant predictor of final displacement. The Opus Magnum was comparable with a conventional suture anchor, but the Mitek Bioknotless RC showed a trend toward early failure. Biomechanical analysis of knotless suture anchor systems can demonstrate trends among implants in an experimental setting. Knowledge of these trends could influence implant selection. (J Shoulder Elbow Surg 2008;17:313-318.)
I
nitial fixation stability is an important variable in suture anchor repairs and can direct postoperative rehabilitation. Fixation failure after primary rotator cuff repair can occur and does so at the implant– bone, implant–suture, or suture–tendon interface.7 Suture anchor repair has been shown to be significantly stronger than bone fixation through transosseous tunnels under physiologic cyclical loading.5 The optimal location of suture anchor placement has been identified.10 Refinements in anatomic restoration of the rotator cuff are still evolving and have included the use of knotless suture anchors, which provide a potential advantage of minimizing gap formation while simultaneously eliminating the technical challenges associated with arthroscopic knot tying. Minimizing gap formation is important because failure to do so reduces contact between tendon and bone, which can adversely affect the healing of tendon to bone, both in the early (6-8 week) postoperative period as well as several months after repair.9 This dynamic biomechanical study compared initial fixation, under physiologic cyclical loading conditions, of 2 knotless anchors, the Magnum (Opus Medical, San Juan Capistrano, CA) and the Bioknotless RC (DePuy Mitek, Norwood, MA), with the TwinFix 5.0 Titanium (Smith & Nephew, Andover, MA) suture anchor, which has previously been identified as the metal suture anchor with the highest absolute pullout strength in cancellous porcine bone.2 Each of the anchors tested is shown in Figure 1.
313
314
Brown et al
J Shoulder Elbow Surg March/April 2008
Figure 1 The three suture anchors tested (from left to right) were the Opus Magnum (undeployed), the Mitek Bioknotless RC, and the Smith & Nephew TwinFix 5.0 Titanium.
MATERIALS AND METHODS Eight matched pairs of proximal humeri from fresh-frozen cadavers (3 women, 5 men), with an average age of 60.4 years (range, 26-89 years), were thawed, dissected free of all soft tissues, and thawed for 24 hours before testing. The humeri were transected 10 cm distal to the most proximal point of the humeral head. Dual energy x-ray absorptiometry (DEXA) was used to collect bone mineral density (BMD) values for each whole humerus in units of g/mm3. Each humeral diaphysis was transfixed with a 0.62 Kirschner wire and potted to a depth of 3 cm in dental cement. The rotator cuff footprint was gently de´brided to simulate in vivo preparation of the footprint. A total of 48 suture anchors, 16 of each type, were inserted in the proximal-anterior, proximal-middle, and proximal-posterior regions of the greater tuberosity according to the manufacturers’ guidelines (see Figure 2), as defined in a published study.10 One anchor from each of the 3 manufacturers was inserted into each humerus. The first suture anchor was chosen at random and was placed in the proximal-anterior region of the greater tuberosity, 5 mm posterior to the bicipital groove and approximately 5 mm medial to the tip of the greater tuberosity. The second was placed in the proximal-middle region of the tuberosity at least 10 mm posterior to the first implant. The third anchor was placed 10 mm posterior in the proximal-posterior region. Suture anchor placement was equally distributed across regions throughout all specimens. Each anchor was inserted according to the manufacturer’s specifications: the Opus Mangum perpendicular to the humerus, the other 2 anchors at 45 angles. After anchor insertion, each was preloaded by hand to verify that it was adequately secured before cyclic testing. The potted humerus specimens were secured to the base of a tensile testing machine in 45 of abduction relative to the loading forces (see Figure 3), representing the abduction angle that produces maximal in vivo tensile forces.3 Each corresponding suture was tied according to the pertinent manufacturer’s recommendations to a 5.3-mm-diameter eyebolt fixed to the loading actuator above. Each suture was tied while the eyebolt was in direct contact with the surface
Figure 2 This drawing demonstrates the regions within the greater tuberosity. All anchors in this study were placed in the proximal regions of the greater tuberosity, in equal distribution. Figure adapted with permission from Tingart et al.10
Figure 3 The testing setup used for our study. Specimens were fixed to the MTS Mini-Bionix machine using custom-made clamps, and the I loop was attached to the servohydraulic actuator.
of the bone at the anchor site. Manufacturer-supplied, nonabsorbable braided suture was used for each anchor. Each anchor used a single suture. The testing commenced with a single pretensioning load to 60 N, which was held for approximately 3 seconds. Then, cyclic loading (sine waveform under load control) commenced using a servohydraulic material testing system (Mini-Bionix, MTS Systems Corp, Eden Prairie, MN) from 10 to 60 N at 2 Hz. These values have been described as within physiologic ranges for rotator cuff loading.4,6 Relative displacement between the eyebolt to which the suture was tied and the stationary humerus was measured after initial pretensioning of each anchor and continuously thereafter during testing. Both initial and final displacement
Brown et al
J Shoulder Elbow Surg Volume 17, Number 2
measures included possible motion or slippage of the anchor, as well as breakage or elongation of the suture material. The anchors were loaded to an arbitrary cutoff of 5000 cycles. If an anchor reached 5000 cycles load to failure with 20 cycles at 60 N, 100 cycles in increasing increments of 30 N was performed until failure. The failure mode and amount of suture pullout of each anchor was documented, as was the number of cycles at which failure occurred. The cycle at which 50% failure (5-mm displacement) and 100% failure (10-mm displacement) occurred was extracted from the original data. The compliance of the potted humerus and fixtures was independently measured for each specimen and later subtracted from the total displacement curves recorded for each anchor test. Repeated measures analysis of variance (ANOVA) and Kruskal-Wallis ANOVA were used to test for treatment effects, after which the Tukey honestly significantly differences (HSD) test or Mann-Whitney test were performed to make specific post hoc comparisons between anchor treatments. Significance was set at P ¼ .05 for rejecting the null hypothesis that no difference in the distribution means exists. Variances were compared using the Levine test for homogeneity of variance. Correlations were presented using Pearson r correlation coefficients.
RESULTS Each anchor treatment group underwent 16 trials. One Mitek trial was unsuccessful owing to insertion failure and was excluded. The mean initial displacement was TwinFx 5.0, 3.82 mm; Bioknotless RC, 4.02 mm; and Magnum, 3.56 mm; the respective mean final displacement for each anchor was 5.37 mm, 5.34 mm,and 4.98 mm. These results are summarized in Table I. Anchors failed in one of several modes: suture breakage, anchor pullout, or severing of the suture by the bone. Of the Magnum anchors, 8 failed by suture breakage, 3 by anchor pullout, 3 by suture cutting, and 2 did not fail. For the Mitek anchor, 2 sutures broke, 11 anchors pulled out, 1 failed to insert, and 2 did not fail during testing. The Twinfix anchors failed 15 times by suture breakage, and 1 did not fail. When all included trial data were considered, no significant difference was detected among the mean displacement of the 3 anchor types (repeated measures ANOVA, P > .05). Nonparametric tests (Kruskal-Wallis ANOVA) also revealed no significant effect of anchor choice on initial or final displacement or on the number of trials reaching 50% or 100% failure levels. The Mitek anchors, however, had a significantly higher variance (Levine test, P < .0001) than the other anchors for both the initial and final displacement. This higher variance was primarily due to a larger number of early failures, where early failures were defined as tests where the anchor pulled out or failed before 5000 cycles. Comparisons of the number of early failures or 100% failure levels between Mitek (early failures ¼ 5) and the other 2 anchors reached nearly significant P values (Fischer exact, P ¼ .07).
315
Table I Initial and final displacements with all specimens included
Anchor
Initial, mean (SD) mm
Final, mean (SD) mm
Smith & Nephew Mitek Opus
3.82 (1.41) 4.02 (3.42) 3.56 (0.75)
5.37 (2.05) 5.34 (3.76) 4.98 (1.74)
Table II Initial and final displacements with early failures excluded
Anchor
Initial, mean (SD) mm
Final, mean (SD) mm
Smith & Nephew Mitek Opus
3.82 (1.41) 1.89 (1.80) 3.56 (0.75)
4.68 (2.09) 1.99 (2.10) 4.24 (1.60)
The mean initial and mean final displacements for all anchors, when early failures were excluded, are summarized in Table II. With early failures excluded (treated as missing values), the Mitek Bioknotless RC had significantly smaller initial and final displacements than the other 2 anchors (repeated measures ANOVA for initial displacement, P ¼ .939 with early failures and P < .001 without early failures; final displacement P ¼ .856 with early failures and P ¼ .003 without early failures.) Post hoc power analysis indicated that, with the number of specimens actually tested and the resulting variances seen in each treatment group, the statistical tests would have a 95% confidence of being able to detect a difference greater than 1.1 mm between the treatment groups. BMD, as measured by DEXA, was correlated negatively to both initial and final displacement (Pearson r ¼ –0.319 and –0.382, respectively). BMD was a significant predictor of both initial and final displacement, even with early failures included (P ¼ .029 and P ¼ .008, respectively). Figure 4 shows the distribution of BMD with respect to the final displacement recorded. The slope of the linear fit was –4.66 mm of final displacement per unit of BMD in g/cm3. Anchors were placed in 3 sites: anterior, mid, and posterior. The mean final displacements grouped by site are shown in Figure 5. The effect of site was examined using ANOVA, with site and anchor type as categoric independent variables and BMD as a continuous predictor variable. Site was significant (P ¼.029) for predicting final, but not initial, displacement (P ¼ .353). Significant differences in the group means for final displacement were found between anterior and posterior sites (P ¼ .011) and between mid and posterior sites (P ¼.043). If P values are adjusted for multiple comparisons, then only the difference between
316
Brown et al
Figure 4 Final displacement vs bone mineral density (BMD), which was shown to be a significant predictor of initial and final displacement (P ¼ .029 and P ¼ .008, respectively). The lines represent the linear fit and the 95% confidence interval for that linear fit.
anterior site and posterior site means remain significant (Tukey HSD test, P ¼.042) for final displacement. In this model, where site was included as a covariate, BMD was only significant for final displacement (P ¼ .021), not initial displacement (P ¼ .075). No other factors were significant. If early failures are not included, then no significant site effect can be detected from the data. This is because no early failures occurred at the posterior placement site. DISCUSSION Rotator cuff tears are a common source of shoulder pain, with an associated disability that is often pronounced. Treatment is intended to minimize morbidity and maximize return of function. This has progressed to all-arthroscopic techniques, which can be less invasive; however, anchor placement, suture management, and arthroscopic knot tying can be technically challenging with these techniques. Novel means of all-arthroscopic rotator cuff repairs include knotless suture anchors, which can reduce difficulty in performing repairs. We analyzed 2 knotless suture anchors and compared them with a conventional suture anchor, which served as our control. Although our evaluation using fresh-frozen cadaveric specimens did not demonstrate significant differences in overall anchor displacement, it did suggest trends toward early failure for the Mitek Bioknotless RC (P ¼ .07), in which 5 of 15 specimens pulled out from the humerus in the initial phase of cyclic testing. This finding did not reach statistical significance but did reveal a trend that could cause concern for clinicians who use this implant. If the implant were to pull out the first time any force were placed on it postoperatively, early gapping and potential failure of the repair could occur.
J Shoulder Elbow Surg March/April 2008
Figure 5 Final displacement vs site. Site was shown to be a significant predictor of final displacement, with the posterior site having the least displacement (P ¼ .029). Ant, Anterior; Mid, middle; Post, posterior. Data are presented with the standard deviation.
One potential cause for the early pullout could be inherent in the design of the anchor. The Bioknotless RC is a triangular implant and relies on subcortical fixation, in which the implant is locked under the humeral cortex. The inability of the implant to become locked could be due to incomplete insertion of the anchor. After the first early failures of the Bioknotless RC, special care was taken to ensure the implant was fully seated within its drill hole, although no noticeable differences were noted between the insertional depth of the implants that failed and those that did not. As evidenced by our other findings, if the Bioknotless RC initial fixation was secure, the implant performed well in testing with low displacement compared with the other anchors. The other anchors demonstrated no trend toward early failure, which could be of benefit for surgeons who use them. If a load were applied before tendon–bone healing, the anchor would lessen the chance of gap formation. We also determined that BMD measured by DEXA was negatively correlated with initial and final displacement and was a significant predictor of both. These data differ from some published studies that detected no difference between bone quality and pullout.1,8 Barber et al1 compared the bone densities at the anchor sites and no correlation was found between pullout strength and site specific BMD. Later work by Tingart et al10 did demonstrate significant differences in anchor pullout strength with regard to bone quality as measured by peripheral quantitative computed tomography scanning. These authors looked at specific areas of the greater and lesser tuberosities, while distinguishing between trabecular and cortical bone in their measurements. In contrast with both of those studies, our study evaluated BMD of the entire proximal humerus and not just specific areas within it. A broader age range and quality of bone seen in
Brown et al
J Shoulder Elbow Surg Volume 17, Number 2
our study may have contributed to our finding a significant effect. Anchor position within the greater tuberosity was evaluated, and placement was found to be significant for predicting final, but not initial, displacement. Barber et al1 also found a significant placement location effect, reporting a 40% lower pullout strength in the anterior greater tuberosity compared with the posterior greater tuberosity. Another more recent study,11 however, concluded just the opposite, that higher pullout strength was associated with more anterior and mid placement of suture anchors compared with posterior placement. Our study demonstrated a significantly lower resistance to slippage or gapping for both anterior- and mid-placed anchors compared with posterior-placed anchors. This result was primarily due to the early pullout with the Mitek anchor, which never occurred at the posterior site. Although pullout strength and slippage leading to gapping are not interchangeable measures, our result is more in line with that of Barber et al.1 No explanation for the differences in results could be found in those previous studies. Differences in testing protocol, outcome measures, anchor types, site location, and insertion techniques make a direct and universal comparison of findings difficult. One potential limitation of this study might be that the rotator cuff was not included as part of the model, thus removing the suture-tendon pullout component of the gapping mechanism. Our primary objective when we developed the experimental protocol was to focus on potential gapping caused by suture elongation or slippage of the anchor in the bone. To do so, we sought to limit the number of factors that could influence these measurements, particularly the potentially highly variable soft tissue–suture interface. Another limitation of this study relates to anchor insertion. After placement, each anchor was preloaded by hand to verify that it was adequately secured below subcortical bone, without objective quantification. A goal of our study was to assess the anchors in an environment that was as realistic as possible. We used physiologic forces in cyclic loading at maximal in vivo tensile force angles. In accordance with this goal, we used manual force to assess anchor stability as it is performed during routine arthroscopic conditions when surgeons attempt to ascertain, by pulling on the suture, if the anchor is secure enough that it will not pull out with a relatively small force. No published data were found on how any of these anchors performed under cyclic loading conditions. The only suture anchor data found was quasi-static tension load-to-failure testing,2 and no published data of any kind were found for the Mitek Bioknotless RC. We determined it would be imperative to elucidate anchor specific performance characteristics better before adding other variables, such as suture configuration
317
(the Opus Magnum anchor has its own suture passer that deploys an incline mattress suture), or the rotator cuff itself. In addition, physiologic cyclic loading of a repaired rotator cuff can lead to failure with as few as 25 cycles,5 and consequently, knowledge of how suture anchors perform under such conditions, with as many variables eliminated as possible, would be useful information. We believe an understanding of anchor-specific characteristics under physiologic conditions can aid surgeons in making a better-informed decision when choosing a specific suture anchor for treating patients with rotator cuff tears. It should be recognized that the anchor is only one component in the entire construct and contributes only partially to the stability or instability of the entire construct. Other potential contributors to gapping include the suture configuration and quality of the purchase that is obtained in rotator cuff, as well as the characteristics of the cuff itself. Further studies including the rotator cuff would provide a more clinically relevant estimation of total gapping expected to be seen when the various anchors are used. Those tests might yield substantially different results, particularly if failure at the cuff occurred at lower loads than those required to displace the anchor or stretch the suture. In such a study, however, the increased number of variables might obscure the specific causes of the gapping, and the variances would be expected to be higher. We acknowledge Smith & Nephew, Mitek, and Opus Medical, which donated the implants used in this study. REFERENCES
1. Barber FA, Feder SM, Burkhart SS, Ahrens J. The relationship of suture anchor failure and bone density to proximal humerus location: a cadaveric study. Arthroscopy 1997;3:340-5. 2. Barber FA, Herbert MA, Richards DP. Suture and suture anchors: update 2003. Arthroscopy 2003;9:985-90. 3. Burkhart SS. The deadman theory of suture anchors: observations along a South Texas fence line. Arthroscopy 1995;11:119-23. 4. Burkhart SS, Johnson TC, Wirth MA, Athanasiou KA. Cyclic Loading of transosseous rotator cuff repairs: tension overload as a possible cause of failure. Arthroscopy 1997;13:172-6. 5. Burkhart SS, Diaz Pagan JL, Wirth MA, Athanasiou KA. Cyclic loading of anchor-based rotator cuff repairs: confirmation of the tension overload phenomenon and comparison of suture anchor fixation with transosseous fixation. Arthroscopy 1997;13:720-4. 6. Craft DV, Moseley JB, Cawley PW, Noble PC. Fixation strength of rotator cuff repairs with suture anchors and the transosseous suture technique. J Shoulder Elbow Surg 1996;5:32-40. 7. Galatz LM, Ball CM, Teefey SA, Middleton WD, Yamaguchi K. The outcome and repair integrity of completely arthroscopically repaired large and massive rotator cuff tears. J Bone Joint Surg Am 2004;86:219-24. 8. Goradia VK, Mullen DJ, Boucher HR, Parks BG, O’Donnell JB. Cyclic loading of rotator cuff repairs: a comparison of bioabsorbable tacks with metal suture anchors and transosseous sutures. Arthroscopy 2001;4:360-4. 9. Lewis CW, Schlegel TF, Hawkins RJ, James SP, Turner AS. The effect of immobilization on rotator cuff healing using modified
318
Brown et al
Mason-Allen stitches: a biomechanical study in sheep. Biomed Sci Instrum 2001;37:263-8. 10. Tingart MJ, Apreleva M, Zurakowski D, Warner JJP. Pullout strength of suture anchors used in rotator cuff repair. J Bone Joint Surg Am 2003;85:2190-8.
J Shoulder Elbow Surg March/April 2008
11. Tingart MJ, Apreleva M, Lehtinen J, Zurakowski D, Warner JJP. Anchor design and bone mineral density affect the pull-out strength of suture anchors in rotator cuff repair: which anchors are best to use in patients with low bone quality? Am J Sports Med 2004; 32:1466-73.
This study evaluated the resistance to gapping and the mode of failure for 2 knotless suture anchor systems used for rotator cuff repair compared with the performance of a conventional titanium anchor system. Eight matched pairs of fresh-frozen humeri were dissected free of all soft tissues and scanned to measure bone mineral density (BMD). The suture anchor systems tested were the TwinFix 5.0 Titanium (Smith & Nephew, Andover, MA), Bioknotless RC (DePuy Mitek, Norwood, MA), and Magnum (Opus Medical, San Juan Capistrano, CA), and each was inserted into each humerus. Cyclic, tensile loading was applied through the suture loop for 5000 cycles, or until failure, by using a servohydraulic testing machine. Gapping distances, defined as increasing elongation of the bone/anchor/suture system, were continuously measured. Total cycles to failure and mechanism of failure were documented. Mean initial (first cycle) and final (last cycle) gapping distances were 3.81 mm and 5.36 mm for the TwinFix 5.0, 4.02 mm and 5.34 mm for the Bioknotless RC, and 3.56 mm and 4.98 mm for the Magnum anchors. No significant difference was detected among mean gap openings (P > .05). However, the Bioknotless RC had more early failures (5) than the other 2 implants (1 each), approaching significance (P ¼ .07). Trials of the Bioknotless RC that did not fail early were found to have significantly less gap opening than the other 2 systems for both initial (1.89 mm vs 3.82 mm for the TwinFix 5.0 and 3.56
a
Sports Medicine and Shoulder Service, Hospital for Special Surgery; bComprehensive Orthopedic Specialists; and cDepartment of Orthopedic Surgery, University of Kansas Medical Center. Supported by a grant from the Marc A. and Elinor J. Asher Orthopedic Research Endowment. Reprint requests: E. Bruce Toby, MD, Peltier/Reckling Chairman, Department of Orthopedic Surgery, University of Kansas Medical Center, 3901 Rainbow Blvd, Kansas City, KS 66160 (E-mail: [email protected]). Copyright ª 2008 by Journal of Shoulder and Elbow Surgery Board of Trustees. 1058-2746/2008/$34.00 doi:10.1016/j.jse.2007.05.016
mm for the Magnum) and final (2.00 mm vs 4.68 mm for the TwinFix 5.0 and 4.24 mm for the Magnum) gap opening. BMD was a significant predictor of initial (P ¼ .029) and final (P ¼ .008) gap opening, whereas the site of anchor insertion was a significant predictor of final displacement. The Opus Magnum was comparable with a conventional suture anchor, but the Mitek Bioknotless RC showed a trend toward early failure. Biomechanical analysis of knotless suture anchor systems can demonstrate trends among implants in an experimental setting. Knowledge of these trends could influence implant selection. (J Shoulder Elbow Surg 2008;17:313-318.)
I
nitial fixation stability is an important variable in suture anchor repairs and can direct postoperative rehabilitation. Fixation failure after primary rotator cuff repair can occur and does so at the implant– bone, implant–suture, or suture–tendon interface.7 Suture anchor repair has been shown to be significantly stronger than bone fixation through transosseous tunnels under physiologic cyclical loading.5 The optimal location of suture anchor placement has been identified.10 Refinements in anatomic restoration of the rotator cuff are still evolving and have included the use of knotless suture anchors, which provide a potential advantage of minimizing gap formation while simultaneously eliminating the technical challenges associated with arthroscopic knot tying. Minimizing gap formation is important because failure to do so reduces contact between tendon and bone, which can adversely affect the healing of tendon to bone, both in the early (6-8 week) postoperative period as well as several months after repair.9 This dynamic biomechanical study compared initial fixation, under physiologic cyclical loading conditions, of 2 knotless anchors, the Magnum (Opus Medical, San Juan Capistrano, CA) and the Bioknotless RC (DePuy Mitek, Norwood, MA), with the TwinFix 5.0 Titanium (Smith & Nephew, Andover, MA) suture anchor, which has previously been identified as the metal suture anchor with the highest absolute pullout strength in cancellous porcine bone.2 Each of the anchors tested is shown in Figure 1.
313
314
Brown et al
J Shoulder Elbow Surg March/April 2008
Figure 1 The three suture anchors tested (from left to right) were the Opus Magnum (undeployed), the Mitek Bioknotless RC, and the Smith & Nephew TwinFix 5.0 Titanium.
MATERIALS AND METHODS Eight matched pairs of proximal humeri from fresh-frozen cadavers (3 women, 5 men), with an average age of 60.4 years (range, 26-89 years), were thawed, dissected free of all soft tissues, and thawed for 24 hours before testing. The humeri were transected 10 cm distal to the most proximal point of the humeral head. Dual energy x-ray absorptiometry (DEXA) was used to collect bone mineral density (BMD) values for each whole humerus in units of g/mm3. Each humeral diaphysis was transfixed with a 0.62 Kirschner wire and potted to a depth of 3 cm in dental cement. The rotator cuff footprint was gently de´brided to simulate in vivo preparation of the footprint. A total of 48 suture anchors, 16 of each type, were inserted in the proximal-anterior, proximal-middle, and proximal-posterior regions of the greater tuberosity according to the manufacturers’ guidelines (see Figure 2), as defined in a published study.10 One anchor from each of the 3 manufacturers was inserted into each humerus. The first suture anchor was chosen at random and was placed in the proximal-anterior region of the greater tuberosity, 5 mm posterior to the bicipital groove and approximately 5 mm medial to the tip of the greater tuberosity. The second was placed in the proximal-middle region of the tuberosity at least 10 mm posterior to the first implant. The third anchor was placed 10 mm posterior in the proximal-posterior region. Suture anchor placement was equally distributed across regions throughout all specimens. Each anchor was inserted according to the manufacturer’s specifications: the Opus Mangum perpendicular to the humerus, the other 2 anchors at 45 angles. After anchor insertion, each was preloaded by hand to verify that it was adequately secured before cyclic testing. The potted humerus specimens were secured to the base of a tensile testing machine in 45 of abduction relative to the loading forces (see Figure 3), representing the abduction angle that produces maximal in vivo tensile forces.3 Each corresponding suture was tied according to the pertinent manufacturer’s recommendations to a 5.3-mm-diameter eyebolt fixed to the loading actuator above. Each suture was tied while the eyebolt was in direct contact with the surface
Figure 2 This drawing demonstrates the regions within the greater tuberosity. All anchors in this study were placed in the proximal regions of the greater tuberosity, in equal distribution. Figure adapted with permission from Tingart et al.10
Figure 3 The testing setup used for our study. Specimens were fixed to the MTS Mini-Bionix machine using custom-made clamps, and the I loop was attached to the servohydraulic actuator.
of the bone at the anchor site. Manufacturer-supplied, nonabsorbable braided suture was used for each anchor. Each anchor used a single suture. The testing commenced with a single pretensioning load to 60 N, which was held for approximately 3 seconds. Then, cyclic loading (sine waveform under load control) commenced using a servohydraulic material testing system (Mini-Bionix, MTS Systems Corp, Eden Prairie, MN) from 10 to 60 N at 2 Hz. These values have been described as within physiologic ranges for rotator cuff loading.4,6 Relative displacement between the eyebolt to which the suture was tied and the stationary humerus was measured after initial pretensioning of each anchor and continuously thereafter during testing. Both initial and final displacement
Brown et al
J Shoulder Elbow Surg Volume 17, Number 2
measures included possible motion or slippage of the anchor, as well as breakage or elongation of the suture material. The anchors were loaded to an arbitrary cutoff of 5000 cycles. If an anchor reached 5000 cycles load to failure with 20 cycles at 60 N, 100 cycles in increasing increments of 30 N was performed until failure. The failure mode and amount of suture pullout of each anchor was documented, as was the number of cycles at which failure occurred. The cycle at which 50% failure (5-mm displacement) and 100% failure (10-mm displacement) occurred was extracted from the original data. The compliance of the potted humerus and fixtures was independently measured for each specimen and later subtracted from the total displacement curves recorded for each anchor test. Repeated measures analysis of variance (ANOVA) and Kruskal-Wallis ANOVA were used to test for treatment effects, after which the Tukey honestly significantly differences (HSD) test or Mann-Whitney test were performed to make specific post hoc comparisons between anchor treatments. Significance was set at P ¼ .05 for rejecting the null hypothesis that no difference in the distribution means exists. Variances were compared using the Levine test for homogeneity of variance. Correlations were presented using Pearson r correlation coefficients.
RESULTS Each anchor treatment group underwent 16 trials. One Mitek trial was unsuccessful owing to insertion failure and was excluded. The mean initial displacement was TwinFx 5.0, 3.82 mm; Bioknotless RC, 4.02 mm; and Magnum, 3.56 mm; the respective mean final displacement for each anchor was 5.37 mm, 5.34 mm,and 4.98 mm. These results are summarized in Table I. Anchors failed in one of several modes: suture breakage, anchor pullout, or severing of the suture by the bone. Of the Magnum anchors, 8 failed by suture breakage, 3 by anchor pullout, 3 by suture cutting, and 2 did not fail. For the Mitek anchor, 2 sutures broke, 11 anchors pulled out, 1 failed to insert, and 2 did not fail during testing. The Twinfix anchors failed 15 times by suture breakage, and 1 did not fail. When all included trial data were considered, no significant difference was detected among the mean displacement of the 3 anchor types (repeated measures ANOVA, P > .05). Nonparametric tests (Kruskal-Wallis ANOVA) also revealed no significant effect of anchor choice on initial or final displacement or on the number of trials reaching 50% or 100% failure levels. The Mitek anchors, however, had a significantly higher variance (Levine test, P < .0001) than the other anchors for both the initial and final displacement. This higher variance was primarily due to a larger number of early failures, where early failures were defined as tests where the anchor pulled out or failed before 5000 cycles. Comparisons of the number of early failures or 100% failure levels between Mitek (early failures ¼ 5) and the other 2 anchors reached nearly significant P values (Fischer exact, P ¼ .07).
315
Table I Initial and final displacements with all specimens included
Anchor
Initial, mean (SD) mm
Final, mean (SD) mm
Smith & Nephew Mitek Opus
3.82 (1.41) 4.02 (3.42) 3.56 (0.75)
5.37 (2.05) 5.34 (3.76) 4.98 (1.74)
Table II Initial and final displacements with early failures excluded
Anchor
Initial, mean (SD) mm
Final, mean (SD) mm
Smith & Nephew Mitek Opus
3.82 (1.41) 1.89 (1.80) 3.56 (0.75)
4.68 (2.09) 1.99 (2.10) 4.24 (1.60)
The mean initial and mean final displacements for all anchors, when early failures were excluded, are summarized in Table II. With early failures excluded (treated as missing values), the Mitek Bioknotless RC had significantly smaller initial and final displacements than the other 2 anchors (repeated measures ANOVA for initial displacement, P ¼ .939 with early failures and P < .001 without early failures; final displacement P ¼ .856 with early failures and P ¼ .003 without early failures.) Post hoc power analysis indicated that, with the number of specimens actually tested and the resulting variances seen in each treatment group, the statistical tests would have a 95% confidence of being able to detect a difference greater than 1.1 mm between the treatment groups. BMD, as measured by DEXA, was correlated negatively to both initial and final displacement (Pearson r ¼ –0.319 and –0.382, respectively). BMD was a significant predictor of both initial and final displacement, even with early failures included (P ¼ .029 and P ¼ .008, respectively). Figure 4 shows the distribution of BMD with respect to the final displacement recorded. The slope of the linear fit was –4.66 mm of final displacement per unit of BMD in g/cm3. Anchors were placed in 3 sites: anterior, mid, and posterior. The mean final displacements grouped by site are shown in Figure 5. The effect of site was examined using ANOVA, with site and anchor type as categoric independent variables and BMD as a continuous predictor variable. Site was significant (P ¼.029) for predicting final, but not initial, displacement (P ¼ .353). Significant differences in the group means for final displacement were found between anterior and posterior sites (P ¼ .011) and between mid and posterior sites (P ¼.043). If P values are adjusted for multiple comparisons, then only the difference between
316
Brown et al
Figure 4 Final displacement vs bone mineral density (BMD), which was shown to be a significant predictor of initial and final displacement (P ¼ .029 and P ¼ .008, respectively). The lines represent the linear fit and the 95% confidence interval for that linear fit.
anterior site and posterior site means remain significant (Tukey HSD test, P ¼.042) for final displacement. In this model, where site was included as a covariate, BMD was only significant for final displacement (P ¼ .021), not initial displacement (P ¼ .075). No other factors were significant. If early failures are not included, then no significant site effect can be detected from the data. This is because no early failures occurred at the posterior placement site. DISCUSSION Rotator cuff tears are a common source of shoulder pain, with an associated disability that is often pronounced. Treatment is intended to minimize morbidity and maximize return of function. This has progressed to all-arthroscopic techniques, which can be less invasive; however, anchor placement, suture management, and arthroscopic knot tying can be technically challenging with these techniques. Novel means of all-arthroscopic rotator cuff repairs include knotless suture anchors, which can reduce difficulty in performing repairs. We analyzed 2 knotless suture anchors and compared them with a conventional suture anchor, which served as our control. Although our evaluation using fresh-frozen cadaveric specimens did not demonstrate significant differences in overall anchor displacement, it did suggest trends toward early failure for the Mitek Bioknotless RC (P ¼ .07), in which 5 of 15 specimens pulled out from the humerus in the initial phase of cyclic testing. This finding did not reach statistical significance but did reveal a trend that could cause concern for clinicians who use this implant. If the implant were to pull out the first time any force were placed on it postoperatively, early gapping and potential failure of the repair could occur.
J Shoulder Elbow Surg March/April 2008
Figure 5 Final displacement vs site. Site was shown to be a significant predictor of final displacement, with the posterior site having the least displacement (P ¼ .029). Ant, Anterior; Mid, middle; Post, posterior. Data are presented with the standard deviation.
One potential cause for the early pullout could be inherent in the design of the anchor. The Bioknotless RC is a triangular implant and relies on subcortical fixation, in which the implant is locked under the humeral cortex. The inability of the implant to become locked could be due to incomplete insertion of the anchor. After the first early failures of the Bioknotless RC, special care was taken to ensure the implant was fully seated within its drill hole, although no noticeable differences were noted between the insertional depth of the implants that failed and those that did not. As evidenced by our other findings, if the Bioknotless RC initial fixation was secure, the implant performed well in testing with low displacement compared with the other anchors. The other anchors demonstrated no trend toward early failure, which could be of benefit for surgeons who use them. If a load were applied before tendon–bone healing, the anchor would lessen the chance of gap formation. We also determined that BMD measured by DEXA was negatively correlated with initial and final displacement and was a significant predictor of both. These data differ from some published studies that detected no difference between bone quality and pullout.1,8 Barber et al1 compared the bone densities at the anchor sites and no correlation was found between pullout strength and site specific BMD. Later work by Tingart et al10 did demonstrate significant differences in anchor pullout strength with regard to bone quality as measured by peripheral quantitative computed tomography scanning. These authors looked at specific areas of the greater and lesser tuberosities, while distinguishing between trabecular and cortical bone in their measurements. In contrast with both of those studies, our study evaluated BMD of the entire proximal humerus and not just specific areas within it. A broader age range and quality of bone seen in
Brown et al
J Shoulder Elbow Surg Volume 17, Number 2
our study may have contributed to our finding a significant effect. Anchor position within the greater tuberosity was evaluated, and placement was found to be significant for predicting final, but not initial, displacement. Barber et al1 also found a significant placement location effect, reporting a 40% lower pullout strength in the anterior greater tuberosity compared with the posterior greater tuberosity. Another more recent study,11 however, concluded just the opposite, that higher pullout strength was associated with more anterior and mid placement of suture anchors compared with posterior placement. Our study demonstrated a significantly lower resistance to slippage or gapping for both anterior- and mid-placed anchors compared with posterior-placed anchors. This result was primarily due to the early pullout with the Mitek anchor, which never occurred at the posterior site. Although pullout strength and slippage leading to gapping are not interchangeable measures, our result is more in line with that of Barber et al.1 No explanation for the differences in results could be found in those previous studies. Differences in testing protocol, outcome measures, anchor types, site location, and insertion techniques make a direct and universal comparison of findings difficult. One potential limitation of this study might be that the rotator cuff was not included as part of the model, thus removing the suture-tendon pullout component of the gapping mechanism. Our primary objective when we developed the experimental protocol was to focus on potential gapping caused by suture elongation or slippage of the anchor in the bone. To do so, we sought to limit the number of factors that could influence these measurements, particularly the potentially highly variable soft tissue–suture interface. Another limitation of this study relates to anchor insertion. After placement, each anchor was preloaded by hand to verify that it was adequately secured below subcortical bone, without objective quantification. A goal of our study was to assess the anchors in an environment that was as realistic as possible. We used physiologic forces in cyclic loading at maximal in vivo tensile force angles. In accordance with this goal, we used manual force to assess anchor stability as it is performed during routine arthroscopic conditions when surgeons attempt to ascertain, by pulling on the suture, if the anchor is secure enough that it will not pull out with a relatively small force. No published data were found on how any of these anchors performed under cyclic loading conditions. The only suture anchor data found was quasi-static tension load-to-failure testing,2 and no published data of any kind were found for the Mitek Bioknotless RC. We determined it would be imperative to elucidate anchor specific performance characteristics better before adding other variables, such as suture configuration
317
(the Opus Magnum anchor has its own suture passer that deploys an incline mattress suture), or the rotator cuff itself. In addition, physiologic cyclic loading of a repaired rotator cuff can lead to failure with as few as 25 cycles,5 and consequently, knowledge of how suture anchors perform under such conditions, with as many variables eliminated as possible, would be useful information. We believe an understanding of anchor-specific characteristics under physiologic conditions can aid surgeons in making a better-informed decision when choosing a specific suture anchor for treating patients with rotator cuff tears. It should be recognized that the anchor is only one component in the entire construct and contributes only partially to the stability or instability of the entire construct. Other potential contributors to gapping include the suture configuration and quality of the purchase that is obtained in rotator cuff, as well as the characteristics of the cuff itself. Further studies including the rotator cuff would provide a more clinically relevant estimation of total gapping expected to be seen when the various anchors are used. Those tests might yield substantially different results, particularly if failure at the cuff occurred at lower loads than those required to displace the anchor or stretch the suture. In such a study, however, the increased number of variables might obscure the specific causes of the gapping, and the variances would be expected to be higher. We acknowledge Smith & Nephew, Mitek, and Opus Medical, which donated the implants used in this study. REFERENCES
1. Barber FA, Feder SM, Burkhart SS, Ahrens J. The relationship of suture anchor failure and bone density to proximal humerus location: a cadaveric study. Arthroscopy 1997;3:340-5. 2. Barber FA, Herbert MA, Richards DP. Suture and suture anchors: update 2003. Arthroscopy 2003;9:985-90. 3. Burkhart SS. The deadman theory of suture anchors: observations along a South Texas fence line. Arthroscopy 1995;11:119-23. 4. Burkhart SS, Johnson TC, Wirth MA, Athanasiou KA. Cyclic Loading of transosseous rotator cuff repairs: tension overload as a possible cause of failure. Arthroscopy 1997;13:172-6. 5. Burkhart SS, Diaz Pagan JL, Wirth MA, Athanasiou KA. Cyclic loading of anchor-based rotator cuff repairs: confirmation of the tension overload phenomenon and comparison of suture anchor fixation with transosseous fixation. Arthroscopy 1997;13:720-4. 6. Craft DV, Moseley JB, Cawley PW, Noble PC. Fixation strength of rotator cuff repairs with suture anchors and the transosseous suture technique. J Shoulder Elbow Surg 1996;5:32-40. 7. Galatz LM, Ball CM, Teefey SA, Middleton WD, Yamaguchi K. The outcome and repair integrity of completely arthroscopically repaired large and massive rotator cuff tears. J Bone Joint Surg Am 2004;86:219-24. 8. Goradia VK, Mullen DJ, Boucher HR, Parks BG, O’Donnell JB. Cyclic loading of rotator cuff repairs: a comparison of bioabsorbable tacks with metal suture anchors and transosseous sutures. Arthroscopy 2001;4:360-4. 9. Lewis CW, Schlegel TF, Hawkins RJ, James SP, Turner AS. The effect of immobilization on rotator cuff healing using modified
318
Brown et al
Mason-Allen stitches: a biomechanical study in sheep. Biomed Sci Instrum 2001;37:263-8. 10. Tingart MJ, Apreleva M, Zurakowski D, Warner JJP. Pullout strength of suture anchors used in rotator cuff repair. J Bone Joint Surg Am 2003;85:2190-8.
J Shoulder Elbow Surg March/April 2008
11. Tingart MJ, Apreleva M, Lehtinen J, Zurakowski D, Warner JJP. Anchor design and bone mineral density affect the pull-out strength of suture anchors in rotator cuff repair: which anchors are best to use in patients with low bone quality? Am J Sports Med 2004; 32:1466-73.