Factors influencing arthrodesis rates in a rabbit posterolateral spine model with iliac crest autograft.

PURPOSE

The rabbit posterolateral intertransverse spine arthrodesis model has been widely used to evaluate spinal biologics. However, to date, the validity and reproducibility of performance of iliac crest bone graft, the most common and critical control group, has not been firmly established. We evaluated original research publications that utilized this model, identified which experimental conditions affected fusion rates, and developed an algorithm to predict fusion rates for future study designs.

METHODS

A MEDLINE search was performed for publications through December, 2011 that utilized this model to evaluate fusion rates elicited by iliac crest autograft. All study parameters were recorded, and logistic regression analyses were performed to determine the effects of these variables on fusion rates as determined by either manual palpation or radiographs.

RESULTS

Seventy studies with 959 rabbits in 102 groups met the inclusion criteria. Excluding studies that measured fusion at 4 or fewer weeks or intentionally tried to decrease the fusion rate, the overall fusion rate for autograft was 58.3 ± 16.3 % (mean ± SD) as determined by manual palpation and 66.4 ± 17.8 % by plain radiographs. Regression analysis demonstrated a difference between these outcome measures with a trend towards significance (p = 0.09). Longer time points and larger volumes of autograft resulted in significantly greater reported fusion rates (p < 0.0001 and p < 0.05, respectively). Neither strain, age, weight, nor vertebral level significantly affected fusion rates.

CONCLUSIONS

Although experimental conditions varied across studies, time point evaluation and autograft volume significantly affected fusion rates. Despite some variability demonstrated across certain studies, we demonstrated that when the time point and volume of autograft were controlled for, the iliac crest control group of the rabbit posterolateral spinal arthrodesis model is both reliable and predictably affected by different experimental conditions.