Ventral approaches to the cervical spine were first introduced in the 1950s by Smith and Robinson,¹ Cloward,² and Bailey and Bagley.³ Early surgical techniques used autologous bone grafts. The longevity of these procedures is a testament to their efficacy. However, the use of autologous bone graft is unavoidably coupled with donor site morbidity. In an attempt to eliminate donor site morbidity, the use of allograft was proposed as early as the late 1950s. Since that time, allograft has been used in spinal fusions and has met with varying degrees of success. Allograft provides a comparable means to autograft by which to obtain spinal fusion while eliminating donor site morbidity.

Review of the literature reveals that allograft provides fusion rates and clinical outcomes similar to those obtained by autograft for single-level interbody cervical fusions.^4–10 Multilevel interbody fusions are associated with increased nonunion rates and graft complication with both autograft and allograft.^4,5,10,11 Nonunion rates and graft complications have been shown to increase with the number of motion segments fused with both autograft and allograft. ^4,5,10,11

Various modifications in surgical technique and the use of rigid external fixation, dorsal cervical instrumentation, and, most recently, ventral cervical instrumentation have been tried in an attempt to improve fusion rates for multilevel interbody cervical fusions. The use of ventral cervical instrumentation as an adjunct to multilevel interbody cervical fusions has been shown to increase fusion rates, improve clinical outcomes, and reduce graft complication.^12–15 For multilevel interbody cervical fusions, the combination of allograft supplemented with ventral cervical instrumentation has been shown to provide similar fusion rates and clinical outcomes to those obtained with autograft and ventral cervical instrumentation.^12,13,15,16

Single-level or multilevel corpectomies are used for the treatment of degenerative, congenital, traumatic, infectious, and neoplastic disorders of the cervical spine. Stabilization of decompressed segments presents spine surgeons with a formidable challenge. Numerous surgical techniques with various types of grafts have been proposed and used with varying degrees of success. As with multilevel interbody cervical fusions, single-level or multilevel corpectomies have an increasing rate of nonunion and graft complications as the number of motion segments fused increases, for both autograft and allograft.^5,11 The literature is inconclusive regarding the ideal graft for single-level or multilevel corpectomies. Some authors find autograft to be significantly superior to allograft, with increased fusion rates and decreased graft complications.^5,11,17 While other authors have found no difference between allograft and autograft, they have observed an increased time to incorporation of the graft.^6,18,19

Modifications in surgical techniques and the use of rigid external fixation, dorsal cervical instrumentation, and, most recently, ventral cervical instrumentation have been used in an attempt to improve fusion rates in single- or multilevel corpectomies. The addition of ventral cervical instrumentation provides immediate cervical spine stability, improves fusion rates, and limits graft complications.^12–1620

Various authors have found no difference in fusion rates, clinical outcome, and graft complications when comparing allograft and autograft supplemented with ventral cervical instrumentation for single-level and multilevel corpectomies.^16,20,21 Most recently, Mahr et al.²¹ reported radiographic stability in 99.2% of their patients who underwent cervical corpectomies with fibular allograft reconstruction and ventral cervical instrumentation.

Allograft eliminates harvest site morbidity. Donor site morbidity has been reported to range from 9% to 29%.^{2,13,15,16,18,20,21} Harvest site complications include wound hematomas, osteomyelitis, wound infections, lateral femoral cutaneous nerve neurapraxia, fractures of the anterior superior iliac spine, and bowel perforations.^13,17,22 In a study by Rawlinson,²² 36% of patients had persistent donor site pain at 1 year after graft harvest. Rawlinson reported that in 45% of the patients in his study, “donor site pain had delayed their mobilization.”¹⁷ Thirty-one percent of patients in his study felt that they could have gone home sooner if not for donor site pain.¹⁶ Other advantages of allograft include decreased operative time and blood loss.^{9,15,16,20,21}

Allograft carries the potential for transmission of infectious diseases. However, the risk of acquiring HIV infection from allograft has been estimated to be between one in 1 million and one in 1.6 million.^16,21 Strict regulations regarding the testing, harvesting, treatment, handling, and storage of allograft should serve to minimize the risk of acquiring infectious disease via the implantation of allograft.

Finally, the cost of bone graft harvest has been shown to be similar to the cost of purchasing allograft and ventral cervical instrumentation.²³

It is not difficult to argue in favor of the use of autogenous bone for ventral cervical spine fusion. As of 2010, the literature is overwhelmingly in favor, radiographically and clinically, of the use of autograft bone for multilevel fusion.

From a biologic perspective, autogenous bone is far superior to allograft bone in promoting arthrodesis. Autogenous bone from the iliac crest is the ideal graft substrate, because it is osteoinductive, osteogenic, and osteoconductive. The processes of allogeneic bone graft recovery, processing, and sterilization effectively remove all viable components and render allograft bone only osteoconductive. Allograft is simply a structural lattice that relies on host tissue ingrowth for successful union, and this underlies the lower fusion rates when compared to autograft bone.

Multiple studies have directly compared autograft to allograft. In 1990, Zdeblick and Ducker¹ compared allograft and autograft retrospectively for single-level and multilevel anterior cervical discectomy and fusion (ACDF) without a plate. The fusion rate for single-level fusion was 95% for both substrates; however, the fusion rate for two-level ACDF with autograft was 83%, compared with a dismal 37% for allograft. Pooling all patients in this study, the overall fusion rates for autograft and allograft were 92% and 78%, respectively. In a similar, but larger, study, Bishop et al. ² evaluated 132 patients who had undergone single-level or multilevel ACDF with either autograft or allograft without a plate. Allograft again was found to be inferior to autograft in terms of fusion for both single-level fusion (87% vs. 97%) and multilevel fusion (53% vs. 87%). In another retrospective study, Samartzis et al.³ compared allograft with autograft for single-level and multilevel ACDF with plate in 80 patients, and like the previous authors, reported fusion rates in favor of autogenous bone, 100% versus 94%. Suchomel et al.⁴ prospectively evaluated 79 patients who underwent single-level or multilevel ACDF with a plate. The fusion rates for either one- or two-level surgery were not found to be different statistically; however, the authors concluded that fusion takes longer with allograft bone.

Several case series have evaluated fusion with either allograft or autograft with a plate. Kostuik et al.⁵ reported 100% fusion rates for single-level ACDF with a plate and autograft in 42 patients. Similarly, Connolly et al.⁶ reported 100% fusion rates for the same surgery in 43 patients following single-level or multilevel ACDF. Fusion rates of 100% also were reported by Schneeberger et al.⁷ for one-, two-, or three-level ACDF with plate and autograft. Wang et al.⁸ reported a 100% success rate for two-level ACDF with autograft and a plate. Wang et al.⁹ also reported a fusion rate of 82% for three-level ACDF with plate and autograft, in contrast to the 71% fusion rate for a three-level ACDF with plate and allograft reported by Yue et al.¹⁰ In a large retrospective series of one- and two-level ACDF with plate and allograft, Martin et al.¹¹ reported fusion rates of 90% and 72%, respectively.

To be fair, single-level fusion using allograft bone and a plate has been shown to result in reasonable rates of fusion. The highest level of evidence available for the use of allograft bone for single-level ACDF with a plate comes from the cervical disc arthroplasty investigational device exemption (IDE) trials. The control arm in the Prestige IDE trial as reported by Mummeneni et al.¹² had a fusion rate of 98.7%. The fusion rate for single-level fusion with allograft and a plate in the Prodisc-C trial was reported to be 90.2%,¹³ and in the Bryan Disc IDE trial, the fusion rate was 94.3%.¹⁴ Although allograft bone may result in sufficient fusion rates for instrumented single-level fusions, several studies have demonstrated that allograft bone is not the ideal substrate for the more biologically challenging multilevel ACDF.^1,2,10,11

The main criteria for success following ventral cervical fusion are ample neurologic decompression and successful fusion. Nonunion has been associated with suboptimal clinical outcomes in multiple studies. Phillips et al.¹⁵ evaluated 48 patients with known pseudarthrosis following ventral cervical fusion surgery. Thirty-two patients (67%) were symptomatic, with either neck pain and/or radiculopathy, and 22 patients (45%) required secondary surgery due to the nonunion. Similarly, poor outcomes were reported by Wright and Eisenstein.¹⁶ In their series, 18 patients developed pseudarthrosis, and 9 of them (50%) required secondary surgeries. Furthermore, only 60% of the patients who had revision surgeries improved clinically. Nonunion was found to be associated with inadequate outcomes in yet another study by Kuhns et al.¹⁷ In that study, 33 patients were treated with dorsal fusion following failed ventral fusion; follow-up was assessed by the Cervical Spine Outcomes Questionnaire, SF-36, and Arthritis Impact Measurement Scale 2. Although all patients achieved fusion following dorsal revision fusion, 20% of the patients reported “discomforting” pain, and 28% of the patients continued to have “moderate to severe” pain. Clearly, the need to achieve fusion cannot be overemphasized.

An argument invariably made for the use of allograft bone is donor-related pain along the ventral iliac crest. However, one must be very critical of the reports that demonstrate high levels of pain following graft harvest. The wide range of pain that is reported is related to the various methods used to harvest iliac crest bone ventrally, as well as the various methods used to assess postoperative pain. Wright and Eisenstein reported that only 2 of 97 (2%) patients reported donor site pain following single-level or multilevel anterior cervical fusion using iliac crest bone graft. Ventral donor site pain is simple to avoid and is entirely related to the technique used. In our experience, such postoperative pain syndromes are extremely rare when the following technique is used. After a methodical dissection to the ventral iliac crest, the periosteum is incised down to bone with electrocautery, and it is then meticulously reflected subperiosteally with the Bovie to a depth of 1.5 cm medially and laterally along the desired length of the graft to be harvested. The borders of the periosteum are then tagged with #0 Vicryl sutures for later repair. This step is important because it allows easy identification of the periosteum after graft recovery. (The tissues retract medial to the crest if it is not tagged, and can be difficult identify.) The periosteum is richly innervated and must be repaired anatomically. The graft is then harvested in the usual fashion using osteotomes or a sagittal saw. A careful, watertight repair of the periosteum is then performed using the previously placed sutures. This prevents a hematoma; if the suture line leaks it is because it was not repaired properly. Opponents of the use of autogenous bone from this site may state that it takes longer; however, this is not the case. Exposure, harvesting, and closure of the iliac crest can be achieved in 15 to 30 minutes in a well-planned, well-executed surgery in most patients.