Local variations in mechanical properties of human hamstring tendon autografts for anterior cruciate ligament reconstruction do not translate to a mechanically inferior strand

A ruptured anterior cruciate ligament (ACL) is often reconstructed with a multiple-strand autograft of a semitendinosus tendon alone or combined with a gracilis tendon. Up to 10% of patients experience graft rupture. This potentially results from excessive local tissue strains under physiological loading which could either result in direct mechanical failure of the graft or induce mechanobiological weakening. Since the original location in the hamstring tendon cannot be traced back from an autograft rupture site, this study explored whether clinical outcome could be further improved by avoiding specific locations or regions of human semitendinosus and/or gracilis tendons in ACL grafts due to potential mechanical or biochemical inferiority. Additionally, it examined numerically which clinically relevant graft configurations experience the lowest strains - and therefore the lowest rupture risk - when loaded with equal force. Remnant full-length gracilis tendons from human ACL reconstructions and full-length semitendinosus- and ipsilateral gracilis tendons of human cadaveric specimens were subjected to a stress-relaxation test. Locations at high risk of mechanical failure were identified using particle tracking to calculate local axial strains. As biochemical properties, the water-, collagen-, glycosaminoglycan- and DNA content per tissue region (representing graft strands) were determined. A viscoelastic lumped parameter model per tendon region was calculated. These models were applied in clinically relevant virtual graft configurations, which were exposed to physiological loading. Configurations that provided lower stiffness - i.e., experiencing higher strains under equal force - were assumed to be at higher risk of failure. Suitability of the gracilis tendon proper to replace semitendinosus muscle-tendon junction strands was examined. Deviations in local axial strains from the globally applied strain were of similar magnitude as the applied strain. Locations of maximum strains were uniformly distributed over tendon lengths. Biochemical compositions varied between tissue regions, but no trends were detected. Viscoelastic parameters were not significantly different between regions within a tendon, although semitendinosus tendons were stiffer than gracilis tendons. Virtual grafts with a full-length semitendinosus tendon alone or combined with a gracilis tendon displayed the lowest strains, whereas strains increased when gracilis tendon strands were tested for their suitability to replace semitendinosus muscle-tendon junction strands. Locations experiencing high local axial strains - which could increase risk of rupture - were present, but no specific region within any of the investigated graft configurations was found to be mechanically or biochemically deviant. Consequently, no specific tendon region could be indicated to provide a higher risk of rupture for mechanical or biochemical reasons. The semitendinosus tendon provided superior stiffness to a graft compared to the gracilis tendon. Therefore, based on our results it would be recommended to use the semitendinosus tendon, and use the gracilis tendon in cases where further reinforcement of the graft is needed to attain the desired length and cross-sectional area. All these data support current clinical standards.