Occipitocervical Fusion

Occipitocervical fusion is indicated when the occipito-atlantal joint is unstable, destroyed, or must be crossed to obtain fixation. The categories below are the ones that recur, and the notes column carries what changes the operation in each.

View larger

The goals. Three, in order:

1. Decompress the neural elements, usually indirectly by reducing the deformity with traction or intraoperative distraction; direct ventral decompression by transoral or endonasal odontoidectomy is reserved for irreducible ventral compression
2. Rigid fixation from occiput to cervical spine in the correct alignment
3. Bony fusion, because instrumentation fails eventually without arthrodesis; the graft runs from occiput to laminae

When not to fuse the occiput. If the occipito-atlantal joint is stable and competent, prefer atlantoaxial (Goel-Harms) fusion alone: it preserves the 25 to 30 degrees of flexion-extension at O-C1 and avoids occipital fixation morbidity. Cross O-C1 only when that joint is incompetent, destroyed or assimilated, or when C1 anchors are impossible.

CT of the occiput and C1. Occipital thickness is mapped on sagittal and axial CT to plan screw lengths and entry points; the anatomy that makes this necessary is in Fixation Options below. At C1, measure lateral mass height and width and look for a ponticulus posticus (arcuate foramen) over the vertebral artery groove, present in roughly 15 percent and a contraindication to a broad C1 entry over the arch.

CT of C2: the isthmus and the artery. Mandel's cadaveric study of 205 C2 vertebrae is where the number comes from, and the criterion is less than 5 mm in either height or width, not 5 mm for one and 4 mm for the other, because that is the point at which a 3.5 mm screw becomes technically difficult. A high-riding vertebral artery on that side is a reason to abandon a transarticular or full-length pars screw and drop back to a deliberately short pars screw or a translaminar screw, which avoids the artery altogether. It is not rare: isthmus height was under 5 mm in 11.7 per cent of specimens and width under 5 mm in 2.4 per cent, and on CT Paramore found 18 per cent of 94 patients had a high-riding transverse foramen prohibiting transarticular screws on at least one side, with a further 5 per cent feasible-but-risky, so roughly one patient in five is unsuitable on at least one side. Two asymmetries are worth carrying: the right isthmus is generally the smaller, and low isthmus height was three times commoner in women (18 of 24 specimens).

Why you will see three prevalences for "high-riding vertebral artery". They are three definitions, not a disagreement, and quoting a figure without its criterion is how this gets muddled. Mandel measures the bony isthmus in height and width with a single 5 mm floor and finds roughly 10 per cent at risk. Paramore reconstructs the CT along the intended screw path and judges suitability, giving 18 to 23 per cent unsuitable on at least one side. The widely used Neo criteria, isthmus thickness under 5 mm or internal height under 2 mm, are the most inclusive and identify about one patient in three (see /topics/vertebral-artery-injury for that series). Higher numbers reflect a broader definition, not a more dangerous population: pick one, state it when you quote a percentage, and measure it yourself on the sagittal reconstruction.

CT angiography where a vertebral artery anomaly is suspected: a dominant artery, an anomalous medial loop, a fenestration.

MRI. Three things to read: cord compression, ventral at the odontoid and dorsal at the foramen magnum rim; cord signal change, because myelomalacia predicts incomplete recovery; and pannus in rheumatoid disease, which regresses after solid fusion.

Dynamic films and reducibility. Flexion-extension laterals, only in the cooperative and neurologically intact patient, tell you whether the deformity reduces. A reducible deformity can be corrected by positioning or traction; irreducible ventral compression may need joint distraction (Goel) or anterior decompression.

Craniometry. Measure the clivoaxial angle (normal 145 to 165 degrees), the Chamberlain and McGregor lines for basilar invagination, and the preoperative O-C2 angle, McGregor line to the inferior endplate of C2, on a neutral standing lateral. That angle is the alignment target for surgery.

View larger

Anaesthesia and monitoring. Rheumatoid cranial settling and atlanto-occipital dislocation are cord-at-risk states, and the airway is managed accordingly: awake fibreoptic intubation avoids extension during laryngoscopy. Neuromonitoring (SSEP and MEP) is baselined supine after induction and repeated after prone positioning in Mayfield pins before incision; signal loss after turning mandates repositioning. In rheumatoid patients, stop biologics per rheumatology guidance and anticipate fragile skin and poor bone.

The constructs have moved from onlay graft and wire to rigid screw-rod fixation, and the halo went with them.

View larger

Occipital anchorage. The occiput is thickest at the external occipital protuberance and the midline nuchal keel, up to 12 to 15 mm, and thins rapidly laterally to 3 to 6 mm. Midline bicortical keel screws therefore give the strongest purchase, a plate that forces lateral screws is buying fixation from bone that is not there, and if a screw is forced laterally it is a short unicortical one.

The sinuses. The torcular Herophili lies deep to the external occipital protuberance and the transverse sinuses run laterally along the superior nuchal line, so the screws go into the keel below the superior nuchal line, at least 2 cm below the EOP, drilled incrementally against a depth stop, with bicortical purchase only where the thickness has been mapped on CT.

View larger

View larger

Cervical anchorage. Goel-Harms C1 lateral mass and C2 pedicle screws are the default. The screw-rod construct exists precisely because a high-riding, medialised vertebral artery makes a transarticular (Magerl) screw unsafe, and because it reduces C1 on C2 after the screws are in. Where the artery forbids a C2 pedicle screw, drop to a short pars screw, a translaminar screw, or no screw on that side: the rod construct survives an asymmetric C2.

The caudal extent. Stop at C2 when the anchors are good: every subaxial level added increases stiffness, adjacent segment load and dysphagia risk without fusion benefit. Extend to C3-C5 lateral masses for rheumatoid osteopenia, tumour, or failed C2 fixation.

View larger

View larger

Position. Prone in Mayfield three-pin fixation, head neutral or in the planned reduced position, in reverse Trendelenburg to reduce venous bleeding. Confirm craniocervical alignment on lateral fluoroscopy before draping. Eyes free, abdomen free.

Imaging and equipment. Biplanar fluoroscopy is ideal, or navigation, with neuromonitoring running throughout. An occipital plate system with polyaxial cervical screws, a high-speed burr, and cell salvage available.

Preparation. Midline mark from the inion to the lower cervical spine, prophylactic antibiotics, tranexamic acid per local protocol.

Approach. A midline posterior incision from the external occipital protuberance to below the caudal instrumented level, through the strictly midline avascular raphe of the ligamentum nuchae. Expose subperiosteally: the occiput below the superior nuchal line, the C1 posterior arch, and the C2 lamina and pars. On the cranial surface of the C1 arch stay no more than 12 to 15 mm lateral to the midline, to protect the vertebral artery in the sulcus arteriosus. The zones at risk are the artery in that sulcus and lateral to the C2 pars, the C2 venous plexus around the C1-C2 joint, and the greater occipital nerve.

Reduction. For reducible basilar invagination, skull traction before or during the operation, or Goel-type C1-C2 joint distraction with intra-articular spacers or cages, translates the odontoid caudally and ventrally away from the brainstem. The rods hold the reduction.

Alignment: the critical step. Before final tightening, obtain a true lateral and confirm that the O-C2 angle and the posterior occipitocervical angle match the preoperative neutral standing film, at or slightly more extended, never flexed, because flexion risks dysphagia and airway compromise. Miyata measured the change in O-C2 angle (postoperative minus preoperative) against the cross-sectional area of the oropharynx in 29 fusions. The relationship is linear and the ends of it are unambiguous: every patient whose angle fell more than 10 degrees lost more than 40 per cent of oropharyngeal area and developed dyspnoea, dysphagia or both, and no patient whose angle increased developed either. So the target is a non-negative change from that patient's own preoperative film, not any particular absolute angle. All five affected patients in that series had short constructs: a short fusion is no protection, because the angle is set at the occiput regardless of how far down the construct runs.

Fixation. Occipital plate with bicortical midline keel screws; C1 lateral mass and C2 pedicle screws, or the bailout options; contour and seat the rods; compress or distract as needed; final tighten only once alignment is confirmed.

Grafting. Decorticate the occiput and the C1 and C2 laminae with a burr. Lay a structural corticocancellous iliac crest or rib strut from occiput to laminae, supplemented with local autograft and cancellous chips, and secure it under the construct or with cables.

Closure. Layered, over prominent hardware: approximate the deep muscle to the nuchal fascia and avoid dead space, drain per preference, and close the skin without tension, with nylon sutures worth considering in fragile rheumatoid skin.

Aftercare. A rigid collar for 6 to 12 weeks depending on construct security; an upright lateral radiograph before discharge; a formal swallowing assessment before oral intake if there is any dysphagia; CT at 6 to 12 months if fusion is in doubt.

View larger

Each structure at risk is paired below with where it is met, how it is injured, how the injury is avoided and what to do if it happens. The artery, the sinuses and the angle each have a fixed response that is worth knowing verbatim.

Related pages. The shorter operation, and when it is enough: Atlantoaxial C1-C2 Fusion — including C1-C2 costs you about half of cervical rotation, so do not cross the occiput without a reason. The indications: Basilar Invagination, Atlanto-occipital Dislocation, Atlantoaxial Instability, Atlantoaxial Arthritis, Atlantoaxial Rotatory Subluxation, Odontoid Fractures and the syndromic instabilities in Down Syndrome and Klippel-Feil Syndrome. What is being decompressed: Cervical Myelopathy and Chiari Malformation. The complication that governs screw choice: Vertebral Artery Injury. Non-operative and adjunct immobilisation: Halo Vest Application and Management.

• Global epidemiology of indications varies markedly: rheumatoid craniocervical disease has declined in high-resource settings with biologic therapy but remains a major indication where access to disease-modifying treatment is limited; congenital craniovertebral junction anomalies (atlas assimilation, os odontoideum, basilar invagination) form a large proportion of practice in South Asian centres, where the Goel joint-distraction philosophy was developed; trauma (atlanto-occipital dislocation) dominates in high-energy road trauma populations.
• Society guidance: AO Spine craniovertebral junction knowledge forum publications and AANS/CNS cervical spine guidelines address atlanto-occipital dislocation (early rigid internal fixation, avoid traction) and odontoid pathology; there is no single dedicated O-C fusion guideline — practice is technique- and evidence-driven.
• Registry evidence: craniocervical procedures are low-volume and not captured by arthroplasty registries; evidence rests on institutional series and multicentre spine study groups, which consistently report fusion rates above 95 percent with modern screw-rod constructs.
• Resource-setting variation: where navigation and CT angiography are unavailable, surgeons rely more on translaminar C2 fixation and generous fluoroscopy; halo immobilisation with wiring constructs remains a legitimate option where modern implants are unaffordable, accepting longer immobilisation and lower union rates.

• Occipital plate versus condyle screws: condyle screws lower profile and preserve squamous occiput but carry hypoglossal canal and VA risk; keel plates remain standard.
• Posterior-only reduction versus anterior decompression for basilar invagination: the Goel distraction school treats most cases posteriorly; some centres still favour endonasal/transoral odontoidectomy for truly irreducible ventral compression — the threshold for "irreducible" is contested.
• Caudal extent of fusion: stop at C2 versus routine extension to C3/C4 in poor bone — longer constructs are stiffer but increase adjacent segment load and dysphagia.
• Allograft, BMP, and graft strategy: structural autograft remains gold standard; off-label BMP use at the craniocervical junction is debated because of swelling risk near the airway.
• Prophylactic O-C fusion in asymptomatic Down syndrome instability: most advocate observation with activity modification unless neurological signs or marked instability — thresholds vary between guidelines.