ARTERIAL SUPPLY OF THE CERVICAL SPINAL CORD AND ITS

RELATION TO THE CERVICAL MYELOPATHY IN SPONDYLOSIS

Arris and Gale Lecture delivered at the Royal College of Surgeons of England
on
8th February 1967
by

B. G. Chakravorty, M.S., Ph.D., F.R.C.S.Eng., F.R.C.S.Ed., F.A.C.S.

Consultant Neurosurgeon and Lecturer, Department of Surgery, Sir Nilratan Sircar
Medical College, Calcutta; Teacher, Department of Surgery, Postgraduate College of
Medicine, University of Calcutta

CERVICAL SPONDYLOTIC MYELOPATHY is now known to be a common

neurological disorder affecting elderly people. In fact it is one of the
commonest, if not the commonest, diseases of the spinal cord after middle
life (Brain, 1954). This condition is the effect of cervical spondylosis,
which is a slowly progressive degenerative condition affecting the intervertebral discs.
The material presented in this lecture is partly anatomical and partly
clinical. An attempt has been made to define the arterial supply of the
spinal cord in the cervical region, and to correlate the anatomical findings
with the clinical features of cervical myelopathy in spondylosis. Operative
treatment is based on this vascular concept, and the results so far have
been satisfactory.
Pathology
The basic pathology of cervical spondylosis is the protrusion of degenerated disc material into the spinal canal. The primary factor in disc
degeneration is still uncertain. It is thought to be a process of degeneration associated with ageing when the nucleus pulposus suffers a progressive
reduction in its fluid content which results in a diminution in its size
together with a loss of elasticity (Puschel, 1930; Saunders and Inman,
1940). This process of degeneration may occur at an earlier age in association with trauma or with pre-existing congenital maldevelopment
either in the bony spine (fusion, spina bifida, hemivertebra) or in the disc
itself.
Once the disc shrinks and loses its resilience the vertebral bodies tend to
approximate. The mobility of the cervical spine puts more strain on the
degenerated annulus, and the disc matter bulges out. Nuclear material
may prolapse through a tear in the annulus (nuclear herniation) or the
annulus itself may bulge out (annular protrusion). The prolapsed disc
matter lifts the periosteum, and the subjacent bone proliferates to form
osteophytes; these with time grow from above and below to enclose the
prolapsed material. Excepting the condition of massive prolapse, when
the spinal canal and the intervertebral foramina may all be equally involved, the localized form of the disc protrusions usually take three
anatomical positions, median, dorsilateral and intraforaminal (Fig. 1).
Although intervertebral disc lesions in the cervical region as a cause of
232

ARTERIAL SUPPLY OF THE CERVICAL SPINAL CORD

acute cord lesion had been recognized as early as 1892 by Victor Horsley,
who treated the condition surgically, it was not until 1926 that Elliott
found that radicular symptoms involving the cervical nerve roots may be
caused by the narrowing of the intervertebral foramina due to arthritic
changes in the vertebral column. Since then many others have worked
on the clinical and pathological aspect of cervical disc lesions (Mixter and
Ayer, 1935; Stookey, 1940; Spurling and Scoville, 1944), and direct pressure on the cord or nerve root by the prolapsed disc substance was held
mainly responsible for myelopathy. Spillane and Lloyd (1951) in a
report of 12 cases reviewed the literature and concluded that multiple
disc degeneration leads to bulging of the discs; these ridges compress the
cord, and produce abrasive effects on it by constant friction. Brain
et al. (1952), in their post-mortem study of cases of cervical spondylosis,
observed adhesions of dura to the posterior longitudinal ligament, and of
the dural sleeves to the margins of the intervertebral foramina. They

Fig. 1. Digram showing the position of a median (I), dorsilateral (II) and intraforaminal (III) cervical disc protrusion. There may be occasionally in-between
positions such as paramedian and lateral.

stressed that the cord was rendered more susceptible to injury by these
adhesions. They also suggested that the interference with the vascular
supply was an important subsidiary factor. Bedford et al. (1952) supported the above workers, and commented that abnormal fixation of the
cord may render it vulnerable to normal neck movements. At autopsy
they observed unusual tension on the thickened denticulate ligaments and
fibrous nerve roots; in addition, the theca was adherent to the posterior
longitudinal ligament. Taylor (1953) suggested a compression effect on
the cord by the ligamenta flava during each extension movement of the
neck against the anterior spondylotic elevations, thus producing gradual
changes of an uncertain nature in the vasculature of the spinal cord.
Clinical features
On analysis of the clinical features of cervical spondylotic myelopathy
it has generally been agreed that the dominating feature of the condition is
weakness and spasticity of limbs, usually starting in the legs (Tablea).
233

B. G. CHAKRAVORTY

This, in combination with extensor plantar responses and exaggerated

deep tendon reflexes, is attributable to lesions in the pyramidal tracts.
Sensory depression, loss of reaction to pain and temperature, with preservation of light touch and tactile discrimination, are caused by lesions
in the spinothalamic tracts. The sphincter disturbances are also attributable to spinothalamic depression when lack of cortical appreciation of
visceral sensation occurs. Anatonmically both these tracts occupy the
lateral white column of the spinal cord.
Pain radiating down the arm, and zones of anaesthesia in dermatome
distribution are due to damage to roots rather than to the cord. The
common involvement of the first three fingers is due to the frequently
affected C5 and C6 nerve roots. Loss of vibration sense and tactile
discrimination, and a positive Rhomberg's sign, relate to damage of the
posterior columns; these signs are occasional and rarely complete,
TABLE I
ANALYSIS OF CLINICAL FEATURES (31 PATIENTS)

No.

1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.

No. of
Symptonms and signs
patienits
.. 20
Weakness and spasticity of lower limbs ..
Simultaneous involvemeait of all limbs with weak..
..
..
14
..
ness and spasticity
..
..
16
Clumsiness of finger movements ..
Spinothalamic (pain and temperature) depression,
.. 14
..
..
..
and paraesthesia ..
..
..
7
..
Sphincter disturbances
6
Pain in the neck with radiation into the arms ..

Limitation of neck move:ments ..

..
Exaggeration of deep je.-ks
..
..
Babinski's sign positive
..
Increase of tone in all limbs
..
Inversion of supinator j(,rks
Atrophy of intrinsic muscles of hand

..
..
..
..
..
..

..
..

.
.
..

7
25
21
16

15
8

signifying a partial involvement. Recently Crandall and Batzdorf (1966)

in an analysis of 199 cases have shown pyramidal and spinothalamic
involvement in all cases, whereas posterior columns were affected only in
16 per cent.
Muscular atrophy and depression of reflexes can be explained by
damage to the anterior horn cells. The inversion of the radial reflex,
which was observed in 15 patients, is characteristic of this syndrome.
It indicated interruption of the reflex arc and associated involvement of the
pyramidal tract, resulting in exaggeration of the flexor movements of the
fingers (Spillane and Lloyd, 1951). Loss of anterior horn cells may also
be responsible for the disturbance of the reflex arc. Rigidity and weakness, a common finding, may indicate some damage to the extrapyramidal corticospinal pathways; the rubrospinal tract occupies the
ateral column just anterior to the main pyramidal tract (pre-pyramidal

tract).
Clinically the lesion is widespread. Compression and traction factors
seem inadequate to explain the extent and order of the damage. There is
sometimes little correlation between clinical findings in cervical myelo234

ARTERIAL SUPPLY OF THE CERVICAL SPINAL CORD

pathy, and the segmental level of radiological changes. The extent of

neurological damage is too great to be explained by the level of
compression.
The small muscles of the hand are often involved and this can be
attributed to the loss of anterior horn cells in the first thoracic segment,
and cannot be explained by compression or traction factors. In large
protrusions, or when the spinal canal has been greatly narrowed (10 mm.
or less compared with 17 mm. normal (Wolf et al. 1956)), direct compression of the cord against the laminae can cause myelopathy, but such a
degree of narrowing is rare (Brain, 1948), although the extent of narrowing
in spondylosis may be as little as 14 mm., still leaving enough room for the
cord (Payne and Spillane, 1957).
A vascular factor
It is agreed that if the direct anteroposterior cord compression is the
basis of myelopathy then the signs and symptoms will be related mainly
to the lesions of anterior horn cells and anterior and posterior columns.
Moreover, the level of the lesion will be localized, or remain localized for
a long time, corresponding to the compressing agent. This fails to explain widespread myelopathy in those cases where the compressing agent
is very small or non-existent.
Recently various workers have laid emphasis on a vascular factor in the
pathogenesis of cervical myelopathy. The theory of vascular insufficiency
in cervical myelopathy postulated by Morton (1950) and Allen (1952)
has also been supported by Mair and Druckman (1953), Girard et al.
(1954) and Clarke (1955). These workers, although putting forward a
common vascular hypothesis, differ in their explanation of the mechanism.
Brain (1948), Frykholm (1951) and Taylor (1964) have also suggested
the possibility of a vascular factor.
This suggestion of a vascular theory is not altogether a new concept.
As early as 1875 Braun thought that osteophytic compression of the periradicular vessels was an essential factor in the production of root symptoms. After a long lapse of time, Greenwood in 1942, while investigating
the pathogenesis of amyotrophic lateral sclerosis, considered it to be the
effect of chronic anoxaemia of the cord resulting from a slowly developing
reduction in the blood flow such as would occur in anterior spinal artery
thrombosis affecting firstly the motor tracts. It might very well be that
many of Greenwood's cases were actually patients suffering from cervical

spondylotic myelopathy.
Although several workers have suggested the hypothesis of anterior
spinal ischaemia in the pathogenesis of cervical myelopathy, the exact
mechanism remains controversial. This led us to find out more about
the vascular factors in cervical myelopathy from anatomical and comparative anatomical studies of the arterial supply of the cervical cord
in cadavers and in laboratory animals.
235

B. G. CHAKRAVORTY

Materials and methods

The material upon which this present study has been made consists of
31 spinal cords obtained at autopsy, as soon as possible after death, from
adults of different ages, mostly the victims of accidental death who in all
probability during life had not suffered from any vascular or nervous
diseases. In order to study the comparative anatomy, in vivo injection
experiments were carried out on four monkeys, three dogs, five rabbits and
four rats.
The arterial supply of the cervical spinal cord was studied following
injection of radio-opaque barium, Indian ink, Monastral blue (I.C.I.),
and coloured neoprene latex solutions. The apparatus for injection
consisted of a small Woulfe's bottle, connected in series to the inlet tube
with a manometer in which pressure during injection was maintained at
120-160 mm. Hg, i.e. at an average systolic blood pressure appropriate
to the age of the specimen, by pumping in air.
In all cadavers, prior to injection, the brain was removed at the midbrain level avoiding damage to the basilar artery and its branches. The
clavicle on the ipsilateral side was removed and the subclavian artery with
its principal branches, i.e. vertebral, thyrocervical, costocervical and internal mammary trunks, were isolated and injected with the media under
positive pressure.
Following injection, the spinal cords with intact dural coverings and
nerve roots were dissected out from the medulla to the third thoracic
level.
After fixation, the meninges were removed and the vessels were examined
macroscopically, radiologically and histologically.
Observations
In this series, complete absence of the right vertebral, the right anterior
spinal or of the main anterior spinal trunk (Duret, 1873; Spiller, 1908)
has not been observed. In each case the vertebral artery entered into the
cranium piercing the atlanto-occipital membrane. Never was it seen to
pass through the second or third cervical intervertebral foramen (Kadyi,
1889), and the abnormality of incontinuous anterior spinal trunk
(Romanes, 1962) has also not been noted.
It was noticed that when injections were carried through the anterior
and posterior spinal branches of the intracranial vertebral artery these
arterial trunks could be filled only in the highest cervical segments; the
rest of these trunks had very poor filling if any at all. The spinal branches
of the intracranial vertebral artery are probably important for the filling
of the upper cervical cord (Cl, C2, C3) only. In all specimens the whole
anterior and posterior arterial trunks had excellent filling when injections
were carried out through isolated radicular arteries (Fig. 2a and b).
The radicular arteries were inferred to be the main source of supply for the
cervical spinal cord except for a few segments at the cranial end where the
236

ARTERIAL SUPPLY OF THE CERVICAL SPINAL CORD

vertebral anterior and posterior spinal branches are important supply.

Anterior spinal artery: The diameter of the anterior spinal trunk always
increases immediately below the junction with a radicular artery, indirect,
evidence that the radicular arteries are feeding vessels, and bring blood
to the anterior spinal artery. In three cases only we noted the Y-shaped

..
I.T.'

ia b

zM,r

.
:'wx-1w

(C)
(t)
(a)
Fig. 2. (a) Shows filling of the anterior spinal artery (paired vessel in this specimen).
Cranial end is not filled, injection was made through the isolated left C5 radicular
artery which also shows Y-shaped bifurcation. (b) Shows the posterior spinal
arterial chains of both sides and its relation to the posterior nerve roots. Injection
was made through the isolated right C5 radicular artery. The cranial end is sparingly filled. (c) Showing the Y-shaped bifurcation of a solitary radicular artery
into an ascending and descending branch to form the anterior spinal artery.

division of the anterior spinal artery, as though its branches were really
the anterior spinal artery going up and down (Fig. 2c). In 19 specimens
the anterior spinal artery was a single trunk, but in 12 these were paired
vessels in the cervical region; such duplications were never noted below

C5/6 segments (Fig. 2a).

237

B. G. CHAKRAVORTY

The anterior spinal artery, when a single trunk, was always contained in
the midline groove. Sometimes deflections occurred at the junction of the
radicular arteries, but these were never more than 1 mm. from the midline
except in those cases where the anterior spinal artery appeared to be the
Y-shaped bifurcation of a major anterior radicular artery, which could be
as much as 5 mm. from the midline (Fig. 2c). In duplications, the wider
vessel more often occupied the midline sulcus, and the other was usually
within 0.5-1.5 mm.; the distance was never more than 2 mm. Knowledge
regarding the position and duplication of the anterior spinal artery is of
great importance in operative procedures like cordotomy.
Posterior spinal arteries: In 17 specimens the origin of the posterior
spinal arteries were dissected free; in 11 they originated from the posterior
inferior cerebellar artery, and in six they were branches of the vertebral
artery. The posterior spinal arteries appeared as anastomotic chains on
either side of the midline close to the line of attachment of the posterior
nerve roots (Fig. 2b). This arterial system was not a continuous channel
of uniform diameter but extremely irregular, and in parts found to be
incomplete; a breach in continuity may occur at the junction of posterior
spinal with the first posterior radicular artery. Injection into the anterior
spinal artery only, never produced filling of the posterior spinal and vice
versa; it is possible that the anastomoses between the anterior and posterior
arterial systems are very sparse and functionally ineffective.
Cervical radicular arteries: The cervical radicular arteries, i.e. the arteries
accompanying the nerve roots entering through the intervertebral
foramina, are capable of filling both anterior and posterior arteries in the
cervical cord when injections are carried through the isolated radicular
arteries obliterating the intracranial vertebral branches (Fig. 2a and b).
These radicular arteries may originate from any branch of the subclavian
artery in the neck, i.e. from the vertebral, costocervical and thyrocervical
trunks. In the upper six segments they can arise from the vertebrals
or from the ascending cervical branch of the thyrocervical trunk, and the
spinal branches of these two vessels always anastomosed (Fig. 3a). It
was apparent that either the vertebral or the ascending cervical can maintain the radicular supplies in the upper cervical region through these
anastomoses if either source is blocked. A radicular of the 7th or 8th
segment is always from the branches of the costocervical trunk. As a rule,
a radicular artery accompanying the 8th root is rare; in only one specimen
in this series was there a radicular artery accompanying the right 8th
root.
Although there may be two branches (one from the vertebral and one
from the ascending cervical) entering one intervertebral foramen, the
radicular artery connecting the anterior spinal trunk is always singular for
that particular segment; it originates as a common radicular artery and
divides into an anterior and a posterior branch. The anterior radicular
has a range of diameter 100-750 p, and can be divided into major
238

ARTERIAL SUPPLY OF THE CERVICAL SPINAL CORD

(500-750 p), minor (250-500 p) and minimal (less than 250 ,u), according
to the diameter. It appeared that vessels having a diameter greater than
250 ,u were the only sources of blood supply to the spinal cord (excepting
the highest segments), because only they joined the anterior spinal artery.

,."

II,/

Fig. 3. (a) Latex corrosion specimen, showing the intervertebral branches from
the ascending cervical artery. These anastomose with the radicular branches of the
vertebral artery, or the radicular arteries may arise directly from the ascending
cervical artery. (b) Corrosion specimen of the vascular pattern in a rabbit. The
formation of the basilar artery is seen. Radicular branches of the right ascending
cervical and the vertebral artery anastomose.

The minimal vessels (i.e. having diameters less than 250 p) were seen in
almost every segment and they ended supplying the tissues around the
roots.
Radicular arteries in spondylosis
In the present series it was observed that, though the anterior radicular
arteries may occur at any level, the significant vessels (i.e. more than 250 i
diameter) were usually between C4 and C6 segments and very rarely
at C3, C7 and C8. In the 31 spinal cords investigated, eight radicular
239

B. G. CHAKRAVORTY

arteries accompanied C4 nerve root, sixteen C5 and nine entered along

with C6 nerves. The average number of radicular arteries (major and
minor) in the cervical cord is two or three. This observation seems to
agree with that of Kadyi (1889) and Suh and Alexander (1939), and there
was very little difference from other workers (Table II). In many cases
the anterior radicular artery may be solitary; in 20 of the 31 cords investigated there was only one anterior radicular artery communicating
with the anterior spinal trunk in the cervical region (Fig. 2c).
Spondylotic changes in the cervical vertebrae commonly affect the
lower cervical region. In Morton's (1950) series of post-mortem examTABLE II
NUMBERS AND POSITIONS OF THE SIGNIFICANT RADICULAR ARTERIES IN THE CERVICAL
CORD AS NOTED BY DIFFERENT OBSERVERS
Anterior radicular
Posterior radicular
arteries
arteries
I or 2; at C6 or C7
Adamkiewicz (1881)
Usually 3; at C4, C5
and C8
I or 2; between C4 and
Kadyi (1889)
Usually 2; between C4
C7
to C7; most commonly
at C5 or C6
1 or 2 in lower and 1 in
Suh and Alexander
(1939)
upper cervical region,
usually between C3 to
C6
A large one at C4
Bolton (1939)
1 or 2 usually at C6 or
Zulch (1954)
C7
I major at C6 or C7; 3
1 at C2; I to 3 between
Woollam and Millen
(1958)
to 5 minor at segments
C5 to C7
above C6
At least 2; larger one at
Gillilan (1958)
C5 or C6, the next one
at C3; almost never at
C8
1 to 5 may occur at all
Perese and Fracasso
(1959)
levels; maximum between C3 to C7
1 or 2 may occur at any
2 or 3 may occur at any
Present series
level, mostly between
level between C2 to C6;
C4 to C6; very rarely at
usually one at C4
C8

inations, the commonest incidence was at C4/5 and C5/6 levels. In

Wilkinson's (1960) series of 17 autopsy examinations, all but one had a
disc protrusion at C5/6 level, and the next common level was C4/5. In
the large series of Crandall and Batzdorf (1966), comprising 199 cases,
the same order of incidence is seen. We also confirm this with our clinical
experience, autopsy findings, and during operative procedures in this
series of 31 patients (Table lll). It is strongly felt that if the radicular
artery (or arteries) is occluded the risk of ischaemia arises. The risk is
greater if there is but a single major radicular artery which is occluded,
and the incidence of solitary radicular artery in the cervical cord, as
already stated, is frequent. By random clipping of C4, C5 and C6 nerve
roots, including the radicular arteries (if any), it had been noticed that in
240

ARTERIAL SUPPLY OF THE CERVICAL SPINAL CORD

all specimens but one there was no significant filling of the anterior spinal
artery.
The intravertebral part of the vertebral artery and the first six nerve
roots are embedded in one mass of connective tissue. The radicular
arteries from vertebral or from ascending cervical originate and pass
through this connective tissue mass. This connective tissue is normally
composed of loose areolar type, and its character does not change with
age, i.e. it retains its areolar character whether it is from a still-born foetus
or from a person over 75. In spondylosis, probably by the irritation of
osteophytes, this areolar tissue changes into mature fibrous tissue which
with time becomes hyaline and sclerotic. It surrounds and envelops
the perineural sheaths of the issuing nerves at the exit foramina producing
a constricting effect on the contents of the i.v. foramina including the
cervical nerves, and the radicular arteries. Although Frykholm (1951)
noticed fibrosis of the dural sheaths, what he called ' root sleeve fibrosis ',
we have found that the real thickening and hyalinization are in the
peridural layer in which the radicular vessels run. The dura itself is
TABLE III
INCIDENCE OF Disc PROTRUSIONS AND SIGNIFICANT RADICULAR
ARTERIES IN RELATION TO CERVICAL SEGMENTAL LEVELS
Levels
Morton
Brain
Wilkinson Stoops
Present
Levels of
et al.
(1950)
(1960)
and
series
significant
31 cases
17 cases
(1952)
King
(1967)
radicular
38 cases
31 cases
arteries
(1962)
31 specimens
49 cases
4
C2/3
5
4
1
2
5
C3/4
20
12
1
6
8
6
19
11
7
12
C4/5
16
6
18
C5/6
16
22
21
9
5
14
C6/7
9
14
18
2
1
2
6
1
1
C7/T1

little thickened, and at operation the sclerotic epidural tissue can be

scraped off leaving a clear shining dural surface. This root sheath
compression can be pre-operatively confirmed by myelography; in
myelogram this is evidenced by the failure of entry of Myodil into the
affected nerve sheaths (Fig. 4).
It becomes apparent that should the blood supplying radicular artery
(or arteries) be affected by lateral foraminal disc protrusion and/or the
reactive fibrosis there will be a major cut-down in the vascular supply to
the anterior spinal trunk, and anterior spinal ischaemia will follow; the
risk is maximal when the radicular artery is solitary and that too is involved in the osteophytic and fibrous strangulation.
The histology of the cervical cord obtained at post-mortem examinations
from cases of spondylotic myelopathy shows widespread degeneration,
extending both upwards and downwards for several segments. Mair and
Druckman (1953) have shown that the lesion principally involved the
anterior horns, lateral columns and the anterior part of the dorsal columns.
The histological changes were similar to those of ischaemia, and they also
showed that the area of lesion following the zone of distribution of the
241

B. G. CHAKRAVORTY

anterior spinal artery. We fully agree with Mair and Druckman. The
clinical features of the patients suffering from spondylotic myelopathy
can be explained satisfactorily on the basis of lesions involving the anterior
spinal arterial zone. The anterior spinal artery supplies the anterior twothirds of the spinal cord, i.e. anterior horns, anterolateral tracts (main
tracts are anterior and lateral, pyramidal and spinothalamic tracts;
rubrospinal tract), central grey matter, and the anterior part of the posterior
columns (Fig. 5a and b).
The atrophy of the small muscles of the hand (lesion of Ti segment) in
cervical myelopathy can only be explained by the vascular hypothesis.
In all specimens we noticed that, on injecting either through the intracranial branch or through the cervical radicular arteries, the filling of the

Fig. 4. Anteroposterior myelogram shows absence of bilateral root sleeve filling.

At operation C6 roots on both sides were found to be compressed by fibrous tissue

only.

anterior and posterior spinal trunks gradually tapered down to the lower
cervical segments; at most it reached up to the second thoracic segment
and never below it. This may indicate a terminal zone of the craniocervical feeding vessels and would support the theory that segments of the
highest thoracic level, where the supply of blood from upper (craniocervical) and lower (thoracic radicular) sources meet, are likely to suffer
from an insufficiency of blood from either source (Fig. 5c). Zulch (1954)
in anatomical and clinical sudies has shown that an occluding lesion of
the anterior spinal artery, or of a major cervical radicular artery, can
cause necrosis of lower cervical and highest thoracic segments, as this
area is the 'last field' zone where the supply of the cervical and thoracic
radiculars come into contact.
242

,7~ ~ atqe

ARTERIAL SUPPLY OF THE CERVICAL SPINAL CORD

I
Ir_A
-i
wL _~ IP|
C iI _gl

tXa s s

Cl.

_ QCl

11M

_a.pwnsl wAve'y
b4sr2;t1.3

WL.

s.i

cs/I&/c-r

TI

(a)

atTS/t'9

(c)

(b)

Fig. 5. (a) Section of the spinal cord showing relative positions of the ascending
and descending tracts, and the areas of distribution of the anterior and posterior
spinal arteries. The leg fibres occupy the outermost position in the lateral corticospinal (main pyramidal) bundle. (1) Dorsal spinocerebellar; (2) lateral corticospinal (pyramidal); (3) ventral spinocerebellar; (4) lateral spinothalamic; (5) ventral
spinothalamic; (6) central branch of anterior spinal artery; (7) posterior spinal
artery; (8) peripheral pial network. (b) Micro-angiography of a C5 segment
showing the joining of an anterior radicular artery to the anterior spinal artery,
and the distribution of the central branch of the anterior spinal in the anterolateral
aspect of the cord. (c) Shows the probable direction of blood flow in the anterior
spinal artery. The areas indicate the ' last field ' zones.

243

B. G. CHAKRAVORTY

The vascular hypothesis can also explain why the legs are first affected
with spasticity and weakness in so many cases and why in some cases
paraesthesia is the first symptom. In the spinal cord, the leg fibres
occupy the outermost position in the main corticospinal bundle; the
spinothalamic tracts are also located peripherally. The distal position of
the leg fibres, and of the spinothalamic tracts, make them more susceptible
to even a minor degree of ischaemia; these zones are similar to a ' last
field' on horizontal cross sections where the central branches of the
anterior spinal artery and the penetrating branches of the pial network
meet (Fig. 5a and b). In progressive ischaemia the area of supply
gradually attains a concentric constriction, and the centre suffers last.
The anterior part of the posterior columns, though supplied by the anterior
spinal artery, is usually spared in ischaemia; Gillilan (1958) has shown
that there is always a zone of overlapping in this area.
Comparative anatomy
It was noticed that the arterial arrangements of the cervical cord in
monkeys and rats were very similar to those of man. In dogs, the branch
from the vertebral artery accompanying the first cervical nerve root is
very small, the larger vertebral branch commonly enters through the
third intervertebral foramen. In none of the human specimens in our
series was this anatomical variation noted, though such an anomaly in the
human is on record (Kadyi, 1889; Sterzi, 1914). In rabbits, the arterial
pattern is somewhat different as the right innominate and left common
carotid leave the arch of the aorta as a single trunk, and the left subclavian
is a separate aortic branch. In number and positions also, the anterior
radicular arteries in dogs and rabbits differed from those of monkeys,
rats and men. The number of radicular arteries was greater in dogs and
rabbits; almost every segment had a radicular artery of more or less
uniformly small diameter. The contribution from the ascending cervical
artery (Fig. 3b), the relationship with the vertebral artery and cervical
nerve roots and the ultimate distribution of vessels in the cord are similar
in all these animals, including man (Sahs, 1942; Woollam and Millen,
1958; Yoss, 1950).
In our animal experiments, interruption of bilateral C5 radicular
arteries produced widespread vertical degeneration of the cord extending
from first cervical to first thoracic segments (Table IV). Necrosis was
most marked at C5 and C6 segments, and the total area of the segments
was involved. In the third and fourth segments, degeneration affected
both anterior and posterior parts, while at distal segments it was related
strictly to the distribution of the anterior spinal artery (Fig. 6). The total
necrosis at the level of the occlusion is due possibly to a secondary necrosis
as a result of occlusion and haemorrhage in the anterior two-thirds of the
segments following the acute ischaemic experiment.
Treatment
Every patient in the present series had a prior trial with non-operative
244

ARTERIAL SUPPLY OF THE CERVICAL SPINAL CORD

conservative measures such as immobilization, traction and limitation of

activity combined with analgesics and various drugs, ranging from weeks
to several years. This regime of management afforded a varying degree of
symptomatic relief for a temporary period in most of the patients. The
neurological deterioration remained progressive though the rate of worsening varied and ultimately, when conservative measures were considered
inadequate, operative treatment was proposed.
The operation procedure adopted in this series aims mainly at relieving
the compressed radicular arteries by decompression of the nerve roots at
those intervertebral foramina where they are affected, and it also frees
the cord from other possible compressing agents such as a big anteriorly
placed bony spur or hypertrophied ligamenta flava.
TABLE IV
EXPERIMENTAL ISCHAEMIA IN DOGS

Method
1.
2.

3.

4.

Procedure
adopted
Both vertebrals
and one common
carotid ligated in
neck
Both vertebrals
tied in neck with
production of artificial thrombosis
in one vertebral
artery
biIntradural
lateral ligation of
C4 nerve roots
and accompanying
radicular arteries
biIntradural
lateral ligation of
C5 nerve roots
and accompanying
radicular arteries

Effects

Histological

Nil

Nil

Immediate diaphragmatic and

intercostal respiratory paralysis

Oedema of cells
and loss of Nissl
granules in the
anterolateral regions of C3, C4
and C5
Necrosis in anterolateral region of all
segments (Cl to
T2), except C5 and
C6 in which
necrosis was total

produced
Nil

Spastic weakness
in limbs and lack
of sensory and visceral appreciation

changes
Nil

Information already obtained by the lateral and oblique X-rays, and

myelography showing encroachment of the intervertebral foramina and
obliteration of root sheaths, helps to indicate the site of exposure (Fig. 4).
Operation comprises of laminectomy and decompression of the nerve roots
by removing the compressing agent, bony or fibrous.
Operative technique
The operation is carried out under general anaesthesia with the patient
in sitting position, which facilitates good respiration and minimizes
congestion and bleeding from the extradural venous plexus. Laminae
of the contiguous vertebrae of the affected intervertebral foramina are
removed. The position of the affected roots, i.e. for the more commonly
involved C4, C5, C6 levels, C5 and C6 roots, are then determined. The
root (or roots) may be swollen and tightly compressed by osteophytes
or by dense fibrosis. A fine probe is passed very gently along the fora245

B. G. CHAKRAVORTY

niina which helps to recognize the roots affected and the degree of compression; this is not attempted if inspection suggests a very high degree
of foraminal constriction. Using an electric dental burr, the affected
root is now decompressed by unroofing its foramen. This involves
removal of the posterior wall of the intervertebral foramen, i.e. contiguous
articular pillars with zygopophyseal joint (Figs. 7a and 8). The condensed fibrous tissue surrounding the dural sheath of nerve root is then
removed as far outwards as possible, leaving the roots free (Fig. 7b).
This process is repeated with each affected nerve root. At the conclusion

Cl (8

C2

CCe~~~~~A

Fig. 6. The shaded areas in the diagram show the areas of necrosis as observed
in the various segments of the spinal cord resulting from occlusion of bilateral C5
radicular vessels in dogs.

the roots should appear reasonably slack, and the probe should pass
smoothly. The dura need not be opened unless some other pathology
in the cord is suspected, and no attempt should be made to remove any
anterior bony spurs.
Results
Complications following the operation were fortunately rare. None
suffered from air embolism; probably meticulous care in haemostasis and
liberal use of saline irrigation during the operation prevented this danger.
There was no post-operative death.
All patients were allowed up on the sixth post-operative day, and re-

246

ARTERLkL SUPPLY OF THE CERVICAL SPINAL CORD

(a)

(b)

Fig. 7. Photograph during operation. (a) Muscles have been separated and
laminae have been removed. Dura is not opened. The electric dental burr is
being used to remove the posterior wall of the left C4/5 intervertebral foramen.
The nerve root is protected with a fine dissector. (b) Shows that C5 roots on both
sides have been completely freed from compression. Posterior bony walls have
been removed, and thickened epidural tissue has been stripped off leaving only
intact dural sheaths of the nerve roots.

II-t

A
'I
I'
I'

-l
-

II

il#

.1 :1'. 1
'I

Fig. 8. Diagrammatic representation of operation on the right side.

247

B. G. CHAKRAVOR1 Y

ceived continued post-operative physiotherapy. They were allowed to

continue their exercises at home after discharge. None needed any
external support.
These patients have been followed up for periods of six months to five
and a half years. The age of the patients and duration of their symptoms
seemed less important than the severity of the pre-operative neurological
abnormalities in determining prognosis.
Post-operative results were assessed on the basis of neurological findings
and also on the degree of return of their power to carry out their normal
activities. Results were considered good when there was complete
restoration of ability, and their normal interests. Of these 31 patients,
19 have shown good results (Table V). In eight, improvement has been
poor, i.e. they have shown marked improvement in muscle tone and power,
but disability is still there although the improvement continued. In three
patients subjective improvement has been slight. Though one has regained sphincter control satisfactorily, and another can move about the
house (this was difficult for him pre-operatively), they feel that there has
not been any significant change in their muscle power.
TABLE V
RESULTS OF FORAMINAL DECOMPRESSION
Cases
Death
Authors
Degree of improvement
Good Fair Slight, no change
or worse
6
3
3
..
Scoville (1961)
1 post-op.
49
28
15
Stoops and King
(2 untraced)
complication
(1962)
3 other causes
14
5
6
3
Epstein et al. (1963)
8
19
3
31
Present series (1967)

(1 untraced)

Post-operative physiotherapy is considered very important, and at least

once a week or fortnight during the first year or two the patient should
receive the supervision and guidance of a trained physiotherapist.
Discussion
Other operative procedures designed to relieve cord compression have
not proved satisfactory. Simple cord decompression by laminectomy
in different series of patients has not yielded such good results as in the
present series where it was combined with root decompression (Table V).
A recent recommendation of laminectomy and leaving the dura open has
also not yielded such satisfactory results. Attempts to chisel off the
ventral bony spurs have proved uniformly disastrous, though it has been
successful in Allen's personal series (1952). Transdural approach to the
median protrusions is also hazardous. Laminectomy with or without
section of the ligamentum denticulatum as advised by Kahn (1947) did
not improve results in Scoville's series (1961), and sometimes severely
worsened the long tract signs. This procedure has also been abandoned
by some workers (Mayfield, 1955; Bradshaw, 1957).
248

ARTERIAL SUPPLY OF THE CERVICAL SPINAL CORD

The anterior approach for removal of a cervical disc with subsequent

fusion is commended for single space disc prolapse with or without
subluxation, and when the diagnosis of spondylosis is absolutely certain.
However, Logue (1967) has encountered spinal cord angiomas, secondary
deposits, and syringomyelia cysts when the pre-operative diagnosis was
spondylotic myelopathy. These conditions would have been missed if an
anterior approach had been made. In spondylosis, calcified protrusions
and osteophytes broaden the intervertebral surface, and the neck becomes
more stable; in these cases even careful laminectomy and excision of the
posterior wall of the intervertebral foramina usually do not impart any
marked instability even when several roots are decompressed.
Laminectomy with foraminal decompression in cases of spondylotic
myelopathy was occasionally practised as early as 1952 (Brain et al.).
Since then the advantage of foraminotomy over other procedures has been
noted by various workers including ourselves (Table V), although the
TABLE VI
RESULTS FOLLOWING LAMINECTOMY

Author
Epstein and Davidoff (1951)
..
Northfield (1955) ..
..
..
Arnold (1955)
Walsh and Mackenzie
..
..
(1956) ..
Segerberg (1956) ..
..
..
Taylor (1964)
Tota!

..

..

..

Improvement

Unchanged

Total

Slight

1
13
2

Fair
1
2

9
1

16
3

4
38
8

13
4
2

3
3

11
4

28
7
15

22

22

20

36

100

Good

or worse
2

explanations are different. Most workers comment that this remarkable

improvement in the result is due to a release of the anchoring effect of the
spinal root sleeves which permit a dorsal migration of the cord (Scoville,
1961; Stoops and King, 1962; Epstein et al. 1963). We have no doubt
that foraminal decompression of roots and removal of the dense peridural
fibrous tissue are responsible for this remarkable change in the results.
It is felt that this is not due to the removal of the tethering effect and dorsal
migration of the cord from the bony spur (which in fact does not occur),
but to release of the important radicular arteries so long compressed, with
a resultant improvement in the blood supply of the cervical spinal cord.
The foraminal compression may be by fibrous tissue only and in some
cases it has been found that bony changes were minimal while fibrosis
was dense.
The radicular arteries are the main feeders of the anterior spinal artery,
and their levels of origin usually correspond to the levels commonly
affected with spondylosis (Table III). In myelopathy vascular occlusive
anoxia is slow and chronic, never acute and rarely complete. This
probably explains why operative radicular decompression brings about
improvement. It is true that if the necrosis and cavitation have already
249

B. G. CHAKRAVORTY

taken place in the cord substance (recognized clinically by features of

nearly complete transection), this operative foraminal decompression will
be of little value.
Even extreme pressure on the nerves does not impede conduction of
nerve impulses. Grundfest (1936) exposed nerves to a range of pressure
2,000-7,000 lb./sq. inch, and still there was no defect in nerve conduction.
On the other hand, Causey (1955) pointed out that vascular obstruction
without interfering with the fibres can block impulse conduction, and that
this ischaemic block was reversible. This theory may explain why the rate
of post-operative improvement is so high after the decompression
operation.
The question may arise that there may be many elderly persons with
gross radiological narrowing of the cervical intervertebral foramina who
are free from myelopathy. On the other hand, there are some patients
with minimal radiological evidence of one-space narrowing with gross
myelopathy. It is possible that the former group have at least one major
radicular artery spared which is sufficient to maintain the nutrition of the
cervical cord, whereas in the latter it may well be that the solitary radicular
artery has been involved.
It is possible that in the pathogenesis of this syndrome no single factor
is responsible, as direct compression, traction, venous stasis and ischaemia
may all play a part. Clinically and histologically the majority of cases
of cervical myelopathy resemble those of the anterior spinal artery
syndrome, but at operation or autopsy they do not show a sufficient
degree of cord compression or traction to explain the condition. As the
operation to release the strangled radicular vessels brings improvement
to the patient, only the ischaemic hypothesis provides an adequate
explanation.
ACKNOWLEDGEMENTS
My sincere thanks are due to Professor H. W. Rodgers and Mr. A. R.
Taylor of the Queen's University of Belfast, who gave me all possible
guidance and co-operation during the anatomical and experimental
studies in this connection. I am also grateful to Professor G. W. Causey
of the Royal College of Surgeons of England for his constructive criticism
in the final preparation of this lecture. I am also indebted to my colleagues in Belfast and in Calcutta for their valuable help, and to my wife,
Dr. Jayanti Chakravorty, for her technical assistance throughout this

study.
REFERENCES
ADAMKIEWICZ, A. (1881) S.B. Akad. Wiss. Wien. (math.-nat. K.L.) 84, 469.
ALLEN, K. L. (1952) J. Neurol. Neurosurg. Psychiat. 15, 20.
ARNOLD, J. G., JNR. (1955) Ann. Surg. 141, 872.
BEDFORD, P. D., BOSANQUET, F. D., and RUSSELL, W. R. (1952) Lancet, 2, 55.
BOLTON, B. (1939) J. Neurol. Psychiat. n.s. 2, 137.
BRADSHAW, P. (1957) Quart. J. Med. 26, 177.
BRAIN, W. R. (1948) Proc. Roy. Soc. Med. 41, 509.

250

ARTERIAL SUPPLY OF THE CERVICAL SPINAL CORD

BRAIN, W. R., NORTHFIELD, D., and WILKINSON, M. (1952) Brain, 75, 187.
(1954) Lancet, 1, 687.
BRAUN, J. (1875) Quoted by CAVE et al. (1955) Lancet, 1, 176.
CAUSEY, G. (1955) Ann. Roy. Coll. Surg. Engl. 16, 367.
CRANDALL, P. H., and BATZDORF, U. (1966) J. Neurosurg. 25, 57.
CLARKE, E. (1955) Lancet, 1, 171.
DURET, H. (1873) Arch. Physiol. 5, 97.
ELLIOTr, G. R. (1926) J. Bone Jt. Surg. 8, 42.
EPSTEIN, J. A., and DAVIDOFF, L. M. (1951) Surg. Gynec. Obstet. 93, 27.
EPSTEIN, B. S., and LAVINE, L. S. (1963) Arch. Neurol. 8, 307.
FRYKHOLM, R. (1951) Acta chir. scand. Suppl. 160, p. 149.
GILLILAN, L. A. (1958) J. comp. Neurol. 110, 75.
GIRARD, P. F., GARDE, A., and DEVIC, M. (1954) Rev. neurol. 80,48.
GREENWOOD, J., JNR. (1942) Trans. Amer. neurol. Ass. 68, 64.
GRUNDFEST, H. (1936) Cold Spring Harbor Symposium on Quantitative Biology, 4, 179.
HORSLEY, V. Quoted by WILKINSON (1960).
KADYI, H. (1889) Denkschriften der Math. Natur. Classe der akad. d. Wissenschaften in
Krakau erschienenen Monographie, aus dem Polnischen uberzetz vom verfasser,
15, p. 152. Lemberg, Gubrynowicz and Schmidt.
KAHN, E. A. (1947) J. Neurosurg. 4, 191.
LOGUE, V. (1967) Personal communication.
MAIR, W. G. P., and DRUCKMAN, R. (1953) Brain, 76, 70.
MAYFIELD, F. H. (1955) Clin. Neurosurg. 2, 83.
MIXTER, W. J., and AYER, J. B. (1935) New Engl. J. Med. 213, 385.
MOODIE, R. L. (1923) Paleopathology, an introduction to the study of ancient evidence
disease. Illinois, Urbana.
MORTON, D. E. (1950) Yale J. Biol. Med. 23, 126.
NORTHFIELD, D. W. C. (1955) Brit. med. J. 2, 1474.
PAYNE, E. E., and SPILLANE, J. D. (1957) Brain, 80,571.
PERESE, D. M., and FRACASSO, J. E. (1959) J. Neurosurg. 16, 314.
PUSCHEL, J. (1930) Beitr. path. Anat. 84, 123.
ROMANES, G. (1962) Personal communication.
RUFFER, M. A., and RIETTI, A. Quoted by CLARKE, 1955.
SAHS, A. L. (1942) J. comp. neurol. 76, 403.
SAUNDERS, J. B., and INMAN, V. T. (1940) Arch. Surg. 40, 389.
SCOVILLE, W. B. (1961) J. Neurosurg. 18, 423.
SEGERBERG, L. H. (1956) Amer. Surg. 22, 227.
SPILLANE, J. D., and LLOYD, G. H. T. (1951) Lancet, 2, 653.
SPILLER, WV. G. (1908) J. nerv. ment. Dis. 35, 775.
SPURLING, R. G., and SCOVILLE, W. B. (1944) Surg. Gynec. Obstet. 78, 350.
STERZI, G. Quoted by HOSKINs, E. R. (1914) Anat. Rec. 8, 371.
STOOKEY, B. (1940) Arch. Surg. 40, 417.
STOOPS, W. L., and KING, R. B. (1962) J. Neurosurg. 19, 986.
SUH, T. H., and ALEXANDER, L. (1939) Arch. Neurol. Psychiat. 41, 659.
TAYLOR, A. R. (1953) Lancet, 1, 717.
(1964) Neurology, 14, 62.
WALSH, L. S., and MACKENZIE, I. (1956) In CLARKE, E., and ROBINSON, P. K., Brain,
79, 483.
WILKINSON, M. (1960) Brain, 83, 589.
WOLF, B. S., KHILNANI, M., and MALIS, L. (1956) J. Mt. Sinai Hosp. 23, 283.
WOOLLAM, D. H. M., and MILLEN, J. W. (1958) Proc. Roy. Soc. Med. 51, 540.
Yoss, R. E. (1950) Univ. Mich. med. Bull. 16, 333.
ZULCH, K. J. (1954) Dtsch. Z. Nervenheilk. 172, 81.

251