In cervical fusion, bone grafts provide both structural and biologic roles. Structurally, they can help bear load anteriorly, reconstruct the anterior column, and restore cervical lordosis. Bone grafts also promote osteoconductive and osteoinductive functions in the fusion process.

Autogenous bone graft fulfills structural and biologic roles necessary for fusion. Usually, autologous bone is taken from the iliac crest for cervical fusion after a discectomy has been done. Cortical fibular grafts and iliac crest have been used frequently for cervical fusion after corpectomy. All these grafts can be fashioned to recreate some lordosis in the cervical spine. In nonosteoporotic patients, they have good compressive strength. Growth factors within the autogenous bone help promote osteoinduction while cancellous trabecular bone forms an osteoconductive substrate.

After surgical implantation, the bone graft undergoes a series of transformations (1). Hematoma forms around the graft and then necrosis occurs during the first 2 weeks. Over the next 2 weeks, the graft becomes vascularized peripherally. The new blood supply brings in mesenchymal cells, which differentiate into osteoblasts, fibroblasts, or chondrogenic cells. Osteoblasts lay down osteoid on the necrotic trabeculae within the graft. The inflammatory cells will remove much of the necrotic tissue. Mineralization and then remodeling of the bone graft occurs as the woven bone is removed and mature lamellar bone is formed. This process of creeping substitution takes about 3 months for cancellous bone. Cortical bone will take longer. During this period, the graft is weakened as it is resorbed. It is critical that new bone is formed to take its place; otherwise, the graft may collapse or break.

Fusion rates for autograft will vary with the construct length, technique used, and plating. Unplated single-level ACDF has a reported fusion rate between 83% and 100%, while multilevel ACDF has fusion rates in the 50% to 75% range. In 1969, Riley et al. (3) retrospectively reviewed the fusion rate and clinical outcome on 93 patients with the Smith-Robinson technique. While 86% had a solid fusion by radiographic criteria, only 72% had a good or excellent result. Other reviews by DePalma and Bohlman demonstrated good results with Robinson technique. DePalma et al. (4) had an 88% fusion rate and 63% good or excellent results. Bohlman et al. (5) also saw an 88% fusion rate and 108/152 of the patients returned to normal activity.

Using a Keystone technique, Simmons and Bhalla (6) reported a very low nonunion rate (3/83 patients) but a similar clinical outcome of 80% good or excellent results.
The high fusion rates with the Keystone technique were also seen in a more recent study by Gore and Sepic in 1984. In their retrospective review of 137 patients, they reported a 97% fusion rate (7).

More recent studies have tried to define the fusion rate based on the length of the fusion construct as fusion rates vary significantly based on the number of levels being fused. Using the Smith-Robinson technique, Wright and Eisenstein found good results with autograft in single-level fusions without instrumentation. Eighty-nine percent fusion rates were seen in 54 patients with autograft in single-level fusions, and 72% fusion rates were seen in 43 patients having two-level fusion. Neck pain improvement was statistically related to ability to achieve fusion. In 90 patients with neck pain, 31% were pain free and 47% were much improved. Only 2/97 patients undergoing fusion had donor site pain (8).

In 1992, Brodke and Zdeblick reported on the results of a modified Smith-Robinson with more complete decortication of the end plates. For single-level autograft fusions, the fusion rate was 97% (32/33), while the fusion rates for two- and three-level fusions were 94% (30/32) and 83% (5/6). Ninety-two percent of their patients reported good or excellent results (9). In 1994, Emery et al. (10) also reported single-level fusion rates in the 90% range with this modification of the Smith-Robinson technique. The added decortication seemed to increase fusion rates particularly for two-level fusions with the Smith-Robinson technique; however, conclusions regarding the three-level fusions could not be drawn due to overall small numbers. Clearly, one-level and to some extent two-level ACDF can be achieved successfully with autograft alone and no plating.

Unfortunately, autograft alone is not as successful in threelevel fusions without instrumentation. Emery et al. retrospectively reviewed 16 patients who underwent modified Smith-Robinson at three levels and found a fusion rate of only 56%. In the nonunion patients, four had moderate to severe pain and three were asymptomatic at an average follow up of 37 months (11). The overall poor fusion rates in multilevel anterior fusion without plating will be demonstrated in multiple comparative studies between allograft and autograft. Based on these results, multilevel (three or more) ventral cervical fusions with autograft and no instrumentation should be avoided.

To increase fusion rates and decrease the incidence of graft collapse and breakage, many surgeons add plate fixation to the anterior fusion. While the use of a plate is still debatable in single-level fusions, plating seems to improve fusion rates in two or more level fusions.

Early studies looking at single-level cervical fusion suggested a decreased fusion rate with plates than without (12). At that time, more rigid plates, which potentially held the disk-graft complex in distraction rather than allowing compression, were thought to be a problem.

Bolesta et al. also looked at cervical fusions with autograft and a plate. For one-level fusion, 11/16 unplated and 2/4 plated fusions healed. In this group, there were 17 good-excellent results. In two-level ACDFs, 15/19 plated fusion and 0/1 unplated fusions healed. This group had 15 good-excellent results. As a result, they concluded that unplated single-level fusions and plated two-level fusion had similar fusion rates and clinical results. For three- and four level cervical discectomy and fusion, the nonunion rates were unacceptably high, and this was not a recommended procedure. In their study, 15 patients with threeor four-level autograft fusions and a plate were followed for at least 2 years. Fifty-eight percent (8/15 patients) did not fuse and three had significant pain to warrant further surgery. Two patients in the union group also had persistent pain and had to have further surgery as well (13).

Similarly, Wang et al. looked at the results of autograft with one-, two-, and three-level cervical fusions. Over a 6-year period, 80 patients undergoing a single level fusion had plated or nonplated surgeries. —95.5% of the plated group and 91.7% of the nonplated group fused. Both groups of patients did well clinically having 88% to 91% good or excellent results. However, as the number of levels increased, the success rate decreased. In two-level fusions, 100% of patients undergoing fusion with autograft and plate fused, whereas 75% of nonplated autograft patients fused. In three-level fusions, the fusion rate was 82% for patients (33/40) with a plate and 63% for patients (12/19) without a plate. Though the difference in fusion rates for three-level fusions was large, it was not statistically significant because of the relatively small number of patients in the three-level group. All 14 of the nonunions from the two- and three-level fusion groups had significant pain and required additional surgery for nonunion (14). In contrast, studies prior to this had suggested that two-thirds of patients with pseudarthrosis had symptoms, and only 25% of them were symptomatic enough to want fusion. Weaknesses of these studies were the lack of clinical outcome data and the fact that the primary surgeon rather than an objective third party evaluated clinical outcome and radiographic fusion. However, the evidence from Bolesta, Wang, and Emery’s studies strongly suggest that autograft and plating are insufficient for patients to achieve bony union across three or more levels anteriorly. Furthermore, though bony fusion correlates with pain relief, a significant number of patients will still have pain even with successful fusion.