Can tape–screw fixation of a quadrupled semitendinosus graft in a full-length tibial tunnel provide superior fixation compared with a doubled semitendinosus–gracilis held with an interference screw? A matched-pair cadaveric biomechanical comparison

Currently, the ideal anterior cruciate ligament reconstruction (ACLR) graft and fixation choice remains elusive, with all common graft options having some disadvantages. Patella tendon autografts have become less popular in recent years [1], possibly due to concerns over donor-site pain and higher risk of osteoarthritis in cohort studies [2]. Allografts have documented higher revision rates [3, 4], however a great advantage is their absence of donor-site morbidity. Hamstring autografts have become more commonly utilized in recent years [1] in some regions, however concerns remain over higher failure rates in both registry studies [5] and randomized controlled trials [6] compared with autograft patella tendon grafts.

Gracilis harvest in addition to semitendinosus has been linked to knee flexor weakness by some authors [7, 8] and low rates of tendon regeneration [9], however the clinical relevance of this has been questioned by others [10].

Magnussen et al. [11] recently published data suggesting that smaller hamstring graft size was linked to higher failure in ACLR, and this has been confirmed by others [12]. While graft constructs of quadrupled semitendinosus are theoretically larger in diameter than doubled semitendinosus–gracilis (ST/G) grafts, reported matched-pair analysis to date surprisingly does not support this [13]. Quadrupled semitendinosus (Quad ST) grafts are also much shorter [14], making them unable to be fixed to the tibia with interference screws. While the shorter Quad ST grafts can be fixed with adjustable suspensory fixation devices on the tibia as well as the femur, concerns have been published over elongation of adjustable suspensory fixation devices under cyclic loading, with vigorous recent debate in literature [15‐19]. In addition, tibial fixation in ACLR remains more problematic than femoral fixation, being described as the weak link in ACL fixation [20, 21]. Interference screw and polyethylene braided terephthalate (PET) tape fixation has been promoted as an alternative method for both tibial and femoral fixation of short quadrupled ST grafts, with supportive biomechanical data [14, 22, 23] for methods using partial length inside-out created tunnels, however this tape–screw method cannot be utilized with a fixed suspensory fixation loop on the femur.

The aim of this study is to investigate an alternative method of tape–screw construct fixation using full-length outside-in created tibial tunnels with tape tied over the screw for additional fixation. This alternative method would offer a graft tibial fixation option to surgeons who wished to preoperatively utilize a quadrupled ST graft to avoid harvesting the gracilis, or intraoperatively if a harvested gracilis tendon was inadequate in diameter or length to be used, while utilizing a fixed suspensory fixation device on the femur.

Two alternative ACLR tibial fixation methods were evaluated by matched-pair cadaveric laboratory biomechanical analysis: a ST/G graft with interference screw (ST/G-screw) and a Quad ST with 5-mm braided PET tape and screw (Quad ST-tape). Ethics approval was obtained from the institution’s Human Ethics Committee.

Table 1 details the harvested tendon lengths, graft length, graft diameter, tibial drill diameter, and screw diameter. The Quad ST grafts had significantly larger mean diameter of 9.5 mm compared with the ST/G grafts (8.7 mm) (p = 0.0041).

All 12 constructs failed during cyclic testing, hence no load-to-failure tests were conducted. Eleven of the constructs failed during the first of the 500-N cycles, hence comparison of displacement results was only conducted for the first phase of cyclic loading, where the peak load was 220 N. The number of cycles to failure for each specimen is shown in Fig. 6. Two specimens survived more than 400 cycles at 220 N before failure, with failure defined as relative displacement between the graft and the tibial tunnel exceeding 5 mm. These two specimens (5 and 6 in Fig. 6) also had the highest bone mineral density (BMD) of all the tested specimen. Three specimens in the Quad-ST tape group and one specimen in the ST/G-screw group did not complete any cycles before the displacement exceeded 5 mm. As these four tests failed at 0 cycles, they do not appear in Fig. 6.

In the Quad ST-tape group, four out of six constructs failed below 100 cycles, and three of those failed in the first cycle. In contrast, the ST/G construct had two failures below 100 cycles, with the remainder of failures occurring after 385 cycles. The distribution of these failure data is presented in Fig. 7.

The displacement results for the ST/G and Quad ST-tape constructs, respectively, were as follows: elongation under initial load of 2.26 mm (SD 3.58 mm) and 4.32 mm (SD 3.73 mm) (Fig. 8), slip of unloaded graft after 500 cycles at 220 N of 3.77 mm (SD 2.05 mm) and 3.15 mm (SD 1.94 mm) (Fig. 3), and elongation at 10/100 cycles of 0.71 mm (SD 0.49 mm)/1.24 mm (SD 0.80 mm) and 2.28 mm (SD 2.46 mm)/4.3 mm (SD 5.70 mm) (Fig. 9).

The average BMD for the ST/G and Quad-ST tape group was 0.92 and 0.91 g/cm³, respectively, similar to the higher BMD in younger adults, in whom this type of surgery is more common [20]. The number of cycles to failure and the displacement results showed strong or very strong correlation with BMD for the Quad ST-tape construct (r < −0.6, r > 0.9) (Fig. 10). The opposite was true for the ST/G construct (−0.5 < r < 0.3) (Table 2).

The mode of failure for all specimens was displacement greater than 5 mm, with the modes of catastrophic failure for the two constructs being consistent (Table 3) but dissimilar. For the ST/G constructs, catastrophic failure was due to the graft slipping past the interference screw with the screw being left in place, except in one instance, where the cortical bone above the screw failed and the graft slipped out. For the Quad ST-tape construct, catastrophic failure occurred in five of the six constructs when the screw was pulled past the cortical bone enough for the knot to slip around the screw and pull through the cancellous bone. In one case the screw was pulled through to the tibial plateau by the knot.

There were no statistically significant differences between the parameters measured for the two tibial fixation methods, however there were high standard deviations, a common observation in ACL fixation literature [14, 21, 23, 29]. When failure due to 5 mm of relative displacement is considered as the critical parameter, these large variations resulted in insufficient samples to accurately determine the significance of the result. The results are clustered into two groups, those that failed within 100 cycles at 220 N, and those that failed after 385 cycles at 220 N. A greater proportion of Quad ST-tape constructs failed earlier than ST/G constructs, which, while not statistically significant, raises concerns about the stiffness and fixation strength of the Quad ST-tape construct.

It is important to note that the image-processing method used cannot isolate the cause of the early displacement failures, hence it was not possible to ascertain whether this failure was due to initial laxity in the tape, a poor interface between the interference screw, tape, and bone, or the elasticity of the graft–tape interface.

The tunnel friction forces are difficult to define, as they depend on several factors including the normal force, the friction coefficient between the materials, and their contact area. Matching the tibial tunnel diameter to the ST/G diameter resulted in friction along the entire length of the graft–tunnel interface, playing a role in resisting the tensile loading for this graft compared with the shorter Quad ST-tape, which had much less graft tunnel contact. Friction along the graft–tunnel interface may have led to longer survival of two low BMD ST/G specimens compared with the matched pair with Quad-ST tape, suggesting that interference fixation would be preferable in patients with lower BMD.

The stability of the ST/G reconstruction also depends on the size and density of the tendons, with poor-quality tendons being more likely to deform plastically or slip past the screw [29]. It is possible that the early failure of a higher BMD specimen from the ST/G group was caused by such problems. In contrast, the tensile strength of the Quad ST-tape construct depends on the compression of the tape to the screw–bone interface [14, 22, 29] and the added restriction of the knot holding the tape in place around that screw [23], but not tunnel friction. One difference between the two groups is with regards to how the measured parameters correlate with the BMD. As noted in the results, BMD showed strong correlation with all results in the Quad-ST tape construct group (peak and residual displacements at 500 cycles are excluded because n = 2). In contrast, the same parameters showed weak correlation with BMD for the ST/G construct.

The difference in mode of catastrophic failure is also noteworthy. The ST/G reconstructions characteristically failed when the graft slipped around the interference screw; i.e., both the static friction along the graft–tunnel interface and the anchoring with the interference screw failed to counter the tension load. The majority of Quad ST-tape constructs failed when the knotted tape slipped around the screw into the cancellous bone. The interference screws used had a different design from other reports for a partial inside-out technique tape fixation technique, which may have resulted in the heterogeneous fixation results in our study. Full-length tunnels were used to mimic an ACLR technique using fixed suspensory loop femoral fixation with tibial tape–screw fixation. If our tunnels had been drilled partial length inside-out, our results may have been different; however, given that the majority of Quad ST-tape failures were due to graft slippage around the screw, despite the tape being tied over the screw head, it remains uncertain if the graft slippage still would not have occurred. If the tunnels had been drilled partial length inside-out so that an isthmus of bone remained at the screw tip, this may have provided additional friction to the tape. However, as noted above, we attempted to add to the tape–screw fixation by tying a knot over the screw in compensation for the full-length tunnel.

One other issue of concern is the tendency of the Quad ST-tape construct to fail by elongation earlier than the ST-G construct. While the sample size was not sufficient to clearly determine which failure results were outliers, elongation as a material property of the braided 5-mm PET tape we utilized has been reported by other authors in situations other than ACL fixation [30], but not described in studies using wider 7-mm PET tape for ACL fixation [14, 22, 29]. Why the braided 7-mm PET tape does not reportedly suffer elongation, while the braided 5-mm PET tape does, remains uncertain. In addition, our testing method could not differentiate elongation of the tape only, elongation of the Quad-ST graft in the tibial tunnel, or both.

One of the strengths of this study is the use of matched human tissue for both the bony and soft components. Magen et al. [31] suggested that animal tissue should not be used to estimate the performance of interference screw fixation in human tissue, and while Colderidge and Amis [27] used bovine tissue, they also stated that younger human cadaveric tibia would have been ideal. In an uncontrolled 7-mm tape and 10-mm screw biomechanical study, Collette et al. [14] utilized cadaveric tissue; however, they used femoral heads, rather than cadaveric tibia, which would have had very different bone density and corticocancellous structure from a tibia. The only other study to use cadaveric tibia in a tape and screw study to date is that of Birmingham et al. [23], who reported mean failure load of 136 N (±136 N) with a 10-mm screw, compared with 288 N (±77 N) for the comparator Endobutton group. When they used a larger 12-mm screw and tied the tape over an additional button, they achieved a statistically significantly greater mean failure load of 668 N (±278 N), with the additional cortical fixation. It should be noted that we tied the tape over an 11-mm screw, rather than to an additional button, hence we did not gain cortical fixation. They reported greater stiffness in the larger screw–button–tape group than the standard screw–tape group and comparator Endobutton group, but the same migration.

The main weakness of the study is the limited number of matched constructs, which when combined with the high variability of measured parameters resulted in insufficient samples to accurately determine the significance of the results when failure due to 5 mm of relative displacement was considered as the critical parameter.

Tibial fixation of ACL quadrupled ST grafts tied over a tape–screw construct with a full outside-in created tunnel was not superior to ST/G grafts fixed with an interference screw.