There are drawbacks to using threaded cylindrical cages (e.g., limited area
for bone ingrowth and metal precluding radiographic visualization of bone
healing). To somewhat offset these drawbacks, a barbell-shaped cage has bee
n designed. The central core of the barbell can be wrapped with collagen sh
eets infiltrated with bone morphogenetic protein. The obvious theoretical a
dvantages of a barbell cage have to be weighed against potential biomechani
cal disadvantages. Our purpose was to compare the biomechanical properties
of an anterior lumbar interbody reconstruction using 18-mm-diameter threade
d cylindrical cages, with a reconstruction using barbell cages (18-mm diame
ter and 6 mm wide at both cylindrical ends, with around 4-mm-diameter bar j
oining the two ends). Twelve cadaveric lumbar motion segments were tested.
Three L5-S1 segments received two threaded cylindrical cages, and three L5-
S1 segments received two barbell cages. Three L3-L4 segments received one t
hreaded cylindrical cage, and three L3-L4 segments received one barbell cag
e. A series of biomechanical loading sequences were carried out on each mot
ion segment, and stiffness curves were obtained. After the biomechanical te
sting, an axial compressive load was applied to the motion segments until f
ailure. They were then radiographed and bisected through the disc, and the
subsidence (or penetration) of the cage(s) in the cancellous bone of the ve
rtebral bodies was measured. There was no difference in terms of stiffness
between the motion segments with the threaded cylindrical cage(s) inserted
and those with the barbell cage(s) inserted (p > 0.15). The average values
of subsidence was 0.96 min for the threaded cylindrical cage group and 0.80
mm. for the barbell cage group (difference not significant: p = 0.38). The
results suggest that a reconstruction using barbell cages is a biomechanic
ally acceptable alternative to one using threaded cylindrical cages.