Resistance to disruption and gapping of peripheral nerve repairs: an in vitro biomechanical assessment of techniques.

One potential cause of suboptimal results after nerve repair is disruption or gapping of the neurorrhaphy in the postoperative period. This study assesses the biomechanical strength of five nerve repair techniques: fibrin glue, simple epineurial sutures, and three other novel neurorrhaphy methods. Fifty rabbit sciatic nerve segments were divided and repaired utilizing one of five different methods, producing five groups of ten specimens. Fibrin glue and four epineurial suture techniques (simple, horizontal mattress, "Tajima," "Bunnell") were employed. Repaired nerve segments were ramp-loaded to failure on an Instron 8300 materials-testing machine at a displacement rate of 5 mm/min. Gapping at the repair site was captured using high-resolution video. Differences among the five groups were assessed for significance using ANOVA and Fisher's protected least squares differences post-hoc testing. The mean force to produce disruption was higher for mattress suture repairs relative to simple repairs, but not significantly so (p = 0.31). Both were significantly stronger than fibrin glue repairs (p < 0.0001). "Tajima" and "Bunnell" repairs were both statistically stronger than glue (p < 0.0001), simple (p < 0.0001), or mattress (p = 0.0004) repairs, but not significantly different from one another (p = 0.48). Data for gapping at the repair site were similar with all suture techniques outperforming fibrin glue (p = 0.003). "Bunnell" repairs demonstrated the most resistance to gapping, compared to glue (p < 0.0001), simple (p = 0.0001), mattress (p = 0.007) and "Tajima" repairs (p = 0.01). These data demonstrate that repairs done utilizing fibrin glue are significantly weaker than all types of suture repairs. Two novel techniques for nerve repair (epineurial "Tajima" and "Bunnell") are significantly more resistant to disruption and gapping. Further evaluation to assess the effect of these repair techniques on function is required.