Biomechanical analysis of screw constructs for atlantoaxial fixation in cadavers: a systematic review and meta-analysis

OBJECT: The unique and complex biomechanics of the atlantoaxial junction make the treatment of C1-2 instability a challenge. Several screw-based constructs have been developed for atlantoaxial fixation. The biomechanical properties of these constructs have been assessed in numerous cadaver studies. The purpose of this study was to systematically review the literature on the biomechanical stability achieved using various C1-2 screw constructs and to perform a meta-analysis of the available data. METHODS: A systematic search of PubMed through July 1, 2013, was conducted using the following key words and Boolean operators: "atlanto [all fields]" AND "axial [all fields]" OR "C1-C2" AND "biomechanic." Cadaveric studies on atlantoaxial fixation using screw constructs were included. Data were collected on instability models, fixation techniques, and range of motion (ROM). Forest plots were constructed to summarize the data and compare the biomechanical stability achieved. RESULTS: Fifteen articles met the inclusion criteria. An average (±SD) of 7.4±1.8 cadaveric specimens were used in each study (range 5-12). The most common injury models were odontoidectomy (53.3%) and cervical ligament transection (26.7%). The most common spinal motion segments potted for motion analysis were occiput-C4 (46.7%) and occiput-C3 (33.3%). Four screw constructs (C1 lateral mass-C2 pedicle screw [C1LM-C2PS], C1-2 transarticular screw [C1-C2TA], C1 lateral mass-C2 translaminar screw [C1LM-C2TL], and C1 lateral mass-C2 pars screw [C1LM-C2 pars]) were assessed for biomechanical stability in axial rotation, flexion/extension, and lateral bending, for a total of 12 analyses. The C1LM-C2TL construct did not achieve significant lateral bending stabilization (p=0.70). All the other analyses showed significant stabilization (p<0.001 for each analysis). Significant heterogeneity was found among the reported stabilities achieved in the analyses (p<0.001; I2>80% for all significant analyses). The C1LM-C2 pars construct achieved significantly less axial rotation stability (average ROM 36.27° [95% CI 34.22°-38.33°]) than the 3 other constructs (p<0.001; C1LM-C2PS average ROM 49.26° [95% CI 47.66°-50.87°], C1-C2TA average ROM 47.63° [95% CI 45.22°-50.04°], and C1LM-C2TL average ROM 53.26° [95% CI 49.91°-56.61°]) and significantly more flexion/extension stability (average ROM 13.45° [95% CI 10.53°-16.37°]) than the 3 other constructs (p<0.001; C1LM-C2PS average ROM 9.02° [95% CI 8.25°-9.80°], C1-C2TA average ROM 7.39° [95% CI 5.60°-9.17°], and C1LM-C2TL average ROM 7.81° [95% CI 6.93°-8.69°]). The C1-C2TA (average ROM 5.49° [95% CI 3.89°-7.09°]) and C1LM-C2 pars (average ROM 4.21° [95% CI 2.19°-6.24°]) constructs achieved significantly more lateral bending stability than the other constructs (p<0.001; C1LM-C2PS average ROM 1.51° [95% CI 1.23°-1.78°]; C1LM-C2TL average ROM -0.07° [95% CI -0.44° to 0.29°]). CONCLUSIONS: Meta-analysis of the existing literature showed that all constructs provided significant stabilization in all axes of rotation, except for the C1LM-C2TL construct in lateral bending. There were significant differences in stabilization achieved in each axis of motion by the various screw constructs. These results underline the various strengths and weaknesses in biomechanical stabilization of different screw constructs. There was significant heterogeneity in the data reported across the studies. Standardized spinal motion segment configuration and injury models may provide more consistent and reliable results.