Neuropathology and Applied Neurobiology 1978, 4, 151-162
T H E C E R V I C A L CORD IN M U L T I P L E S C L E R O S I S
D. R. OPPENHEIMER Department of Neuropathology, Radcliffe Infirmary, Oxford Accepted for publication 17 October 1977
Oppenheimer D.R. (1978) Neuropathology and Applied Neurobiology 4, 151-162 The cervical cord in multiple sclerosis The spinal cords were examined i n eighteen cases of multiple sclerosis, with special attention t o the cervical enlargement. It was found that (1) lesions i n the cervical cord are about twice as common as at lower levels, (2) i n this region there is a striking preponderanceof fan-shaped lesions in the lateral columns. It is argued that both these findings are explicable on the theory that mechanical stresses play a part i n determining the site of lesions; that such stresses are commonly transmitted t o the cord via the denticulate ligaments during flexion of the spine; and thatmany of the lesions are attributable to vascular leakages due to tension in the denticulate ligaments. It is concluded that in patients with multiple sclerosis neck flexion is dangerous-especially in cases where Lhermitte’s sign has occurred.
Introduction
It is reasonable to hope that increased knowledge of the causes of multiple sclerosis will lead, during the next few decades, to effective measures reducing t,he incidence of the disease. For the effective management of patients, however, what is needed is a better understanding of the mechanisms of plaque formation, rather than of the causes of the disease. Knowledge on this subject is a t present rudimentary, and progress is hampered by the lack of an acceptable animal model of the disease. The first step in this direction should be the careful and imaginative examination of lesions, by whatever methods are available. These include the older techniques of dissection and histopathology. It is now over a hundred years since Charcot described the pathology of the disease, and it might be thought that there was nothing left to be learnt from these methods. I do not think this is the case. Among the most disabling lesions of multiple sclerosis are those which interfere with conduction in the long tracts of the spinal cord. The following observations provide suggestive evidence on how they arise, and why they are commoner in some situations than in others. 0305-l846/78/0400-0l51 $02.00 Q 1978 Blackwell Scientific Publications
151
152
D. R. Oppenheimer
Materials and methods Betwecn the years 1971 and 1976 I acquired the spinal cords, i n a good state of preservation, from twenty cases of multiple sclerosis. Two of these were excluded from the present study: i n one, the whole of the cervical enlargement, was demyelinated, and there was nothing t o be gained by mapping the lesions, while t h e other was a case of fulminating type, with intense oedema of t h e cord, making it impossible t o delineate the edges o f plaques. Of the remaining eighteen cases, fourteen were male, four female. Ages at death ranged from 40 to 83, six being over 70 years. In one case (iVo. 561) the disease was subacute, with a course lasting 18 months. In the rest it was chronic. In all but two cases the cord was fixed within the spinal column-a procedure which avoids post rnortern distortion, and enables one to examine nerve roots and ganglia, as well as the bony column itself. After fixation the cords were routinely exposed from behind, and dissected out of the spinal canal still inside the dura. Notes were made of any fibrous adhesions between the dura and the posterior longitudinal ligament; of osteophytic protrusions opposite disc spaces ; and of signs of anteroposterior compression of the spinal cord. The upper part of the spine was then sawn sagittally, and notes made on changes i n intervertebral discs. The cords were examined i n tranverse sections a t ten or more levels, always including the lower four cervical and the first thoracic segments. The distribution of plaques a t different levels was noted, and camera lucida tracings were made from myelin-stained sections of the cervical enlargement (Figure 2). Only lesions with clearly defined margins were included i n the tracings. Shadow plaques (which i n any case were rare) and myelin loss from Wallerian degeneration were disregarded.
Observations Spinal column Severe degenerative changes in the cervical spine were found in Cases 608,790 and 3304 (Figure 1).In Case 3304, this had resulted in some anteroposterior flattening of cord segments C8 and T1. In Case 1633 there were multiple posterior osteophytic bars, opposite degenerate discs, again causing flattening of the cord at levels C 8 and T1. Minor degrees of disc degeneration were very common, especially in the elderly subjects. Associated with this was a tendency of the spinal dura to form fibrous adhesions with the posterior longitudinal ligament. In these cases a kzife was needed in order to separate the dura from the anterior wall of the spinal canal.
Distribution of lesions in CNS In every case except 2619 there were multiple plaques in the brain. In all but two cases, it was noted that the most conspicuous cord lesions were in the cervical segments, often extending to the topmost thoracic segments. In Cases 133 and 1951 these were the only lesions found in the cord. The lower thoracic levels tended to be spared. Cases 790,1632 and 2619 showed lesions in the cervical and lumbar enlargements, but not in between. Case 520 was exceptional in that the whole thoracic region was severely affected, with visible wasting of the cord. In no case was there any doubt of the diagnosis of multiple sclerosis.
Cervical cord i n multiple sclerosis
Figure 1. Cervical spines, sawn sagittally, with the surface of the spinal canal facing outwards. (a) Case 608, (b) Case 790, (c) Case 3304. In (a) there is loss of cervical lordosis. Anterior osteophytes are prominent, but the canal is not greatly narrowed. In (b) there is exaggeration of the normal lordosis. In (c) there is a fairly normal curvature, but the canal is narrowed, and the cord has undergone some antero-posterior flattening (see Figure 2, Case 3304).
153
154
133 M 40
D.R. Oppenheimer 5 2 0 M 51
561 F 50
608 M 68
785 F 70
790 F 83
@@@
@-m e m @
982 M . 4 3
1551 M.77
1632 M.55
1633 M.72
1893 M.4 0
1951 M.M)
Cervical cord in multiple sclerosis
2563 M.55
2619 F: 63
2646 M.45
155
3304 M.73
@
-U
Figure 2. Camera lucida tracings from myelin-stained sections of spinal cord from 18 cases of multiple sclerosis. Magnification x 6. Case number at top. Tracings in top row from level C5, i n second row from C6, and so on; bottom row, T1. Demyelinated areas i n black. The attachments of the denticulate ligaments, where present in the sections, are indicated in the tracings. Levels C6 and C8 in Case 2563 were not available, having been used for other purposes.
Patterns of lesions in the cervical enlargement
A first glance at Figure 2 gives the impression that the positions, shapes, and sizes of the lesions are completely haphazard. They do not respect the boundaries between grey matter and white, or between neighbouring tracts in the white matter, and show no correlation with territories of arterial supply or of venous drainage. In a number of the cases, however, there is a tendency, which can hardly be accidental, towards a rough bilateral symmetry at one or more levels (Cases 133,561, 1551, 1633, 1893,2563, 2646, 3303, 3304). As to location, there is a preponderance of lesions in the lateral
columns, abutting on the lateral surface, and having a roughly fan-shaped outline in transverse section. The single lesion in Case 790 is of this type. In all eighteen cases, the seventh cervical segment contains at least one lesion. Figure 3 gives a composite picture of plaques at this level in all eighteen cases, and clearly shows the relative frequency of lesions in the lateral columns. Lesions in the posterior columns tend to lie in the midline (Cases 561, 1893, 2563, 3138, 3304). Lesions are relatively uncommon in the anterior columns; when present, they tend to be symmetrical (Cases 133, 520,2619). Another recurring pattern, seen in
156
D. R. Oppenheimer
Figure 3. Composite photograph, showing location of plaques in the seventh cervical segment in eighteen cases. The lesions have been drawn, as in Figure 2, within a standard outline of the cord. There is an obvious preponderance of fan-shaped lesions i n the lateral columns
Cases 133, 520, 561, 785 and 1633, is a narrow band of demyelination running between the lateral tip of a n anterior grey horn and the lateral surface of the cord. No attempt has been made to reconstruct individual plaques in three dimensions from serial sections. From naked-eye inspection of the cords it is clear that many of the lesions-particularly those in the lateral columns-extend upwards and downwards through several cord segments and tend to taper a t their upper and lower ends. Other lesions are confined within 2 or 3 mm of the length of the cord. A rough idea of this variation in length of plaques can be got by looking up and down the vertical columns in Figure 2.
Relation of lesions to vessels, meninges, and nerve roots The fact that many plaques arise on the course of a small vein is now too wellestablished t o need reiteration. An example is given in Figure 4. This relationship can only be shown convincingly for small plaques, since larger plaques contain several venules (Figure 5). Only a few of the lesions in this material are clearly related to veins, and these are mostly around small radially-running veins in the lateral columns (Figure 6). Regarding the numerous fan-shaped plaques abutting on the pial surface of the lateral white columns, one can hardly ignore the frequent apparent relationship t o the line of attachment of a denticulate ligament (examples: Case 520, C8; 561, C6; 608, C6; 790, C7; 1633, C6, 7 and 8; 1893, C8; 1951, C8 andT1; 3138, C5). Thisimpression is reinforced by the composite picture (Figure 3), which shows the lateral column lesions extending equally behind and in front of the denticulate attachment. In Cases 561, 1551, 1633, 1893, 2563, 3303 and 3304 plaques in this situation are the principal, if not the only, lesions. Suspicion that the denticulate ligaments may be related to plaque formation is increased, in some cases, by the outline of the cord, which appears
Cervical cord in multiple sclerosis
157
Figure 4. Case 1633, level T5. Plaque related to a radial vein in the posterior columns. Klu verBarrera stain. x 32.
Figure 5. Case 1633, leveI C8. Plaque i n lateral column. Denticulate ligament on left. The lesion contains numerous small veins. Kluver-Barrera stain. x 32.
as if distorted by a n outward tug a t the line of attachment of the ligament (examples: Case 561, C6; 608, 6C; 1633, C8;1951, T1; 3138, C5; 3303, C 6 ; 3304, C5, 6, 7 and 8). In two cases (1951, T1; 2646, C6) plaques appear to be related to the tips of the POSterior horns, ie. the point of entry of posterior nerve roots. In other cases, there may well have been small lesions in this situation which were overlooked, because there
158
D. R. Oppenheimer
Figure 6. Case 520, level C6. Plaque i n anterolateral column, related to a radial vein. Denticulate ligament on right. Kluver-Barrera stain. x32.
is so little myelin in this region in the normal cord. There is no clear relationship of plaques to the exit zones of motor roots; dubious instances include Cases 133, 561 and 2646, all a t level T1.
Comparison of various cord levels Lesions in the upper cervical cord do not appear to differ, in frequency or in location, from those in the cervical enlargement. Out of eighteen blocks, sixteen (89%) show MS lesions, the corresponding figures for the cervical enlargement being seventy-nine out of eighty-eight (90%). I n one case (2646) there is a n almost total transverse demyelination a t level C3. The frequency of lesions in the cord below this is significantly less. Sixty-one blocks of levels T2 t o L2 inclusive show only twenty-eight (46%) with lesions, and thirty-two blocks from L3 downwards show thirteen (41%) with lesions. In other words, lesions in the cervical cord are about twice as frequent as those a t lower levels. In view of the suspicion, mentioned above, that many of the plaques i n the cervical cord could be related to the lines of attachment of denticulate ligaments, a search was made for similar lesions in the lumbosacral region. The denticles which anchor the ligament to the spinal dura cease in the upper lumbar region. Below this, the ligament continues downwards as a ribbon of fibrous tissue attached to the pia on each side of the cord. It is not certain whether, or in what circumstances, this ribbon comes under tension. It was found that lesions in the lateral columns, corresponding t o those in the cervical cord, were present in seven out of thirty-two blocks (22%) of lower lumbar and sacral segments. All these lesions were small.
Summary of observations All eighteen cases had lesions in the cervical enlargement, the seventh cervical segment being the most frequently affected. The commonest site was in the lateral white columns, extending upwards and downwards. Lesions in the cervical cord were
Cervical cord in multiple sclerosis
159
about as numerous as at lower levels. A few of the lesions were clearly related to veins. Many of the lesions in the lateral columns appeared to be related to denticulate ligaments.
Discussion The best known work on the topography of plaques in the spinal cord is that of Fog (1950). Fog examined the cords of eight cases of multiple sclerosis in great detail, and drew maps of the lesions, many of them from serial sections, at numerous levels. He noted certain characteristic shapes, including the fan-shaped plaques in the lateral columns which are described above. From this mass of work he drew one conclusion, and one only: that plaques originate in the immediate vicinity of veins. (Rather confusingly, Fog uses the term ‘venous territory’ to mean the area immediately surrounding a vein and its tributaries, rather than in its common meaning of ‘area of venous drainage’). He did not speculate on why plaques should be related to veins, or on why some veins should be more commonly affected than others; and he did not consider any factors involved in plaque formation other than proximity to veins. In the material presented here, some small plaques are clearly related to veins, but it is by no means clear that all of them are so related. Many of them are far too large to be clearly related to any single small structure. What is more, the pattern of lesions seen in cases of perivenous encephalomyelitis (Figure 7), with multiple narrow sleeves of myelin destruction around venules, was never seen in this material. The fact that many plaques, in both brain and cord, are obviously centred on small veins immediately suggests that a substance toxic to myelin has diffused from
Figure 7. Case of acute perivenous encephalomyelitis. Upper lumbar cord. Multiple sleeves of perivenous. locally confluent, myelin destruction. Kliiver-Barrera stain. x 15.
160
D.R. Oppenheimer
the blood stream into the parenchyma. This idea, which was current a t least 70 years ago (Marburg 1906, Hallervorden & Spatz, 1932-3), remains highly plausible, especially since the demonstration by Broman (1947) of abnormally leaky vessels in plaques of multiple sclerosis and by Bornstein (1963) of the effect on myelinating tissue cultures of sera from patients with multiple sclerosis. The hypothesis implies merely the presence of ‘myelinotoxin’ in the blood. This could reach the parenchyma, not only from leaky veins, but from leaky capillaries or via the cerebrospinal fluid. For obvious reasons, the relationship of a plaque to a leaky capillary would be almost impossible t o demonstrate; but there is no theoretical objection to the idea that some plaques are due to capillary leakage. If one asks what kinds of disturbance might cause local defects in the blood-brain barrier, the obvious suggestions are (a) a local rise in venous pressure, (b) focal inflammation and (c) mechanical stresses, especially longitudinal stretching of vessels. Assuming (a large but reasonable assumption) that plaques originate, and enlarge, a t sites of vascular leakage, we have t o enquire why such leakages should be more frequent in the cervical cord than in other parts, and occur most frequently i n the lateral columns. There is no reason to suppose that the lateral columns of the cervical cord are particularly vulnerable to changes in venous pressure or t o inflammation; on the other hand it is fairly certain that adaptive deformation of the cord is greatest where the vertebral column is most mobile, ie. in the cervical region; and that the principal transmitters of forces between the dura and the cord are the denticulate ligaments. Direct observations of the effects of neck movements on the contents of the spinal canal (Reid, 1958; 1960; Breig, 1960; Adams & Logue, 1971) have established that the canal becomes longer in flexion of the neck, and that the dura and spinal cord lengthen correspondingly. In extension, the cord is shorter and fatter, and the nerve roots and denticulate ligaments are slack, while in full flexion the cord is longer and thinner, and the denticulate ligaments, and i n some circumstances the posterior roots, become tense. There is clinical and experimental evidence that such tension may be sufficient to interfere with the functions of long tracts. Alajouanine, Thurel & Papaioanou (1949) described ten cases, with lesions of the cervical spine and cord other than multiple sclerosis, in which ‘Lhermitte’s sign’ had been observed. They conclude (my translation) : ‘Whereas Lhermitte regards demyelination as the necessary condition for the production of the sensation of electric shock provoked by head flexion, we are much more inclined to incriminate meningeal adhesions, and the tugs exerted by these on the posterior columns during flexion movements of the head’. A similar conclusion was reached by Bedford, Bosanquet & Russell (1952), who described a case of myelopathy due to cervical spondylosis, in which thickened denticulate ligaments and degenerated nerve roots were closely related to an area of cord damage. Cusick, Ackmann & Larson (1977) describe a case of cervical myelopathy, unaffected by simple laminectomy, but showing marked improvement following cervical dentatotomy. They followed this up with experiments on dogs, in which they showed changes in sensory evoked potentials, caused by mild stretching of the cord, which were lessened by dentatotomy. Just how tension in the denticulate ligaments results
Cervical cord in multiple sclerosis
161
in disturbances of nervous conduction is not quite clear; but presumably the force is transmitted through the pia mater and the blood vessels, which constitute the most solid elements of the cord. It would not be surprising if small and transient leakages of plasma occurred when the cord and its vessels are deformed during movements of the neck; and it is consistent with what is known of the pathogenesis of plaques t o relate their distribution in the spinal cord to vascular leakage caused by mechanical stresses. The part, if any, played by cervical spondylosis in provoking or aggravating the lesions of multiple sclerosis remains uncertain. Brain & Wilkinson (1957) discuss the not uncommon association of the two conditions, and describe seventeen patients suffering from both. Necropsy on two of these showed a preponderance of plaques in the cervical cord, opposite the spondylotic protrusions, suggesting that they had been provoked by local pressure. If, as the present study suggests, this part of the cord is the most liable to contain plaques even i n the absence of cervical spondylosis, this theory loses its force. In the present series, three out of eighteen cases showed severe spondylosis (see Figure 1);but the cord lesions in these three do not differ in any obvious way from the rest. They do not appear to be related to points of compression by spondylotic ‘bars’, and in Case 3304, where anteroposterior flattening was most marked, the main lesions are of the common type, in the lateral columns. If, i n fact, spondylosis plays a part in determining the situation of plaques in the cervical cord, it is probably not by direct compression, but rather on the lines suggested by Bedford et al. (1952)-ie., by the tethering effect on the dura of fibrous adhesions between it and the posterior longitudinal ligament. If the dural attachments of the denticulate ligaments are fixed relative to the bony canal, they will become taut at a n earlier stage of neck flexion. More needs to be known about the actual tensions in the denticulate ligaments during flexion in normal subjects and in those with spondylosis. Too little is at present known about the mechanical effects of lateral bending and twisting movements of the neck to justify speculation on the stresses set up in the cord during these movements. Summary and conclusions The lesions of multiple sclerosis may occur in any part of the spinal cord, but some parts are more vulnerable than others. Plaques are about as common in the cervical cord as in lower segments. Within the cervical enlargement, plaques are commonest in the lateral columns, the typical lesion being fan-shaped i n cross-section, tapering upwards and downwards, and abutting on the lateral surface of the cord. The form and position of these plaques strongly suggest that the denticulate ligaments play a part in their formation. The hypothesis that mechanical stresses, transmitted to the cord through the denticulate ligaments, are partly responsible for plaque formation does not conflict with the old view that plaques arise in relation to small intramedullary veins. If, as seems probable, perivenous lesions are due to leakage of plasma from small vessels, mechanical stresses are among the most likely causes of such leakage.
162
D.R. Oppenheimer
The kind of stresses to which the cervical cord is exposed may be inferred from two well-attested facts: (1)the spinal column is most mobile in the cervical region, and (2) during flexion and extension of the neck the cord undergoes deformation, being longer in flexion and fatter in extension. The denticulate ligaments (perhaps also the posterior nerve roots) become taut in full flexion, and in some pathological conditions this tension may cause damage to the cord. If this theory is correct, it has a n obvious practical application-namely that patients with multiple sclerosis should in some way be protected against the effects of full flexion of the head and neck. Patients who have experienced Lhermitte’s sign should be regarded as those most in need of such protection.
References ADAMSC.B.T. & LOGUEV. (1971) Studies in spondylotic myelopathy. I. Movement of the cervical roots, dura and cord, and their relation to the course taken by the extrathecal roots. Brain 94, 557-568 ALAJOUANINE TH., THUREL R. & PAPAIOANOU C. (1949) La douleur a type de dbcharge dectrique, provoqu6e par la flexion de la t6te e t parcourant le corps de haut e n bas. Revue Neurologique 81, 89-97 BEDFORD P.D., BOSANQUET F.D. & RUSSELLW.R. (1952) Degeneration of the cervical cord associated with cervical spondylosis. Lancet ii, 55-59 BORNSTEIN M.B. (1963) A tissue-culture approach t o demyelinative disorders. National Cancer Institute Monograph 11, 197-214 BRAINR. & WILKINSON M. (1957) The association of cervical spondylosis and disseminated sclerosis. Brain 80, 456-478 BREIGA. (1960) Biomechanics of the Central Nervous Sjstem. Almqvist & Wiksell, Stockholm BROMAN T. (1947) Supravital analysis of disorders in the cerebral vascular permability. 11. Two cases of multiple sclerosis. Acta psychiatrica et neurologica Scandinavica Suppl. 46 CUSICKJ.F., ACKMANN J.J. & LARSON S.J. (1977) Mechanical and physiological effects of dentatotomy. Journal of Neurosurgery 46, 767-775 FOGT. (1950) Topographical distribution of plaques i n the spinal cord i n multiple sclerosis. Archives of Neurology and Psychiatry 63, 382-414 HALLERVORDEN J. & SPATZH. (1932-3) Uber die konzentrische Sklerose. Archiv fiir Psychiatrie und Neruenkranhheiten 98, 641-701 MARBURG 0 . (1906) Die sogenannte ‘akute multiple Sklerose’. Jahrbuch fur Psychiatrie uizd Neurologie 27, 211-312 REIDJ.D. (1958) Ascending nerve roots and tightness of dura mater. New Zealand Medical Journal 57, 16-26 REIDJ.D. (1960) Effects of flexion-extension movements of the head and spine upon the spinal cord and nerve roots. Journal of Neurology, Neurosurgery and Psychiatry 23, 214-221
T H E C E R V I C A L CORD IN M U L T I P L E S C L E R O S I S
D. R. OPPENHEIMER Department of Neuropathology, Radcliffe Infirmary, Oxford Accepted for publication 17 October 1977
Oppenheimer D.R. (1978) Neuropathology and Applied Neurobiology 4, 151-162 The cervical cord in multiple sclerosis The spinal cords were examined i n eighteen cases of multiple sclerosis, with special attention t o the cervical enlargement. It was found that (1) lesions i n the cervical cord are about twice as common as at lower levels, (2) i n this region there is a striking preponderanceof fan-shaped lesions in the lateral columns. It is argued that both these findings are explicable on the theory that mechanical stresses play a part i n determining the site of lesions; that such stresses are commonly transmitted t o the cord via the denticulate ligaments during flexion of the spine; and thatmany of the lesions are attributable to vascular leakages due to tension in the denticulate ligaments. It is concluded that in patients with multiple sclerosis neck flexion is dangerous-especially in cases where Lhermitte’s sign has occurred.
Introduction
It is reasonable to hope that increased knowledge of the causes of multiple sclerosis will lead, during the next few decades, to effective measures reducing t,he incidence of the disease. For the effective management of patients, however, what is needed is a better understanding of the mechanisms of plaque formation, rather than of the causes of the disease. Knowledge on this subject is a t present rudimentary, and progress is hampered by the lack of an acceptable animal model of the disease. The first step in this direction should be the careful and imaginative examination of lesions, by whatever methods are available. These include the older techniques of dissection and histopathology. It is now over a hundred years since Charcot described the pathology of the disease, and it might be thought that there was nothing left to be learnt from these methods. I do not think this is the case. Among the most disabling lesions of multiple sclerosis are those which interfere with conduction in the long tracts of the spinal cord. The following observations provide suggestive evidence on how they arise, and why they are commoner in some situations than in others. 0305-l846/78/0400-0l51 $02.00 Q 1978 Blackwell Scientific Publications
151
152
D. R. Oppenheimer
Materials and methods Betwecn the years 1971 and 1976 I acquired the spinal cords, i n a good state of preservation, from twenty cases of multiple sclerosis. Two of these were excluded from the present study: i n one, the whole of the cervical enlargement, was demyelinated, and there was nothing t o be gained by mapping the lesions, while t h e other was a case of fulminating type, with intense oedema of t h e cord, making it impossible t o delineate the edges o f plaques. Of the remaining eighteen cases, fourteen were male, four female. Ages at death ranged from 40 to 83, six being over 70 years. In one case (iVo. 561) the disease was subacute, with a course lasting 18 months. In the rest it was chronic. In all but two cases the cord was fixed within the spinal column-a procedure which avoids post rnortern distortion, and enables one to examine nerve roots and ganglia, as well as the bony column itself. After fixation the cords were routinely exposed from behind, and dissected out of the spinal canal still inside the dura. Notes were made of any fibrous adhesions between the dura and the posterior longitudinal ligament; of osteophytic protrusions opposite disc spaces ; and of signs of anteroposterior compression of the spinal cord. The upper part of the spine was then sawn sagittally, and notes made on changes i n intervertebral discs. The cords were examined i n tranverse sections a t ten or more levels, always including the lower four cervical and the first thoracic segments. The distribution of plaques a t different levels was noted, and camera lucida tracings were made from myelin-stained sections of the cervical enlargement (Figure 2). Only lesions with clearly defined margins were included i n the tracings. Shadow plaques (which i n any case were rare) and myelin loss from Wallerian degeneration were disregarded.
Observations Spinal column Severe degenerative changes in the cervical spine were found in Cases 608,790 and 3304 (Figure 1).In Case 3304, this had resulted in some anteroposterior flattening of cord segments C8 and T1. In Case 1633 there were multiple posterior osteophytic bars, opposite degenerate discs, again causing flattening of the cord at levels C 8 and T1. Minor degrees of disc degeneration were very common, especially in the elderly subjects. Associated with this was a tendency of the spinal dura to form fibrous adhesions with the posterior longitudinal ligament. In these cases a kzife was needed in order to separate the dura from the anterior wall of the spinal canal.
Distribution of lesions in CNS In every case except 2619 there were multiple plaques in the brain. In all but two cases, it was noted that the most conspicuous cord lesions were in the cervical segments, often extending to the topmost thoracic segments. In Cases 133 and 1951 these were the only lesions found in the cord. The lower thoracic levels tended to be spared. Cases 790,1632 and 2619 showed lesions in the cervical and lumbar enlargements, but not in between. Case 520 was exceptional in that the whole thoracic region was severely affected, with visible wasting of the cord. In no case was there any doubt of the diagnosis of multiple sclerosis.
Cervical cord i n multiple sclerosis
Figure 1. Cervical spines, sawn sagittally, with the surface of the spinal canal facing outwards. (a) Case 608, (b) Case 790, (c) Case 3304. In (a) there is loss of cervical lordosis. Anterior osteophytes are prominent, but the canal is not greatly narrowed. In (b) there is exaggeration of the normal lordosis. In (c) there is a fairly normal curvature, but the canal is narrowed, and the cord has undergone some antero-posterior flattening (see Figure 2, Case 3304).
153
154
133 M 40
D.R. Oppenheimer 5 2 0 M 51
561 F 50
608 M 68
785 F 70
790 F 83
@@@
@-m e m @
982 M . 4 3
1551 M.77
1632 M.55
1633 M.72
1893 M.4 0
1951 M.M)
Cervical cord in multiple sclerosis
2563 M.55
2619 F: 63
2646 M.45
155
3304 M.73
@
-U
Figure 2. Camera lucida tracings from myelin-stained sections of spinal cord from 18 cases of multiple sclerosis. Magnification x 6. Case number at top. Tracings in top row from level C5, i n second row from C6, and so on; bottom row, T1. Demyelinated areas i n black. The attachments of the denticulate ligaments, where present in the sections, are indicated in the tracings. Levels C6 and C8 in Case 2563 were not available, having been used for other purposes.
Patterns of lesions in the cervical enlargement
A first glance at Figure 2 gives the impression that the positions, shapes, and sizes of the lesions are completely haphazard. They do not respect the boundaries between grey matter and white, or between neighbouring tracts in the white matter, and show no correlation with territories of arterial supply or of venous drainage. In a number of the cases, however, there is a tendency, which can hardly be accidental, towards a rough bilateral symmetry at one or more levels (Cases 133,561, 1551, 1633, 1893,2563, 2646, 3303, 3304). As to location, there is a preponderance of lesions in the lateral
columns, abutting on the lateral surface, and having a roughly fan-shaped outline in transverse section. The single lesion in Case 790 is of this type. In all eighteen cases, the seventh cervical segment contains at least one lesion. Figure 3 gives a composite picture of plaques at this level in all eighteen cases, and clearly shows the relative frequency of lesions in the lateral columns. Lesions in the posterior columns tend to lie in the midline (Cases 561, 1893, 2563, 3138, 3304). Lesions are relatively uncommon in the anterior columns; when present, they tend to be symmetrical (Cases 133, 520,2619). Another recurring pattern, seen in
156
D. R. Oppenheimer
Figure 3. Composite photograph, showing location of plaques in the seventh cervical segment in eighteen cases. The lesions have been drawn, as in Figure 2, within a standard outline of the cord. There is an obvious preponderance of fan-shaped lesions i n the lateral columns
Cases 133, 520, 561, 785 and 1633, is a narrow band of demyelination running between the lateral tip of a n anterior grey horn and the lateral surface of the cord. No attempt has been made to reconstruct individual plaques in three dimensions from serial sections. From naked-eye inspection of the cords it is clear that many of the lesions-particularly those in the lateral columns-extend upwards and downwards through several cord segments and tend to taper a t their upper and lower ends. Other lesions are confined within 2 or 3 mm of the length of the cord. A rough idea of this variation in length of plaques can be got by looking up and down the vertical columns in Figure 2.
Relation of lesions to vessels, meninges, and nerve roots The fact that many plaques arise on the course of a small vein is now too wellestablished t o need reiteration. An example is given in Figure 4. This relationship can only be shown convincingly for small plaques, since larger plaques contain several venules (Figure 5). Only a few of the lesions in this material are clearly related to veins, and these are mostly around small radially-running veins in the lateral columns (Figure 6). Regarding the numerous fan-shaped plaques abutting on the pial surface of the lateral white columns, one can hardly ignore the frequent apparent relationship t o the line of attachment of a denticulate ligament (examples: Case 520, C8; 561, C6; 608, C6; 790, C7; 1633, C6, 7 and 8; 1893, C8; 1951, C8 andT1; 3138, C5). Thisimpression is reinforced by the composite picture (Figure 3), which shows the lateral column lesions extending equally behind and in front of the denticulate attachment. In Cases 561, 1551, 1633, 1893, 2563, 3303 and 3304 plaques in this situation are the principal, if not the only, lesions. Suspicion that the denticulate ligaments may be related to plaque formation is increased, in some cases, by the outline of the cord, which appears
Cervical cord in multiple sclerosis
157
Figure 4. Case 1633, level T5. Plaque related to a radial vein in the posterior columns. Klu verBarrera stain. x 32.
Figure 5. Case 1633, leveI C8. Plaque i n lateral column. Denticulate ligament on left. The lesion contains numerous small veins. Kluver-Barrera stain. x 32.
as if distorted by a n outward tug a t the line of attachment of the ligament (examples: Case 561, C6; 608, 6C; 1633, C8;1951, T1; 3138, C5; 3303, C 6 ; 3304, C5, 6, 7 and 8). In two cases (1951, T1; 2646, C6) plaques appear to be related to the tips of the POSterior horns, ie. the point of entry of posterior nerve roots. In other cases, there may well have been small lesions in this situation which were overlooked, because there
158
D. R. Oppenheimer
Figure 6. Case 520, level C6. Plaque i n anterolateral column, related to a radial vein. Denticulate ligament on right. Kluver-Barrera stain. x32.
is so little myelin in this region in the normal cord. There is no clear relationship of plaques to the exit zones of motor roots; dubious instances include Cases 133, 561 and 2646, all a t level T1.
Comparison of various cord levels Lesions in the upper cervical cord do not appear to differ, in frequency or in location, from those in the cervical enlargement. Out of eighteen blocks, sixteen (89%) show MS lesions, the corresponding figures for the cervical enlargement being seventy-nine out of eighty-eight (90%). I n one case (2646) there is a n almost total transverse demyelination a t level C3. The frequency of lesions in the cord below this is significantly less. Sixty-one blocks of levels T2 t o L2 inclusive show only twenty-eight (46%) with lesions, and thirty-two blocks from L3 downwards show thirteen (41%) with lesions. In other words, lesions in the cervical cord are about twice as frequent as those a t lower levels. In view of the suspicion, mentioned above, that many of the plaques i n the cervical cord could be related to the lines of attachment of denticulate ligaments, a search was made for similar lesions in the lumbosacral region. The denticles which anchor the ligament to the spinal dura cease in the upper lumbar region. Below this, the ligament continues downwards as a ribbon of fibrous tissue attached to the pia on each side of the cord. It is not certain whether, or in what circumstances, this ribbon comes under tension. It was found that lesions in the lateral columns, corresponding t o those in the cervical cord, were present in seven out of thirty-two blocks (22%) of lower lumbar and sacral segments. All these lesions were small.
Summary of observations All eighteen cases had lesions in the cervical enlargement, the seventh cervical segment being the most frequently affected. The commonest site was in the lateral white columns, extending upwards and downwards. Lesions in the cervical cord were
Cervical cord in multiple sclerosis
159
about as numerous as at lower levels. A few of the lesions were clearly related to veins. Many of the lesions in the lateral columns appeared to be related to denticulate ligaments.
Discussion The best known work on the topography of plaques in the spinal cord is that of Fog (1950). Fog examined the cords of eight cases of multiple sclerosis in great detail, and drew maps of the lesions, many of them from serial sections, at numerous levels. He noted certain characteristic shapes, including the fan-shaped plaques in the lateral columns which are described above. From this mass of work he drew one conclusion, and one only: that plaques originate in the immediate vicinity of veins. (Rather confusingly, Fog uses the term ‘venous territory’ to mean the area immediately surrounding a vein and its tributaries, rather than in its common meaning of ‘area of venous drainage’). He did not speculate on why plaques should be related to veins, or on why some veins should be more commonly affected than others; and he did not consider any factors involved in plaque formation other than proximity to veins. In the material presented here, some small plaques are clearly related to veins, but it is by no means clear that all of them are so related. Many of them are far too large to be clearly related to any single small structure. What is more, the pattern of lesions seen in cases of perivenous encephalomyelitis (Figure 7), with multiple narrow sleeves of myelin destruction around venules, was never seen in this material. The fact that many plaques, in both brain and cord, are obviously centred on small veins immediately suggests that a substance toxic to myelin has diffused from
Figure 7. Case of acute perivenous encephalomyelitis. Upper lumbar cord. Multiple sleeves of perivenous. locally confluent, myelin destruction. Kliiver-Barrera stain. x 15.
160
D.R. Oppenheimer
the blood stream into the parenchyma. This idea, which was current a t least 70 years ago (Marburg 1906, Hallervorden & Spatz, 1932-3), remains highly plausible, especially since the demonstration by Broman (1947) of abnormally leaky vessels in plaques of multiple sclerosis and by Bornstein (1963) of the effect on myelinating tissue cultures of sera from patients with multiple sclerosis. The hypothesis implies merely the presence of ‘myelinotoxin’ in the blood. This could reach the parenchyma, not only from leaky veins, but from leaky capillaries or via the cerebrospinal fluid. For obvious reasons, the relationship of a plaque to a leaky capillary would be almost impossible t o demonstrate; but there is no theoretical objection to the idea that some plaques are due to capillary leakage. If one asks what kinds of disturbance might cause local defects in the blood-brain barrier, the obvious suggestions are (a) a local rise in venous pressure, (b) focal inflammation and (c) mechanical stresses, especially longitudinal stretching of vessels. Assuming (a large but reasonable assumption) that plaques originate, and enlarge, a t sites of vascular leakage, we have t o enquire why such leakages should be more frequent in the cervical cord than in other parts, and occur most frequently i n the lateral columns. There is no reason to suppose that the lateral columns of the cervical cord are particularly vulnerable to changes in venous pressure or t o inflammation; on the other hand it is fairly certain that adaptive deformation of the cord is greatest where the vertebral column is most mobile, ie. in the cervical region; and that the principal transmitters of forces between the dura and the cord are the denticulate ligaments. Direct observations of the effects of neck movements on the contents of the spinal canal (Reid, 1958; 1960; Breig, 1960; Adams & Logue, 1971) have established that the canal becomes longer in flexion of the neck, and that the dura and spinal cord lengthen correspondingly. In extension, the cord is shorter and fatter, and the nerve roots and denticulate ligaments are slack, while in full flexion the cord is longer and thinner, and the denticulate ligaments, and i n some circumstances the posterior roots, become tense. There is clinical and experimental evidence that such tension may be sufficient to interfere with the functions of long tracts. Alajouanine, Thurel & Papaioanou (1949) described ten cases, with lesions of the cervical spine and cord other than multiple sclerosis, in which ‘Lhermitte’s sign’ had been observed. They conclude (my translation) : ‘Whereas Lhermitte regards demyelination as the necessary condition for the production of the sensation of electric shock provoked by head flexion, we are much more inclined to incriminate meningeal adhesions, and the tugs exerted by these on the posterior columns during flexion movements of the head’. A similar conclusion was reached by Bedford, Bosanquet & Russell (1952), who described a case of myelopathy due to cervical spondylosis, in which thickened denticulate ligaments and degenerated nerve roots were closely related to an area of cord damage. Cusick, Ackmann & Larson (1977) describe a case of cervical myelopathy, unaffected by simple laminectomy, but showing marked improvement following cervical dentatotomy. They followed this up with experiments on dogs, in which they showed changes in sensory evoked potentials, caused by mild stretching of the cord, which were lessened by dentatotomy. Just how tension in the denticulate ligaments results
Cervical cord in multiple sclerosis
161
in disturbances of nervous conduction is not quite clear; but presumably the force is transmitted through the pia mater and the blood vessels, which constitute the most solid elements of the cord. It would not be surprising if small and transient leakages of plasma occurred when the cord and its vessels are deformed during movements of the neck; and it is consistent with what is known of the pathogenesis of plaques t o relate their distribution in the spinal cord to vascular leakage caused by mechanical stresses. The part, if any, played by cervical spondylosis in provoking or aggravating the lesions of multiple sclerosis remains uncertain. Brain & Wilkinson (1957) discuss the not uncommon association of the two conditions, and describe seventeen patients suffering from both. Necropsy on two of these showed a preponderance of plaques in the cervical cord, opposite the spondylotic protrusions, suggesting that they had been provoked by local pressure. If, as the present study suggests, this part of the cord is the most liable to contain plaques even i n the absence of cervical spondylosis, this theory loses its force. In the present series, three out of eighteen cases showed severe spondylosis (see Figure 1);but the cord lesions in these three do not differ in any obvious way from the rest. They do not appear to be related to points of compression by spondylotic ‘bars’, and in Case 3304, where anteroposterior flattening was most marked, the main lesions are of the common type, in the lateral columns. If, i n fact, spondylosis plays a part in determining the situation of plaques in the cervical cord, it is probably not by direct compression, but rather on the lines suggested by Bedford et al. (1952)-ie., by the tethering effect on the dura of fibrous adhesions between it and the posterior longitudinal ligament. If the dural attachments of the denticulate ligaments are fixed relative to the bony canal, they will become taut at a n earlier stage of neck flexion. More needs to be known about the actual tensions in the denticulate ligaments during flexion in normal subjects and in those with spondylosis. Too little is at present known about the mechanical effects of lateral bending and twisting movements of the neck to justify speculation on the stresses set up in the cord during these movements. Summary and conclusions The lesions of multiple sclerosis may occur in any part of the spinal cord, but some parts are more vulnerable than others. Plaques are about as common in the cervical cord as in lower segments. Within the cervical enlargement, plaques are commonest in the lateral columns, the typical lesion being fan-shaped i n cross-section, tapering upwards and downwards, and abutting on the lateral surface of the cord. The form and position of these plaques strongly suggest that the denticulate ligaments play a part in their formation. The hypothesis that mechanical stresses, transmitted to the cord through the denticulate ligaments, are partly responsible for plaque formation does not conflict with the old view that plaques arise in relation to small intramedullary veins. If, as seems probable, perivenous lesions are due to leakage of plasma from small vessels, mechanical stresses are among the most likely causes of such leakage.
162
D.R. Oppenheimer
The kind of stresses to which the cervical cord is exposed may be inferred from two well-attested facts: (1)the spinal column is most mobile in the cervical region, and (2) during flexion and extension of the neck the cord undergoes deformation, being longer in flexion and fatter in extension. The denticulate ligaments (perhaps also the posterior nerve roots) become taut in full flexion, and in some pathological conditions this tension may cause damage to the cord. If this theory is correct, it has a n obvious practical application-namely that patients with multiple sclerosis should in some way be protected against the effects of full flexion of the head and neck. Patients who have experienced Lhermitte’s sign should be regarded as those most in need of such protection.
References ADAMSC.B.T. & LOGUEV. (1971) Studies in spondylotic myelopathy. I. Movement of the cervical roots, dura and cord, and their relation to the course taken by the extrathecal roots. Brain 94, 557-568 ALAJOUANINE TH., THUREL R. & PAPAIOANOU C. (1949) La douleur a type de dbcharge dectrique, provoqu6e par la flexion de la t6te e t parcourant le corps de haut e n bas. Revue Neurologique 81, 89-97 BEDFORD P.D., BOSANQUET F.D. & RUSSELLW.R. (1952) Degeneration of the cervical cord associated with cervical spondylosis. Lancet ii, 55-59 BORNSTEIN M.B. (1963) A tissue-culture approach t o demyelinative disorders. National Cancer Institute Monograph 11, 197-214 BRAINR. & WILKINSON M. (1957) The association of cervical spondylosis and disseminated sclerosis. Brain 80, 456-478 BREIGA. (1960) Biomechanics of the Central Nervous Sjstem. Almqvist & Wiksell, Stockholm BROMAN T. (1947) Supravital analysis of disorders in the cerebral vascular permability. 11. Two cases of multiple sclerosis. Acta psychiatrica et neurologica Scandinavica Suppl. 46 CUSICKJ.F., ACKMANN J.J. & LARSON S.J. (1977) Mechanical and physiological effects of dentatotomy. Journal of Neurosurgery 46, 767-775 FOGT. (1950) Topographical distribution of plaques i n the spinal cord i n multiple sclerosis. Archives of Neurology and Psychiatry 63, 382-414 HALLERVORDEN J. & SPATZH. (1932-3) Uber die konzentrische Sklerose. Archiv fiir Psychiatrie und Neruenkranhheiten 98, 641-701 MARBURG 0 . (1906) Die sogenannte ‘akute multiple Sklerose’. Jahrbuch fur Psychiatrie uizd Neurologie 27, 211-312 REIDJ.D. (1958) Ascending nerve roots and tightness of dura mater. New Zealand Medical Journal 57, 16-26 REIDJ.D. (1960) Effects of flexion-extension movements of the head and spine upon the spinal cord and nerve roots. Journal of Neurology, Neurosurgery and Psychiatry 23, 214-221
