J Neurosurg Spine 13:451–460, 2010

Normal anatomical variations of the V3 segment of the

vertebral artery: surgical implications

Laboratory investigation
Arthur J. Ulm, M.D.,1 Monica Quiroga, M.D., 3 Antonino Russo, M.D., 2
Vittorio M. Russo, M.D.,1,2 Francesca Graziano, M.D.,1,2 Angel Velasquez, M.D., 4
and Erminia Albanese, M.D. 2

Department of Neurosurgery, Louisiana State University, New Orleans, Louisiana; 2Department of

1

Neurosurgery, University of Catania, Italy; 3Department of Neurosurgery, University of Costa Rica, San José,
Costa Rica; and 4Department of Neurosurgery, M. Catarino Rivas Hospital, San Pedro Sula, Honduras

Object. The authors undertook this cadaveric and angiographic study to examine the microsurgical anatomy of
the V3 segment of the vertebral artery (VA) and its relationship to osseous landmarks. A detailed knowledge of these
variations is important when performing common neurosurgical procedures such as the suboccipital craniotomy and
the far-lateral approach and when placing atlantoaxial instrumentation.
Methods. A total of 30 adult cadaveric specimens (59 sides) were studied using magnification × 3 to × 40 after
perfusion of the arteries and veins with colored silicone. Seventy-three vertebral angiograms were also analyzed. The
morphological detail of the V3 segment was described and measured in both the cadavers and angiograms. Transar-
ticular screws were placed into 2 cadavers and the relationship of the trajectory to the V3 segment was analyzed.
Results. The authors identified 4 sites along the V3 segment that are anatomically the most likely to be injured
during surgical approaches to the craniovertebral junction. In 35% of the cadaveric specimens the vertical portion of
V3 formed a posteriorly oriented loop that could be injured during surgical exposures of the dorsal surface of C-2. The
mean distance from the midline to the most posteromedial edge of the loop was 25.6 ± 3.5 mm (range 20–35 mm) on
the left side and 30.4 ± 3.8 mm (range 23–36 mm) on the right side. On lateral angiograms, this loop projected poste-
riorly, with a mean distance of 9.8 ± 3.5 mm (range 0–15.7 mm) on the right side and 11.7 ± 1.2 mm (range 10–13.6
mm) on the left side. The horizontal segment of V3 can be injured when exposing the lower lateral occipital bone and
when the C-1 arch is exposed. The mean distance from the inferior border of the occipital bone to the superior surface
of the horizontal segment of V3 was 6 ± 2.8 mm on the right side and 5.6 ± 2.3 mm on the left. In 12% of cases the
authors found no space between the horizontal portion of V3 and the occipital bone. The medial edge of the horizontal
segment of V3 was located 23 ± 5.5 mm (range 10–30 mm) from the midline on the right side and 24 ± 5.7 mm (range
15–32 mm) on the left side. The transition between the V2–V3 segments after exiting the C-2 vertebral foramen is the
most likely site of injury when placing C1–2 transarticular screws or C-2 pars screws.
Conclusions. The normal variation of the V3 segment of the VA has been described with quantitative measure-
ments. An awareness of the anatomical variations and the relationships to the surrounding bony anatomy will aid in
reducing VA injury during suboccipital approaches, exposure of the dorsal surfaces of C-1 and C-2, and when placing
atlantoaxial spinal instrumentation. (DOI: 10.3171/2010.4.SPINE09824)

Key Words      •      vertebral artery      •      V3 segment      •      microsurgical anatomy      •     

atlantoaxial complex      •      craniovertebral junction      •      vertebral artery injury

A
natomically, the VA has been described as having
4 segments.2,10,11 The first segment (V1) extends
from its origin to the C-6 transverse process. The
second segment (V2) includes the part of the artery that
courses through the C-6 to C-2 transverse processes. The
third segment (V3) extends from the C-2 transverse pro-
Abbreviations used in this paper: PICA = posterior inferior cer-
ebellar artery; VA = vertebral artery.
cess to the artery’s entry through the dura mater. Finally,
the fourth segment (V4) is formed by the intradural part
of the artery up to its confluence with the basilar artery
(Figs. 1 and 2). The V3 or suboccipital segment can be
further subdivided into 3 parts.12 The vertical part is lo-
cated between the C-2 and C-1 transverse processes and
contains a proximal loop. The horizontal part is formed
by the segment of the artery that occupies the groove of
the posterior arch of the atlas, and it projects superome-

J Neurosurg: Spine / Volume 13 / October 2010 451

 A. J. Ulm et al.

Fig. 1.  Cadaveric dissection of the VA in different angles of view. Anatomically, the VA has been described as having 4 seg-
ments. The first segment extends from its origin to the C-6 transverse process. The second segment includes the part of the
artery that courses through the C-6 to C-2 transverse processes. The third segment extends from the C-2 transverse process to
the artery’s entry through the dura mater. Finally, the fourth segment is formed by the intradural part of the VA up to its confluence
with the basilar artery.   A: Cadaveric dissection of anterior cervical spine and craniovertebral junction. The vertical segment
of V3 of both VAs extended superiorly from the C-2 transverse process, through the C-1 transverse process. When the VA exits
the C-2 transverse foramen, the artery courses within a groove formed along the lateral and inferior aspect of the C-2 pars. At
this point, the artery forms a bend prior to exiting the groove (dashed line). After exiting the C-2 groove, the artery projects later-
ally a variable distance and forms a proximal loop.   B: Cadaveric dissection (lateral view) of cervical spine and craniovertebral
junction. The V2 is shown with its entry point into the C-2 transverse foramen. The vertical part of V3 is located between the C-2
and C-1 transverse processes and contains a proximal loop. The horizontal part of V3 is formed by the segment of the artery
that occupies the groove of the posterior arch of the atlas and projects superomedially from the C-1 arch en route to the dura
mater.   C: Cadaveric dissection of the right side of craniovertebral junction (posterior view). In a few cases (5 of 34 cadaveric
dissections) the horizontal segment of V3 coursed dorsally through a medial bony foramen (foramen arcuale).   D: Cadaveric
dissection (oblique view) of craniovertebral junction. The proximal loop is shown with its lateral extension away from the level of
the C1–2 facet articulation. The distal loop travels over the C-1 lamina and is in close proximity to the inferior border of the oc-
cipital bone. The vertical red dotted line, parallel or slightly medial to the occipitomastoid suture, represents a surgical incision for
retrosigmoid and far-lateral approaches and the horizontal black dotted line coincides with a tangent line to the foramen magnum.
The 2 lines’ intersection point corresponds to the site of potential injury of the V3 horizontal segment. A. = artery; Atl. = atlanto;
Br. = branch; Digast. = digastric; Dist. = distal; Dors. = dorsal; Ext. = external; For. = foramen; Junct. = junction; L. = loop; Musc.
= muscular; Occip. = occipital; Occipitomast. = occipitomastoid; Parietomast. = parietomastoid; Plex. = plexus; Proc. = process;
Prot. = protuberance; Prox. = proximal; Sup. = superior; Trans. = transverse; Vent. = ventral; Vert. = vertebral.

452 J Neurosurg: Spine / Volume 13 / October 2010

The V3 segment of the vertebral artery

Fig. 2.  Cadaveric dissection of the V3 segment of a left VA in different angles of view. The V3 or suboccipital segment can
be subdivided into 3 parts. The vertical part is located between the C-2 and C-1 transverse processes and contains a proximal
loop. The horizontal part is formed by the segment of the artery that occupies the groove of the posterior arch of the atlas, and
the oblique part extends up to the point where the artery pierces the dura mater. The horizontal segment extending over the C-1
arch forms the distal loop.   Left: Posterior view. The ventral and dorsal ramus of C-2 have been cut. The horizontal segment of
V3 courses dorsally and medially over the lamina of C-1 after exiting the C-1 transverse process. The distal loop of the horizontal
segment wraps around the occipital condyle and projects above the C-1 arch prior to coursing anterior, superior and medially
as the oblique segment en route to the dura mater. The C-2 ganglion is located just below the inferior border of the C-1 lamina
at the level of the C1–2 facet joint. The internal carotid artery and the internal jugular vein are located anterior and lateral to the
vertical part of V3.   Right: Oblique view. The V2–V3 junction lies between the exit of the artery from the C-2 transverse foramen
and the exit at the lateral border of the C-2 groove. This transition segment runs along a groove in the inferolateral border of the
C-2 pars or isthmus. There is great variability of this segment depending on the size and tortuosity of the vessel. Car. = carotid;
For. = foramen; Gang. = ganglion; Int. = internal; Jug. = jugular; V. = vein.

dially from the C-1 arch en route to the dura mater. The
horizontal segment of V3 forms a distal horizontal loop
(Fig. 2).
Iatrogenic injury to the third segment can occur dur-
ing surgical procedures that involve the occipitoatlanto-
axial region.1,13,22,24,28,31,32,34,37 The risk of injury can be re-
duced with the aid of preoperative imaging that includes
the evaluation of osseous cervical structures and vascu-
lar anatomy.14,28,35,37 However, these imaging techniques
are not routinely performed for all approaches and are
not available at all institutions or in emergent situations.
Therefore, the surgeon must possess detailed knowledge
of the anatomy and variability of the V3 segment of the
VA to reduce the risk of complications related to surgical
approaches and procedures involving the posterior cran-
iovertebral junction.
The purpose of this study is to anatomically describe
the V3 segment of the VA and to highlight the potential
sites of injury that can occur during common neurosurgi-
cal procedures involving this region.
Methods
A total of 30 adult cadaveric specimens (59 sides)
whose arteries and veins were infused with colored sili-
cone were used in this study. Using × 3 to × 40 magnifica-
tion, the V3 segments of the VA were dissected on each
side. In 17 specimens (34 sides) the following quantita-
tive anatomical measurements were made: the diameter
of the VA measured from the outer surface, the distance
between the occipital bone and the horizontal part of V3,
and the distance between the most posteromedial edge
of the horizontal and vertical VA and the posterior mid-
line. In 13 additional specimens (25 sides) the distance
and angle of lateral projection of the proximal loop were
measured from the transverse foramen of C-2 to the most
lateral position of the loop between the C-1 and C-2 trans-
verse processes. Also the relationship of the V2-V3 junc-
tion to the C-2 pars was quantified, including the presence
and degree of angle of the transition segment.
The angiographic anatomy of V3 was also analyzed.
The results of consecutive vertebral angiographic exami-
nations, performed by the senior author (A.J.U.) between
2007 and 2008 on a biplanar angiography system (Axiom
Artis, Siemens), were studied using the hospital’s radiologi-
cal electronic database. The angiograms were obtained for
routine clinical indications, including aneurysm, vascular
malformations, transient ischemic attacks, or acute stroke.
Angiograms obtained in patients with vascular pathology
in the vertebrobasilar system or patients younger than 18
years of age were excluded from this study. Forty digitally
subtracted angiograms that included the entire V3 segment
were incorporated into the study. The angiography was
performed with the patients’ heads in neutral position. On
lateral view angiograms, the distance of the proximal loop
was measured from its most posterior margin on a perpen-
dicular line that intersects a line extending from the caudal
to the rostral bases of the loop. On the anteroposterior view
angiograms, the length of the lateral extension of the proxi-
mal loop was measured. Additionally, 72 unsubtracted VA
angiograms (obtained in 48 patients) were used to calculate
the diameter of the VA, measured from the intimal sur-
face, and the distance between the superior border of the
horizontal segment of V3 and the inferior border of the oc-
cipital bone.
The mean, range, and SD were calculated for all of

J Neurosurg: Spine / Volume 13 / October 2010 453


the cadaveric and angiographic measurements. The statis-
tical analysis was performed by the analysis of variance.
Probability values < 0.05 were considered significant and
those < 0.01 were considered highly significant.
Results
In the 30 cadaveric specimens, all the V3 segments of
the right and left VAs extended superiorly from the C-2
transverse process, through the C-1 transverse process,
and then proceeded dorsally over the vertebral groove,
finally passing through the dura between the foramen
magnum and the posterior arch of C-1 (Figs. 1 and 2).
No V3 segment anomalies such as a fenestrated VA or a
persistent first intersegmental artery were found in our
specimens. One extradural origin of the PICA was found
in the 59 cadaveric dissections and in 1 of the 43 angio-
graphic examinations. The range, mean, and SDs for each
anatomical parameter measured are presented in Table 1.
The angiographic results are displayed in Table 2.
The Vertical Portion of V3
Anatomical Study. After passing through the C-2
transverse foramen, the VA first ascends and then makes
a lateral bend before ascending en route to the C-1 trans-
verse foramen. When the VA exits the C-2 transverse fo-
ramen, the artery courses within a groove formed along
the lateral and inferior aspect of the C-2 pars.17 At this
point, the artery forms a bend prior to exiting the groove
(Figs. 1 and 2). Half of the VA specimens presented a
90° bend, 37.5% made a slightly more acute angle, and
TABLE 1: Anatomical measurements
Measurement
distance from most medial edge of proximal
loop to dorsal midline
diameter of the horizontal loop of the V3
segment
distance btwn horizontal VA & occipital bone
distance btwn most medial edge of distal
loop & dorsal midline
454
A. J. Ulm et al.
12.5% formed an obtuse bend inside the groove. After
exiting the C-2 groove, the artery projects laterally a vari-
able distance before ascending toward the C-1 transverse
foramen. The segment between the C-1 and C-2 trans-
verse foramina forms a proximal loop (Figs. 1 and 2). The
average length of the lateral projection of the proximal
loop from its exit from the C-2 transverse groove was 16
± 2.0 mm.
In 25 V3 dissections we looked specifically at the
proximal loop, which is tethered between the bony C-1
and C-2 foramina (Figs. 1A and 2A). The proximal loop
projected posteriorly in 12 cases (48%) and directly later-
ally in 5 cases (20%); in 8 cases (32%) there was little or
no loop present and the artery coursed in a straight line
between C-1 and C-2. There was no case in which the
proximal loop projected medially or anteriorly. The mean
distance from the most medial edge of the loop, along the
lamina of C-2, to the dorsal midline was 25.6 ± 3.5 mm
(range 20–35 mm) on the left and 30.4 ± 3.8 mm (range
23–36 mm) on the right.
Angiographic Study. Forty vertebral angiograms that
included the V3 segment were reviewed. On the lateral
views, all but one of the studies demonstrated a bend ori-
ented posteriorly at the vertical portion of V3. The dis-
tance of this bend from its most posterior margin on a
perpendicular line that intersects a line extending from
the caudal to the rostral bases of the loop ranged from
0 to 15.7 mm (mean 9.8 ± 3.5 mm) on the right side and
from 10 to 13.6 mm (mean 11.7 ± 1.2) on the left side.
The specimen that did not have a posterior angiographic
bend was one in which the bend extended laterally with-
Right (mm) Left (mm)
Mean ± SD Range Mean ± SD Range
30.4 ± 3.8 23–36 25.6 ± 3.5 20–35
4.6 ± 1.4 1–7 5.1 ± 1.5 2–8
6 ± 2.8 0–10 5.6 ± 2.3 0–10
23 ± 5.5 10–30 24 ± 5.7 15–32
J Neurosurg: Spine / Volume 13 / October 2010
The V3 segment of the vertebral artery
TABLE 2: Angiographic measurements*

Right (mm) Left (mm)

Measurement Mean ± SD Range Mean ± SD Range

A 9.8 ± 3.5 0–15.7 11.7 ± 1.2 10–13.6

B 4.0 ± 1.3 2.6–8.1 4.1 ± 1.0 1.4–6

C 6.4 ± 3.9 0–13.9 8.3 ± 2.9 0–13.9

*  The measurements A, B, and C, as illustrated here on lateral-view unsubtracted VA angiograms (red letters), are defined as
follows. A: The length of the proximal loop at the vertical segment (right end of double-headed arrow), measured as the distance
from its most posterior margin on a perpendicular line that intersects a line extending from the caudal to the rostral bases of the
loop. B: The diameter of the VA. C: The distance between the most rostral border of the horizontal segment of the VA and the
inferior border of the occipital bone.

out an anterior or posterior deviation (Fig. 3A). On the AP
angiograms, the vertical segment extended laterally 13.3
± 3.2 mm (range 9.5–18.1 mm) before turning rostrally
toward C-1.
The Horizontal and Oblique Portions of V3
Anatomical Study. The V3 segment enters the trans-
verse foramen of C-1 and turns medially, coursing over
the superior surface of C-1 for a variable distance before
coursing anteromedially and superiorly toward its entry
into the dura mater (Figs. 1 and 2). The mean diameter of
the horizontal segment was 4.6 ± 1.4 mm (range 1–7 mm)
for the right VA and 5.1 ± 1.5 mm (range 2–8 mm) for the
left. There was no statistically significant difference be-
tween the mean diameters of the right and left VAs. The
mean distance between the occipital bone and the superior
border of the horizontal part of V3 was 6 ± 2.8 mm (range
0–10 mm) on the right and 5.6 ± 2.3 mm (range 0–10
mm) on the left. In 4 of the 34 sides studied (3 right sides,
1 left) we found no space between the occipital bone and
the horizontal segment. The VA coursed dorsally through
a fully foramen-shaped arterial groove25 in 5 specimens
(4 left sides, 1 right) (Fig. 1C). The mean distance from
the most medial edge of the VA on its groove, along the
lamina of the atlas, to the dorsal midline was 24 ± 5.7 mm
(range 15–32 mm) on the left, and 23 ± 5.5 mm (range
10–30 mm) on the right.
Angiographic Study. The mean diameter of the VA
was 4.0 ± 1.3 mm (range 2.6–8.1 mm) on the right side
and 4.1 ± 0.1 mm (range 1.4–6 mm) on the left side. The
mean distance between the occipital bone and the supe-
rior border of the horizontal VA was 6.4 ± 3.9 mm (range
0–13.9 mm) on the right and 8.3 ± 2.9 mm (range 0–13.9
mm) on the left. In 6.9% of the specimens (4 right sides,
1 left) we found no space between the occipital bone and
the VA (Fig. 3D). In 2.7%, a unilateral extradural origin
of the PICA was found (Fig. 3C).
Discussion
Surgical approaches to the posterolateral craniover-
tebral junction entail working in close proximity to the V3

J Neurosurg: Spine / Volume 13 / October 2010 455

 A. J. Ulm et al.

Fig. 3.  A: Right VA angiogram (anteroposterior view). The VA projects laterally from its exit from the C-2 groove, ascends
between C-1 and C-2 prior to traversing the C-1 transverse foramen. After exiting the C-1 transverse process, the artery courses
medially over the C-1 arch before turning anteriorly, medially, and superiorly as the oblique segment that enters the lateral dorsal
surface of the spinal dura.  B: Right VA angiogram (lateral view). The proximal loop projected posteriorly in all but one direct
lateral angiogram.  C: Left VA angiogram (lateral view) showing the PICA originating off the extradural horizontal segment
of V3. In 2.7% of VA angiograms, the PICA had an extradural origin.  D: Left VA angiogram (lateral view) demonstrating the
relationship of the horizontal segment/distal loop to the lower occipital bone. In 12% of sides and 6.9% of lateral angiograms
demonstrated no space between the lower occiput and the upper surface of the horizontal segment.  E: Left VA angiogram
(lateral view). Demonstrates pseudoaneurysm caused by iatrogenic injury to the distal loop during a far-lateral approach. The
injury was the result of unipolar cautery used off midline while exposing the C-1 arch. B. = basilar; Horiz. = horizontal; Pseudoan.
= pseudoaneurysm; Segm. = segment.

segment of the VA. This segment may be injured during
approaches or procedures that involve fusion or decom-
pression of the craniovertebral junction, or while expos-
ing the V3 segment when treating vascular pathology and
tumors.4,7,13,15,17,32,33,37 Injury to the VA, especially when
the dominant artery is involved, may result in serious
complications such as VA occlusion, formation of an ar-
teriovenous fistula, a pseudoaneurysm, or massive bleed-
ing that could result in stroke or even death.6,22,26,27 When
VA injury occurs and ligation is mandatory, the risk of
ischemia-related neurological injury depends on, among
other factors, the size of the contralateral artery and its
compensatory flow capability. Not only is VA asymmetry
a normal variant, but it can also accompany congenital
skeletal anomalies such as anomalies of the craniover-
tebral junction.29,30,36 No V3 segment anomalies such as a
fenestrated VA or a persistent first intersegmental artery
were found in our specimens. An incidence rate of 5.4%
of V3 segment anomalies has been previously described
based on CT angiography.14
Preoperative planning using high-resolution CT scan-
ning and imaging of the VA is important in preventing
VA injury.13,14,18,19,23 However, a detailed knowledge of VA
anatomy, the normal variants of the VA, and its relation-
ship to nearby osseous structures is critical when expos-
ing the lateral craniovertebral junction, when performing
procedures requiring exposure of the V3 segment, and
when placing C-2 spinal instrumentation.
The anatomy of the V3 segment is complex. The seg-
ment extends from the C-2 VA foramen to the dura mater.
The V3 segment consists of a vertical portion between C-2
and C-1, a horizontal portion that courses over the con-

456 J Neurosurg: Spine / Volume 13 / October 2010

The V3 segment of the vertebral artery

dyle and the upper surface of the C-1 arch, and an oblique
portion that ascends from the C-1 arch to penetrate the
dura. The V3 segment has redundancy that results in loops
and bends in the artery, which accommodate the mobility
of the head with rotation. The vertical and horizontal seg-
ments contain, respectively, a proximal and a distal loop.
These loops have the potential to bring the V3 segment
into close proximity to the occipital bone, the posterior
surface of the C-2 vertebral body, and the midline along
the C-1 arch (Figs. 1 and 2). In addition, the relationship
of the V2–V3 junction to the pars of C-2 impacts the safety
of C-2 instrumentation (Fig. 4). The purpose of this study
was to characterize the normal anatomical variations of
the V3 segment with paying particular attention to the ef-
fect these loops may have on common surgical approach-
es to the region.
Sites of Potential Injury
There are 4 sites along the normal course of the V3
segment that anatomically constitute the points of great-
est risk of iatrogenic injury during surgery involving the
craniovertebral junction (Figs. 4 and 5).
Medial Edge of the Horizontal V3 Segment Along the C-1
Arch
The horizontal portion of the V3 segment begins
as the artery exits the C-1 transverse foramen and ends
where the artery penetrates the dura. In between, the ar-
tery courses along the superior aspect of the posterior C-1
arch. The distal horizontal loop lies between the exit of
the V3 segment from the C-1 transverse foramen and the
point where it courses anteriorly, medially, and superiorly
toward the dura mater. The medial border of the distal Fig. 4.  The V2–V3 transition zone occurs above the C-2 transverse
horizontal loop can be injured when exposing the C-1 foramen and medial to the exit of the V3 segment lateral to the C-2 body.
In between, the artery lies adjacent to the C-2 groove, which is infero-
arch (Fig. 5). In 29 of 34 cadaveric dissections, the hori- lateral to the C-2 pars or isthmus. The V2–V3 transition zone (asterisk)
zontal loop coursed within a groove on C-1, while in the coincides with the fourth site along the V3 segment prone to iatrogenic
other 5 dissections the artery coursed through a medial injury.   A: Ideal screw trajectory in a normal VA. The relationship be-
bony foramen (Fig. 1C). The medial edge of the distal tween the C-2 isthmus and the screw trajectory is shown.   B: Slightly
horizontal loop was found 23 ± 5.5 mm from midline on inferolateral trajectory combined with a “high-riding” VA. The V2–V3
the right and 24 ± 5.7 mm from midline on the left, with junction above the C-2 transverse foramen is the site of VA injury when
a range between 10 and 32 mm. This medial border of the placing C-2 pars or C1–2 transarticular screws.
distal loop can be injured when exposing the posterior ment (Fig. 3D). The superior border is at greatest risk with
arch of C-1 during midline procedures such as Chiari de- paramedian incisions carried out to expose the foramen
compression and C1–2 spinal surgery and when exposing magnum or lower cerebellopontine angle (Figs. 1D and
the midline occiput (Fig. 5).8,9 Care must be taken when 6A).8,31,32 The VAs of elderly patients and dominant VAs
exposing the C-1 lamina from medial to lateral, espe- have the tendency to be ectatic, to bulge out from the C-1
cially when using unipolar cautery (Fig. 3E). Any lateral groove, and therefore to be at greater risk during lateral
exposure of the craniovertebral junction also places this exposure of the posterior skull base. Commonly, the ret-
point on the artery at risk of injury. rosigmoid approach requires dissection down to the level
Superior Border of the Horizontal V3 Segment Along the of the foramen magnum. Two common instances include
C-1 Arch retrosigmoid craniotomies for microvascular decompres-
sion of the seventh cranial nerve and for large tumors such
The superior surface of the horizontal segment is in as acoustic neuromas (Fig. 6A).32 In these instances, care
close proximity to the lower lateral surface of the occipi- must be taken to avoid unipolar cautery when approach-
tal bone (Figs. 3D and 6B and C). On average, we found ing the C-1 arch. The horizontal segment lies along a line
the superior surface of this segment 6 ± 3 mm beneath that runs parallel and lateral to the foramen magnum and
the occipital bone. Surprisingly, in 12% of cadaveric dis- inferior and medial to the mastooccipital suture (Fig. 1D).
sections (4/39) and in 7% of direct lateral angiograms, To minimize the risk of VA injury during the retrosig-
we found no separation between the lower surface of the moid approach, we meticulously dissect the muscular
occiput and the superior surface of the horizontal V3 seg- layers using bipolar cautery and Metzenbaum scissors in

J Neurosurg: Spine / Volume 13 / October 2010 457

 A. J. Ulm et al.

a layer-by-layer fashion when approaching the foramen

Fig. 5.  Three of 4 sites along the V3 segment at greatest risk of iatro-
genic injury during surgical approaches to the craniovertebral junction
(each indicated by an asterisk): 1) Medial border of the distal horizontal
loop. Exposure of the C-1 arch places this site at risk. Particular care
must be used when using unipolar cautery as the dissection proceeds
from medial to lateral along C-1. 2) Superior surface of the distal hori-
zontal loop. Injury can occur to this segment while performing soft-
tissue dissection during the exposure of the lateral suboccipital region.
The segment often bulges posteriorly into the suboccipital muscular tri-
angle and is particularly prone to injury when extending exposure down
to the level of the foramen magnum. 3) Medial border of the proximal
vertical loop. It projects posteriorly toward the dorsal surface of C-2 in
more than two-thirds of cases. Injury can occur during the exposure
of the lateral dorsal surface of C-2, particularly when the exposure ex-
tends beyond the lateral border of C-2.
magnum and craniovertebral junction. The far-lateral ap-
proach is often performed using a paramedian incision,
and care must likewise be taken when performing the mus-
cle dissection at the level of the craniovertebral junction
to avoid injuring the VA along the C-1 groove (Fig. 6C).
Patient positioning can minimize the risk to the horizontal
segment. It is important to open up the interval between the
artery and the occipital bone with adequate neck flexion,
head rotation, and dropping the vertex of the head toward
the floor. These maneuvers displace the superior surface of
the horizontal segment away from the lower occiput.5,11,16
Medial Surface of the Vertical Segment Lateral to the
Dorsal Surface of C-2
The V3 segment begins as the artery exits the groove
along the medial border of the C-2 pars (Fig. 2). We found
a proximal loop in the artery between C-1 and C-2 in 70%
of cadaveric dissections. The proximal loop is tethered
between the exit from the C-2 groove and the entrance
into the C-1 transverse foramen. The proximal loop was
found to course posteriorly toward the dorsal surface of
C-2 in approximately half of the cadaveric specimens. In
cases of large, tortuous arteries, the posterior bend pro-
jected to the level of the dorsal surface of C-2 (35% of the
cadaveric specimens, 48% of the angiograms). Exposure
of the lateral dorsal surface of C-2, particularly when us-
ing unipolar cautery, places this segment at risk for iatro-
genic injury. Care must be taken when extending expo-
sure beyond the lateral dorsal surface of C-2 to prevent
injury to this segment (Fig. 5).
Transition of V2–V3 at the Level of the C-2 Transverse
Foramen
The VA transitions between the V2 and V3 segments

Fig. 6.  A: Superficial anatomical dissection of the suboccipital region. The superficial muscular layer (sternocleidomastoid,
trapezium, semispinalis, splenius capitis major, and longissimus capitis muscles) has been removed. The horizontal part of the
V3 segment of the VA and the C-1 arch are visible through the suboccipital muscular triangle. The latter is formed superiorly and
medially by the rectus capitis posterior major muscle, superiorly, and laterally by the superior oblique muscle, and inferiorly by
the inferior oblique muscle. The rectus capitis posterior major and the inferior oblique muscles attach at the midline to the C-2
spinous process, while both oblique muscles attach laterally to the C-1 transverse process. The asterisk shows the site of poten-
tial iatrogenic injury to the horizontal V3 segment while performing the ideal surgical incision for retrosigmoid and far-lateral ap-
proaches (black dotted line).   B: The suboccipital muscles have been removed to show the V3 segment, the surrounding venous
plexus, and the C-2 nerve exiting between the C-1 and C-2 laminae. Note the proximity of the artery to the occipital bone.   C:
Far-lateral craniotomy with removal of the C-1 posterior arch. The V3 horizontal distal loop wraps around the posterior occipital
condyle. The C-1 transverse foramen has been unroofed, which allows mobilization of the VA, improving access to the condyle
for removal. Cap. = capitis; Inf. = inferior; Lat. = lateral; Lev. = levator; M. = muscle; Maj. = major; Min. = minor; Musc. = muscular;
Obl. = oblique; Post. = posterior; Rec. = rectus; Scap. = scapula; Sig. = sigmoid.

458 J Neurosurg: Spine / Volume 13 / October 2010

The V3 segment of the vertebral artery
at the level of the C-2 vertebral body. The V2 segment
ends at the C-2 transverse foramen and courses a vari-
able distance within a groove along the lateral surface of
the C-2 pars before exiting as the V3 segment (Figs. 1A
and 2B). The segment of artery running within the C-2
groove is a transition segment. Significant variability is
found within this segment. We found it to form a 90° bend
in 50% of cadaveric dissections, a bend of less than 90°
in 37.5%, and a bend of more than 90° in 12.5%. Large-
caliber, dominant VAs and ectatic VAs of the elderly are
often referred to as “high-riding” VAs, so defined when
the isthmus height at C-2 is less than 5 mm or the internal
height of the lateral mass (measured from the roof of the
VA groove to the surface of the superior facet) is less than
2 mm.3,36 These high-riding arteries project superiorly
into the inferolateral surface of the C-2 pars prior to turn-
ing inferolaterally and exiting the C-2 groove (Fig. 4B).
These arteries are associated with a narrowed isthmus of
the C-2 pars. The transition segment between the V2 and
V3 segments, within the C-2 groove is the site of injury
when placing C1–2 transarticular screws. The combina-
tion of a high-riding dominant VA with a narrowed C-2
isthmus poses the greatest risk when placing C-2 instru-
mentation.18,21,23,36 The atlantoaxial joint lies deep to the
VA and the point along the C-2 pars where the VA runs
within the C-2 groove. Therefore, an equal amount of care
should be given to this anatomy when placing either C-2
pars screws or C1–2 transarticular screws, as this segment
can be injured with either technique.1,13,17,20,24,34 The most
likely mistake leading to a VA injury, when planning the
screw trajectory for the placement of C1–2 transarticular
screws, is a trajectory that deviates laterally or inferiorly
(Fig. 4B). In cases of a narrow isthmus, C-2 pars screws
can be used safely as long as the depth of the screw does
not extend to the level of the C-2 groove. A safe screw
depth for a C-2 pars screw can usually be determined by
studying a CT reconstruction in the sagittal plane.
Conclusions
The V3 segment of the VA has a variable normal anat-
omy that impacts common surgical procedures involving
the craniovertebral junction. We identified 4 sites along
the V3 segment that are anatomically the most likely to be
injured during surgical approaches to the craniovertebral
junction. Injury to this segment of the VA can be greatly
reduced with a thorough understanding of the normal
variations of the V3 segment.
Disclosure
The authors report no conflict of interest concerning the mate-
rials or methods used in this study or the findings specified in this
paper.
Author contributions to the study and manuscript prepara-
tion include the following. Conception and design: Ulm, A Russo,
VM Russo, Graziano, Albanese. Acquisition of data: Quiroga, A
Russo, VM Russo, Graziano, Velasquez, Albanese. Analysis and
interpretation of data: all authors. Drafting the article: Ulm, Quiroga,
A Russo, VM Russo, Graziano, Albanese. Critically revising the
article: Ulm. Reviewed final version of the manuscript and approved
it for submission: all authors. Statistical analysis: A Russo, VM
Russo, Graziano, Albanese. Administrative/technical/material sup-
port: Ulm. Study supervision: Ulm.
J Neurosurg: Spine / Volume 13 / October 2010
References
  1.  Aota Y, Honda A, Uesugi M, Yamashita T, Baba N, Niwa T, et
al: Vertebral artery injury in C-1 lateral mass screw fixation.
Case illustration. J Neurosurg Spine 5:554, 2006
  2.  Argenson C, Francke JP, Sylla S, Dintimille H, Papasian S, di
Marino V: The vertebral arteries (segments V1 and V2). Anat
Clin 2:29–41, 1980
  3.  Bloch O, Holly LT, Park J, Obasi C, Kim K, Johnson JP: Effect
of frameless stereotaxy on the accuracy of C1-2 transarticular
screw placement. J Neurosurg 95 (1 Suppl):74–79, 2001
  4.  Bruneau M, Cornelius JF, George B: Antero-lateral approach
to the V3 segment of the vertebral artery. Neurosurgery 58 (1
Suppl):ONS29–ONS35, 2006
  5.  Bruneau M, George B: Foramen magnum meningiomas: de-
tailed surgical approaches and technical aspects at Lariboisière
Hospital and review of the literature. Neurosurg Rev 3:19–33,
2008
  6.  Coric D, Branch CL Jr, Wilson JA, Robinson JC: Arterio-
venous fistula as a complication of C1-2 transarticular screw
fixation. Case report and review of the literature. J Neurosurg
85:340–343, 1996
  7.  Curylo LJ, Mason HC, Bohlman HH, Yoo JU: Tortuous course
of the vertebral artery and anterior cervical decompression: a
cadaveric and clinical case study. Spine 25:2860–2864, 2000
  8.  de Oliveira E, Rhoton AL Jr, Peace D: Microsurgical anatomy
of the region of the foramen magnum. Surg Neurol 24:293–
352, 1985
  9.  Ebraheim NA, Xu R, Ahmad M, Heck B: The quantitative
anatomy of the vertebral artery groove of the atlas and its rela-
tion to the posterior atlantoaxial approach. Spine 23:320–323,
1998
10.  Francke JP, Dimarino V, Pannier M, Argenson C, Libersa C:
The vertebral arteries (arteria vertebralis): the V3 atlanto-ax-
oidal and V4 intracranial segments-collaterals. Anat Clin 2:
229–242, 1981
11.  George B: Exposure of the upper cervical vertebral artery, in
Dickman CA, Spetzler RF, Sonntag VKH (eds): Surgery of
the Craniovertebral Junction. New York: Thieme, 1998, pp
545–567
12.  George B, Cornelius J: Vertebral artery: surgical anatomy, in
Spetzler RF (ed): Operative Techniques in Neurosurgery.
Philadelphia: WB Saunders, 2001, pp 168–181
13.  Gluf WM, Schmidt MH, Apfelbaum RI: Atlantoaxial transar-
ticular screw fixation: a review of surgical indications, fusion
rate, complications, and lessons learned in 191 adult patients.
J Neurosurg Spine 2:155–163, 2005
14.  Hong JT, Lee SW, Son BC, Sung JH, Yang SH, Kim IS, et
al: Analysis of anatomical variations of bone and vascular
structures around the posterior atlantal arch using three-di-
mensional computed tomography angiography. J Neurosurg
Spine 8:230–236, 2008
15.  Hotta S, Morita A, Seichi A, Kirino T: Aberrant vertebral ar-
tery course in a case of Chiari malformation and occipitoat-
lantal assimilation. Case report. J Neurosurg Spine 3:246–
248, 2005
16.  Kawashima M, Tanriover N, Rhoton AL Jr, Ulm AJ, Matsu-
shima T: Comparison of the far lateral and extreme lateral
variants of the atlanto-occipital transarticular approach to an-
terior extradural lesions of the craniovertebral junction. Neu-
rosurgery 53:662–675, 2003
17.  Kazan S, Yildirim F, Sindel M, Tuncer R: Anatomical evalua-
tion of the groove for the vertebral artery in the axis vertebrae
for atlanto-axial transarticular screw fixation technique. Clin
Anat 13:237–243, 2000
18.  Lee JH, Jahng TA, Chung CK: C1-2 transarticular screw fixa-
tion in high-riding vertebral artery: suggestion of new trajec-
tory. J Spinal Disord Tech 20:499–504, 2007
19.  Liu J, Shafiq Q, Ebraheim NA, Karkare N, Asaad M, Wolden-
459

berg L, et al: Value of intraoperative true lateral radiograph
of C2 pedicle for C1-2 transarticular screw insertion. Spine J
5:434–440, 2005
20.  Madawi AA, Casey AT, Solanki GA, Tuite G, Veres R, Crock-
ard HA: Radiological and anatomical evaluation of the atlanto-
axial transarticular screw fixation technique. J Neurosurg 86:
961–968, 1997
21.  Mandel IM, Kambach BJ, Petersilge CA, Johnstone B, Yoo
JU: Morphologic considerations of C2 isthmus dimensions for
the placement of transarticular screws. Spine 25: 1542–1547,
2000
22.  Neo M, Fujibayashi S, Miyata M, Takemoto M, Nakamura T:
Vertebral artery injury during cervical spine surgery: a survey
of more than 5600 operations. Spine 33:779–785, 2008
23.  Neo M, Matsushita M, Iwashita Y, Yasuda T, Sakamoto T,
Nakamura T: Atlantoaxial transarticular screw fixation for a
high-riding vertebral artery. Spine 28:666–670, 2003
24.  Neo M, Sakamoto T, Fujibayashi S, Nakamura T: The clinical
risk of vertebral artery injury from cervical pedicle screws in-
serted in degenerative vertebrae. Spine 30:2800–2805, 2005
25.  Paraskevas G, Papaziogas B, Tsonidis C, Kapetanos G: Gross
morphology of the bridges over the vertebral artery groove on
the atlas. Surg Radiol Anat 27:129–136, 2005
26.  Prabhu VC, France JC, Voelker JL, Zoarski GH: Vertebral ar-
tery pseudoaneurysm complicating posterior C1-2 transartic-
ular screw fixation: case report. Surg Neurol 55:29–34, 2001
27.  Rhoton AL Jr (ed): The far-lateral approach and its transcon-
dylar, supracondylar, and paracondylar extensions, in: Cra-
nial Anatomy and Surgical Approaches. Philadelphia: Lip-
pincott, Williams and Wilkins, 2003, p 639
28.  Rocha R, Safavi-Abbasi S, Reis C, Theodore N, Bambakidis
N, de Oliveira E, et al: Working area, safety zones, and angles
of approach for posterior C-1 lateral mass screw placement: a
quantitative anatomical and morphometric evaluation. J Neu-
rosurg Spine 6:247–254, 2007
29.  Sanelli PC, Tong S, Gonzalez RG, Eskey CJ: Normal variation
of vertebral artery on CT angiography and its implications for
diagnosis of acquired pathology. J Comput Assist Tomogr
26:462–470, 2002
30.  Sawlani V, Behari S, Salunke P, Jain VK, Phadke RV: “Stretched
loop sign” of the vertebral artery: a predictor of vertebrobasilar
460
A. J. Ulm et al.
insufficiency in atlantoaxial dislocation. Surg Neurol 66:298–
304, 2006
31.  Sen CN, Sekhar LN: An extreme lateral approach to intradural
lesions of the cervical spine and foramen magnum. Neuro-
surgery 27:197–204, 1990
32.  Tedeschi H, Rhoton AL Jr: Lateral approaches to the petro-
clival region. Surg Neurol 41:180–216, 1994
33.  Tubbs RS, Shoja MM, Acakpo-Satchivi L, Wellons JC III,
Blount JP, Oakes WJ, et al: Exposure of the V1-V3 segments of
the vertebral artery via the posterior cervical triangle: a cadav-
eric feasibility study. J Neurosurg Spine 5:320–323, 2006
34.  Wright NM, Lauryssen C: Vertebral artery injury in C1-2
transarticular screw fixation: results of a survey of the AANS/
CNS section on disorders of the spine and peripheral nerves.
J Neurosurg 88:634–640, 1998
35.  Yamaguchi S, Eguchi K, Kiura Y, Takeda M, Kurisu K: Pos-
terolateral protrusion of the vertebral artery over the posterior
arch of the atlas: quantitative anatomical study using three-di-
mensional computed tomography angiography. J Neurosurg
Spine 9:167–174, 2008
36.  Yamazaki M, Koda M, Aramomi MA, Hashimoto M, Masaki
Y, Okawa A: Anomalous vertebral artery at the extraosseous
and intraosseous regions of the craniovertebral junction: anal-
ysis by three-dimensional computed tomography angiogra-
phy. Spine 30:2452–2457, 2005
37.  Yoshida M, Neo M, Fujibayashi S, Nakamura T: Compari-
son of the anatomical risk for vertebral artery injury associ-
ated with the C2-pedicle screw and atlantoaxial transarticular
screw. Spine 31:E513–E517, 2006
Manuscript submitted October 12, 2009.
Accepted April 8, 2010.
Current address for Dr. Quiroga: Georgia Neurosurgical Institute
and Mercer University, Macon, Georgia.
Address correspondence to: Arthur J. Ulm III, M.D., Louisiana
State University, School of Medicine, Louisiana State University
Department of Neurosurgery, Health Sciences Center, 2020 Gravier
Street, 336A, New Orleans, Louisiana 70112. email aulm@lsuhsc.
edu.
J Neurosurg: Spine / Volume 13 / October 2010