Ann. rheium. Dis.

(1972), 31, 345

Destructive lesions of vertebral

bodies in ankylosing spondylitis
M. I. D. CAWLEY., T. M. CHALMERS., J. H. KELLGREN, AND J. BALL
Rheumatism Research Clinic, University of Manchester

This paper describes a clinico-pathological study of at the outset to distinguish this lesion from those
destructive lesions of the vertebrae in ankylosing generally regarded as destructive lesions-a point to
spondylitis (AS), sometimes referred to as 'spondylo- which we will return later.
discitis'.

Review of the literature

The first report of such lesions is attributable to
Andersson (1937), who described sclerosis and
destruction around the disc margins in lumbar and
thoracic vertebrae in radiographs of two patients.
The earliest significant clinical observations were
made by Edstr6m (1940). In a paper on the nature of
AS, he referred to a male patient who, some years
after the onset of the disease, developed severe pains a b
in the back during a spell of unusually heavy work;
a year later radiographs showed vertebral destruction FIG. 1 Anterior spondylitis (a) erosion of attachment of
around the T12/L1 disc. Tuberculosis was suspected outer annulus (b) healed erosion producing syndesmophyte
but the lesion was not investigated pathologically.
In retrospect this clinical sequence is in some respects Jacqueline (1956) classified destructive lesions into
exemplary, yet it receives little emphasis in subsequent two groups according to whether they occurred in the
reports. Indeed, the purely clinical aspects of patients early stage of spinal ankylosis or in spines with
with destructive lesions are not well documented in advanced ossification, and he pointed out that in the
the literature and have not been made the subject of latter severe lesions might be provoked by trauma, a
a formal study. In 1951 the lesions were again briefly clear history of which was given by two of his three
mentioned by Guest and Jacobson (1951) and Fores- patients in this group. In this connection it may be
tier, Jacqueline, and Rotes-Querol (1951: trans. noted that Wilkinson and Bywaters (1958), in the
1956), the latter authors having encountered two first British contribution to the subject, described the
cases in their review of 200 patients with AS. The radiographic appearance in their patient with a
following year Romanus and Yden (1952) described spontaneous severe destructive lesion as a crush
the superficial, short-lived, erosion ofthe anterior and fracture.
antero-lateral surface of the vertebral rim which they The clinical presentation of certain destructive
called 'anterior spondylitis'. Their systematic radio- lesions late in the course of the disease was again
logical studies were prompted by their having emphasized by Wholey, Pugh, and. Bickel (1960),
encountered patients with AS and severe destructive whose ten patients had all had AS for at least 10
lesions of vertebral bodies; having discovered minor years before the lesions were discovered.
vertebral erosions close to discs (anterior spondylitis) During the last decade several radiological studies
in the early stages of AS, they speculated that severe -,(Seaman and Wells, 1961 ; Louyot, Gaucher, Mathieu,
destructive lesions might be exceptional extensions and Miquel, 1963; Coste, Delbarre, Cayla, Massias,
of the primary inflammatory phase of the disease. and Beasley, 1963; Streda, 1964; Jacqueline, 1965;
Since recent pathological studies (Ball, 1971) Serre, Simon, and Claustre, 1965; Schulitz, 1968,
indicate that anterior spondylitis, as defined above, 1969; Dihlmann, 1968; Rivelis and Freiberger, 1969;
is probably the precursor of the characteristic Daneo and Di Vittorio, 1970) have established that,
ossification of the annulus (Fig. 1), it seems important whereas destructive lesions invariably affect vertebral
Accepted for publication January 26, 1972
346 Annals of the Rheumatic Diseases

bodies close to a disc, the extent of involvement varies lesion may be seen late in the course of the disease,
considerably. Most investigators have recognized Type e lesions are rarely seen in unankylosed spines.
two main types of lesion, one localized and the other Published data on the pathology of destructive
involving more or less the whole of the disc-bone lesions is limited to seven reports of biopsies of ten
border. lesions. In none was there any histological evidence
These may be conveniently (and perhaps signi- of bacterial infection and bacterial cultures were
ficantly) classified into three main groups according negative in the six cases tested. The first report was
to whether the lesion involves the discal surface of the that of Wholey and others (1960), who described
vertebral rim (Types a and b), the cartilaginous part their findings in two patients: one showed 'changes
of the vertebral end-plate (Types c and d), or both consistent with an inflammatory reaction', the other
(Type e) (Fig. 2). 'findings consistent with a chronic inflammation'.
Radiological estimates of the prevalence of destruc- Coste and others (1963) presented a more detailed
tive lesions have varied between 1 and 28 per cent. account of the pathology in one of their cases. They
(Table I). Estimates of the prevalence and distribution found replacement of bone by fibrous tissue contain-
of lesions have probably been influenced by the ing new bone and haemosiderin deposits, and
method used, the extent of spinal survey undertaken, oedematous marrow containing perivascular collec-
lack of agreed diagnostic criteria, and the duration tions of lymphocytes and plasma cells. They con-
of the disease. Nevertheless, severe lesions (Type e) cluded that the lesions had an inflammatory com-
are probably uncommon and, while they may occur ponent but left the question of pathogenesis open.
at any level, the dorso-lumbar region (T10-SI) seems Serre and others (1965) briefly reported finding 'a
to be specially susceptible (Louyot and others, 1963; non-specific granulo-histiocytic infiltrate', and
Streda, 1964). Another point emerging from radio- Hackenbroch (1967) wrote of a 'non-specific inflam-
logical surveys is that, while all types of destructive mation of moderate or minimal intensity with
occasional round cell infiltration'. This evidence has
Table I Prevalence of destructive vertebral lesions apparently been accepted by some as supporting a
primary inflammatory origin for the destructive
Author Year Total cases Prevalence lesions, but the consideration of three more recent
per cent reports casts some doubt on this view.
Forestier 1951 200 1-0
(1) Hansen, Taylor, Honet, and Lewis (1967)
Coste and others 1963 171 4-7 described a patient with AS of 20 years' duration who
felt pain and a snapping sensation in the back after
Streda 1964 250 15-2 stepping off a kerb. The pain, though not severe,
Jacqueline* 1965 165 28-5 persisted, and 7 months later radiographs showed a
Type e lesion at the T12/L1 disc and a biopsy revealed
Serre and others 1965 150 10-0 'non-specific granulomatous inflammatory tissue'.
Hackenbroch 1967 255 2-7 A bone graft was inserted and there were no further
symptoms until 5 years later when, rising from bed,
Dihlmann 1968 549 2-7 he felt a 'pop' in his back followed by severe pain.
1968 50 18-0 The following day radiographs showed a fracture
Schulitz 1969 371 3-0 dislocation at T11/12. The question raised by this
case (though not mentioned by the authors) is
* We have excluded his Type I lesions which appear to represent an-
terior spondylitis.
whether the first destructive lesion (at T12/L1) was
t Tomography. the late result of a similar but less severe fracture.

( -."N

a b C d e
FIGi. 2 Types of destructive lesions: lateral view, anterior to right. .x cartilage plate I - osteosclerosis
 Destructive lesions of vertebral bodies in ankylosing spondylitis 347

CIo (A CA
Z I
w
+ + I t
CX
t CU Co CU C U

|)
u
CU CUo
t co co o , , U co
CU c-I K c-~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ :~x c

G
|ki
^u Z t R I t I ~~~~~~~~t
- I2

E|
Y ] FC | z zCi U) Ci Ci |) | 4IC ~|~z |F |> o|
CJCu

0
0 + 0 0 0 + + 0 0~~t 0 0 0 1

0WI n +n + + + 0 0~ 0ns
CU C~vC4
C c

C' C

-C U

Ci~~~~~~~~~~C

E-' _I tI I o I o I tI W co I
~
~~~ ~ ~ ~ n en 0
ON c- 00 en e 0)

o ~~~~ov CasW)W

~~ ~~ UA'ren v) U en +

0-4
%, Ca~~~~~~~~~~~~~~~~~~~~~~U
U C

- - - c CA-
c- c-
348 Annals of the Rheumatic Diseases

(2) Kanefield, Mullins, Freehafer, Furey, Horen- duration of AS at that time ranged from 4 to 45 years. All
stein, and Chamberlin (1969) described three patients extensive (Type e) lesions occurred in patients with a disease
with severe lesions in whom tissue obtained at duration of 15 years or more. Other types were seen in
biopsy was described respectively as 'non-specific both early and late stages.
granulation tissue', 'non-specific chronic inflamma- The occupations of our patients were of interest by
comparison with those of 100 unselected male patients
tion', and 'fibrous tissue ... with new-bone ... with AS. Table III shows that, at the time of the onset of
fibroblasts, and fibrinoid degeneration and macro- back pain (as distinct from the onset of AS), 67 per cent.
phages'. In two of these cases there was evidence in of patients with destructive lesions were employed in
radiographs taken at the time oj onset of localized heavy manual work as compared with 28 per cent. of the
back pain that the destructive changes seen later had unselected group.
originated in a non-dislocated fracture of an ankylosed
segment. The authors concluded that the 'mechanism Table III Occupational history
of granuloma formation might be traumatic'.
(3) In the report of Rivelis and Freiberger (1969), two Occupation Present series Unselected series
additional points emerged. In their Case 1, tissue
from the posterior spinal joint revealed reactive Manual 10 28
Light manual 5 37
fibroblastic tissue, bone resorption, and reactive bone Sedentary 0 35
deposition, and tissue from the corresponding disc
showed proliferative connective tissue, microscopic Total 15 100
sequestra, and reactive new-bone formation. These
findings were considered to represent a pseud-
arthrosis. Secondly, in their Cases 2 and 3, it was History
asserted that a localized destructive lesion in the In fourteen of the fifteen cases seen clinically we were
impressed by a characteristic change in symptoms which
anterior part of the disc had occurred in the only had occurred 2 to 12 months before referral. Usually,
unankylosed segment in the lumbar region. They backache in AS, as is well recognized, is diffuse in distribu-
concluded that the destructive lesions in their cases tion, tends to be increased by inactivity, and is relieved by
might be explained by excessive forces imposed on an mobilization. The patients in this study developed, often
isolated mobile segment in a rigid spine. quite suddenly, a more localized pain, usually in the area
The reports of Kanefield and others and of Rivelis of distribution from a single spinal segment. Secondly, the
and Freiberger seem to show that a destructive lesion pain was exacerbated by physical activity, especially
heavy work, and was relieved by rest. In only three of our
may be the late result of a stress fracture through an patients was there any evidence of concurrent exacerbation
ankylosed segment or may be produced by abnormal of disease activity. Four patients related the onset of their
stress on a mobile segment lying between ankylosed new symptoms to a specific traumatic incident of a minor
regions. nature, including a light blow on the lumbar region in one
Since there is a paucity of clinical information and case, and sitting up suddenly in bed in another. One patient
views about the pathogenesis of destructive lesions (Case 4) was referred to us because the pain, which had
are conflicting, it seemed worth while to review our been attributed to a relapse of AS, had failed to respond to
patients, paying particular attention to the clinical radiotherapy. Another (Case 13) had undergone full
and pathological aspects. Our clinical, radiological, urological investigation for severe loin pain, with negative
result, before referral.
and pathological findings are consistent with a
traumatic aetiology. Examination
In addition to the usual signs of AS, local tenderness on
Material and methods percussion or palpation at the level of the destructive
We have studied seventeen male patients with established lesion was observed in thirteen patients. In the two re-
AS, of whom fifteen presented to us with clinical and maining patients these signs were absent, and in the two
radiological features of destructive lesions. Ten have been post mortem cases they had not been recorded. In eleven
followed up for 1 to 3 years and one for 16 years. Tissue patients there was a localized visible or palpable spinal
has been obtained at operation from four of these 15 deformity or gibbus. In all patients movement of the
patients, and post mortem from two others whom we did affected segment reproduced the localized pain. In
not have the opportunity to examine clinically. Tissue patients with spinal ankylosis, this was most convincingly
from thirteen un-united fractures in patients without demonstrated by pressure on the adjacent part of the spine
AS has also been studied for comparison. with the patient prone. It is important to differentiate this
type of local pain from a more superficial tenderness
which is due to the characteristic ligamentous attachment
Clinical features lesions which may occur over the spinous processes.
The important data on all the patients studied are sum- There was no evidence that iritis, carditis, or other non-
marized in Table II (p. 347). The ages of our seventeen skeletal manifestations occurred more commonly in
cases, all male, ranged from 29 to 63 years at the time patients with destructive lesions than in those without.
radiographs disclosing destructive lesions were taken. The One patient (Case 12) had had pulmonary tuberculosis
 Destructive lesions of vertebral bodies in ankylosing spondylitis 349

one year before the onset of AS and had been successfully of the thoracic spine or in the lumbar spine
treated by chemotherapy and upper lobe resection. There (Table II).
was no evidence of recurrence of tuberculosis at the time
of the destructive lesion some 12 years later.
The erythrocyte sedimentation rate (E.S.R. Westergren) LOCALIZED LESIONS
ranged from 23 to 115 mm./lst hr when the destructive
lesion was diagnosed, but there was no evidence that this Thoracic spine
index of disease activity had increased with the develop- Lesions (Types a and b) were found predominantly
ment of the destructive lesion. anteriorly in kyphotic spines. They varied from
Nine patients (Cases 3, 4, 5, 6, 7, 10, 11, 14, 15) had thinning of the anterior third of a single disc with
previously had radiotherapy to the spine, including the irregularity and sclerosis of neighbouring bone (Figs
level at which the destructive lesion was found; of these, 3 and 4) to local defects involving all but the outer-
four had extensive lesions and five minor lesions. Only most part of the anterior vertebral rim on either side
one patient (Case 5) had received corticosteroids (in low of several discs (Figs 5 and 6, overleaf). In the single
intermittent dosage for 4 years) before the discovery of
the destructive lesion. case with more severe thoracic lesions (Case 16),
ossification of the same area of the discs was present
in neighbouring vertebrae (Fig. 5).
Radiography Lumbar spine
The various radiological changes in the vertebral Localized lesions (Types b, c, and d) at the disc-bone
bodies conform to those described in the literature border were seen in anterior, antero-lateral, postero-
(Fig. 2). In addition we have observed fractures of lateral, and posterior situations. Anterior or posterior
the neural arch in association with both localized and lesions were easily recognized in plain lateral radio-
extensive lesions. With the exception of Case 6, all graphs. The detection oflesions at other sites required
the lesions studied were situated in the lower half tomography or, in post mortem specimens, radio-
graphs of slab sections. The lesions (Types b, c, and d)
typically presented as a saucer- or cup-shaped defect
at the disc-bone border delineated by a zone of
vertebral density (Figs 7, 8, 9, overleaf). In lesions
involving the discal surface of the vertebral rim (Type
b), it was noted that the outermost part of the rim and
the anterior vertebral surface just below the rim edge
appeared to be intact. Occasionally a Type b lesion
was bordered externally by an incomplete syndesmo-
phyte. In one instance (Case 1), a Type b lesion was
associated with apparently ankylosed apophyseal
joints and a fracture of the neural arch (Fig. 9). It

*.:.. 4>;* *& *;.-

o~~~~~~~~~~~j
.; .-...
Wj*
....

~ *,*t t -

A;iBitt.%''

FIG. 3 Case 8. Type a lesion. X-ray shows opacity around

anterior part of narrowed T8/9 disc. Section (above) shows FIG. 4 Case 8. Detail of Fig. 3, showing reactive bone,
disc replaced byfibrous tissue and cartilaginous nodules fibrous tissue, and cartilage. Haematoxylin and eosin x 120
350 Annals of the Rheumatic Diseases

FIG. 5 Case 16. Type b lesion. Slab post mortem radio- FIG. 7 Case 13. Type b lesion. Tomogram, showing
graph, showing punched-out defect of anterior vertebral rim saucer-shaped defect (arrow) in lower anterior rim of T 1I
of T5/6 and 8/9. Corresponding area is ossified in T6/7, not apparent in plain radiograph
T718, and T9/10

FIG. 6 Case 16. Histological section of T5/6 disc shown in Fig. 5. All but the outer edge of the
rim and the intervening disc are replaced by vascular fibrous tissue. Lesion delineated by thick
bone trabeculae. No anterior spondylitis. Haematoxylin and eosin x 4
 Destructive lesions of vertebral bodies in ankylosing spondylitis 351

F I G. 8 Case 17. Type c

lesion L2/3 disc. Section
A: 2 ^mjshows discal hernia through
upper cartilage plate of L3,
surrounded by cancellous
bone enclosing oedematous
hypocellular marrow. De-
v fect in rim of L2 occupied by
avascular disc-like tissue.
Haematoxylin and eosin x 3
INSET
radiograph ofPost-mortem
L2/3 disc

was also evident from post mortem studies (Fig. 8)

that Type c lesions involving the cartilaginous part
of the end-plate may occur in an unankylosed lumbar
spine.
EXTENSIVE LESIONS
These (Type e) typically involved a single segment,
above and below which the spine showed advanced
ankylosis. The affected segment was, in some in-
stances, demonstrably mobile. Dislocation, lateral
or sagittal usually of a relatively minor degree, was
observed in some cases. Characteristically there was
irregular destruction of the central and peripheral
parts of the disc-bone border extending for a few
millimetres into the vertebral body and bordered by
a narrow zone of radio-opacity. Both the upper and
lower surfaces were commonly involved, although
usually one was more severely affected than the other.
In two of the five patients with this type of lesion there
was a fracture through the corresponding neural arch
(Figs 10 and 11, overleaf).

Pathology
LOCALIZED LESIONS
FIG. 9 Case 1. Type b lesion in upper anterior rim of L4. Type a (Case 8) A clinical radiograph and the histo-
Apophyseal joints ankylosed. There appears to be afracture logical appearances of the biopsy are shown in Figs 3 and
in the neural arch (arrow) 4. Two cancellous bone fragments (the bone above and
27
352 Annals of the Rheumatic Diseases

F IG. 1O Case 3. Type e lesion at L4/5 with gross destruc-

tion. Apoplhyseal joints ankylosed. Arrow points to fracture
in neural arch
FIG. 11 Case 4. Type e lesions at L2/3 with posteriol
displacement. Apophysealjoints ankylosed. Arrow points to
below the disc) are joined by fibrocartilaginous nodules fracture throlugh neural arch. Discs abore and below lesion
and fibrous tissue. In the latter (Fig. 4) new-bone deposi- ankylosed
tion can be seen. Scanty lymphocytes and plasma cells are
present in the oedematous marrow in the cancellous bone. evidence of anterior spondylitis or aseptic necrosis. The
The overall appearance resembles an incompletely healed central and posterior parts of the vertebral bodies are
fracture. porotic. This process has produced thinning and focal loss
of the subchondral bony end-plate and of the bony support
Type b (Case 16) Serial slab sections (Fig. 5) of this of the vertebral rim (Fig. 12a, opposite).
thoracic spine show destructive lesions predominantly
involving anterior vertebral rims at T5/6, 8/9, and 11/12. Type c (Case 17) The lumbar spine in this patient, as
The most severe lesion (T8/9) lies between ankylosed judged by radiographs and histological sections of discs
regions, T6-8 above and T9/ 10 below. The ankylosis of the and posterior spinal joints, was mobile and lacked
vertebral bodies in these regions is similar to that seen in syndesmophytes (Fig. 8). Radiographs of slab sections
senile kyphosis. Each destructive lesion is associated reveal destructive lesions (not easily recognized in radio-
anteriorly with small beak-like osteophytes. Histo- graphs of the whole specimen) in the left postero-lateral
logically (Fig. 6), all the anterior lesions show focal part of the L2/3 disc and the right lateral part of the
replacement of the disc and disc-bone border by vascular L5/S1 disc. Histologically the lesions are virtually iden-
fibrous tissue containing groups of haemosiderin-laden tical. At the L2/3 level (Fig. 8), there are two defects in
phagocytes and very scanty lymphocytes and plasma the cartilaginous plate of L3. Here avascular disc tissue
cells-changes resembling those described by Schmorl can be seen to stream into the vertebral body where it
and Junghanns (1959) in the early stages of senile kyphosis. merges with a narrow zone of reactive fibrocartilage
The narrow zone of radiodensity delimiting the lesions attached to the surrounding bone. At the edges of the
consists of thickened trabeculae of mainly lamellar bone, lesion there is some reactive bone; but the sclerotic zone
the marow spaces in which show recent haemorrhage. delimiting the lesion consists mainly of lamellar bone
The V-shaped translucency in the lesion at T8/9 is occupied trabeculae enclosing hypocellular oedematous marrow.
by cartilaginous and fibrous tissue, the latter containing Pathologically this type of bone sclerosis suggests per-
reactive new bone. The radiodensity occupying the V- sistent excessive pressure. Opposite to this lesion in L3
shaped area consists of trabeculae of mainly lamellar bone. there is a similar defect in the lower border of L2 at the
At the edges of the V-shaped area there are foci of junction of the cartilage plate and vertebral rim. The
crumpled bone-clear evidence of trauma. There is no lumbar vertebral bodies are generally porotic and as in
 Destructive lesions oJ vertebral bodies in ankylosing spondylitis 353

..:..

a 1.
......

b!

-
.f
pw-,., 't.
p .0-.r -*C,*- ;"." ".

4. .-16- -,* 4. w
?I*
I" a
10"'Y"tft- --
-,
ow., V,.
4. .11A*'.w
,,

:
FIG . 12 (a) Case 16. Osteoporotic thinning andfocal loss FIG. 13 Case 9. Type e lesion. Zone a Disc remnant;
Zone b Erosion at disc-bone border; Zone c Osteosclerosis.
of subchondral bone plate in ankylosed spine Haenatoxylin and eosin x 6
(b) Case 17 as for (a) in an unankylosed spine.
(c) Non-spondylitic control case of osteoporosis. Haem-
atoxylin and eosin x 95
VC..
Case 16 there is osteoporotic thinning and osteoporotic
focal loss of the bony end-plate (Fig. 12b). There is no
evidence of anterior spondylitis or aseptic necrosis.
EXTENSIVE LESIONS K. '.
Type e (Cases 3, 4, and 9) The biopsy specimen
included the disc-bone border and adjacent vertebral
bone. All are similar histologically and may be described
as a group. In none is there any evidence of bacterial C ,... A6\s
L_^%N~~~~~,i
infection, tuberculous or otherwise; nor is there evidence
of aseptic necrosis or radiation damage to bone. The lesion
(Fig. 13) can be divided into three merging zones, a-c,
moving from the disc towards the vertebral bone. Zone a
(Fig. 14) consists of disc tissue in which the most notable
change is the presence of capillary congeries and micro-
scopic foci of eosinophilic degeneration which, incidentally
are quite unlike the formations seen in a rheumatoid
nodule. There are no inflammatory cells in zone a.
Zone b (Figs 15, 16, 17) includes the disc-bone border.
Here the vertebral bony end-plate and discal cartilage are
invaded and largely or completely replaced by immature
connective tissue composed of proliferating capillaries
and fibroblasts (granulation tissue) in which inflammatory
cells are inconspicuous or absent. There is both recent and
FIG . 14 Case 9. Type e lesion. Zone a Capillaries invade
disc remnant showing eosinophilic degeneration (upper
right) Haematoxylin and eosin x 95
354 Annals of the Rheumatic Diseases

_! WR aQ~~~~'fX~:

FIG. 16 Case 3. Type e lesion. Zone b Disc-bone border

replaced by vascularfibrous tissue in which fibrocartilage has
FIG .15 Case9. Type e lesion. Zone b Remnant of formed (arrow). Perisinusoidal lymphocytes and plasma
original cartilaginous end-plate erodedby granulation cells in subjacent marrow. Haematoxylin and eosin x 95
tissue. Inflammatory cells very scanty. Reactive
fibrocartilage below centre
and upper right. Haematoxylin
and eosin x 95

FIG. 17 Case 3. Type e lesion. w

Zone b Irregular bony fragments 4x_
embedded in vascular fibrous tissue -;=
replacing vertebral end-plate. No in- :'W
flammatory cell infiltration. Haem-i;'/t,..,_
atoxylin and eosin x 95
 Destructive lesions of vertebral bodies in ankylosing spondylitis 355

old haemorrhage and evidence at weak attempts at UN-UNITED FRACTURES

callus formation in the form of scattered foci of reactive In eight of thirteen cases of un-united fractures in patients
fibrocartilage, cartilage, and rarely reactive (woven) bone without AS, small scanty aggregates of lymphocytes and
in the fibrous tissue. plasma cells were seen in oedematous marrow neighbour-
Zone c (Figs 18, 19) consists of a narrow region of ing the fracture site (Fig. 19).
intense bone remodelling with lamellar trabecular sclero-
sis. The marrow spaces in this zone are oedematous and Treatment
contain scattered perivascular collections of lymphocytes
and plasma cells. This zone merges gradually with verte- The therapeutic measures required to relieve symp-
bral bone which is normal apart from some oedema and toms in each case are indicated in Table II. Patients
hypocellularity of the marrow. with minor lesions and relatively mild symptoms
were advised to modify their occupation and avoid
flexion strains. Where this was impractical or un-
successful, a spinal support was fitted. This was of the
Milwaukee, Fisher, or modified Goldthwaite type.
In three patients, all with severe (Type e) lesions, this
was unsuccessful, and surgical treatment, involving
spinal interbody fusion by bone-graft, was carried
out. In one other patient (Case 8) surgical treatment
was undertaken at the patient's request to correct
thoracic kyphosis.
This therapeutic approach invariably resulted in
relief of pain. Follow-up, with one exception, has not
yet exceeded 3 years (Table II). Recurrence of
symptoms and progression of radiographic changes
have not been observed after effective immobiliza-
tion. In two patients (Cases 1 and 5), the follow-up
radiographs are suggestive of repair, although we
have no histological confirmation of this. Radio-
graphs of the patient followed up for 16 years (Case 2)
showed unequivocal healing with massive osteophytes
and bony sclerosis.
Discussion
In the present study patients with AS who presented
with a history of localized back pain of relatively
recent onset, made worse by movement and relieved
F G. 18 Case 4 Type e ksion Zone c Bone trabeculae by rest, have invariably been found to have a destruc-
showing appositional sclerosis (upper left) and active tive lesion of a vertebral body corresponding to the
osteoclastic resorption (arrow). Focal lymphocyte and
plasIa cell aggregates in oedematois marrow (centre right). painful region.
Although the ESR was often raised, as
Haematoxylin and eosin x 95
expected in AS, there was no evidence of a general

* V',V,. _
I,
I
*0
FIG. 19 Lymphocytic and plasma cell in-
filtration in oedematous marrow
(a) Destructive lesion (Case 4)
(b) Nonunited fracture of first phalanx
of great toe after 12 months in plaster.
Haematoxylin and eosin x 235
Isolation of synovialfluid cells
356 Annals of the Rheumatic Diseases

increase in disease activity concomitant with the lesion in the bone just beneath the subchondral plate.
change in symptoms. In contrast to patients without We think this unlikely, because studies of unanky-
destructive lesions, all our patients were physically losed lumbar spines in AS have shown no evidence of
active at the time of onset of symptoms, and in some focal osteitis (Ball, 1971).
the onset clearly dated from a minor traumatic Discal herniation (in the sense of Schmorl and
episode. The onset of pain was usually abrupt and Junghanns, 1959) is restricted to the cartilaginous part
there was invariably a delay of some months before of the vertebral end-plate. When herniation is situated
the patient was seen in hospital. At this time local at the periphery of the plate, the inner part of the rim
tenderness and pain could usually be elicited by may be involved, as was evident in one of the lesions
direct pressure over the site of the lesion where there studied by us. But discal herniation cannot account
was sometimes a gibbus. These clinical features can for Type b lesions which produce cup-shaped defects
be found in whole or in part in the case reports of in the discal surface of the vertebral rim where the
Edstrom (1940), Jacqueline (1956), Coste and others avascular part of the annulus is attached. The Type b
(1963), Hansen and others (1967), Hicklin (1968), and lesions in the anterior vertebral rims of the kyphotic
Kanefield and others (1969). They constitute a thoracic spine of one of our cases were histologically
syndrome sufficiently different from that of a spondyli- similar to those described by Schmorl and Junghanns
tic relapse to suggest that some other pathogenic (1959) in elderly non-spondylitic kyphotic individuals
factor is involved. and attributed by them to thecompressive forces gener-
Although destructive lesions may be mistaken ated by the kyphosis. Progressive kyphosis, according
clinically for pyogenic or tuberculous osteomyelitis to Schmorl and Junghanns, leads to collapse of the
(Forestier and others, 1951; Romanus and Yden, vertebral rim and invasion and replacement of the
1952; Pfluger, 1959), the problem of their patho- disc by vascular fibrous tissue which ultimately
genesis centres on whether they are essentially part ossifies. Since thoracic kyphosis is a characteristic
of the inflammatory process of AS or simply due to feature of AS, it is hardly surprising that lesions
trauma. similar to those of senile kyphosis are encountered
in AS.
LOCALIZED LESIONS Type b lesions may, however, also involve the
These may be seen in both ankylosed and unanklyosed anterior vertebral rim in the lumbar spine. No
spines (Dihlmann, 1968). As might be expected, histological data are available on these lesions in this
clinical symptoms and signs were in general less pro- situation, but the assumption that they are primarily
nounced than in patients with Type e lesions. It is inflammatory can be questioned on two counts.
therefore possible that they may be overlooked, Previous histological studies (Ball, 1971) indicate that
especially as some localized lesions can be detected the primary inflammatory process in the vertebral
only by tomography. bodies in AS specifically involves the attachments of
Coste and others (1963) and Streda, (1964) have the vascularized outer annulus to the anterior or
noted the radiological resemblance of some of these antero-lateral surface of the vertebral rim (anterior
lesions to one or other form of discal herniation. spondylitis). Radiologically this area seems to remain
Histological studies of localized lesions have not been intact in Type b lesions, which is not what one would
reported. In one of our patients a localized lesion in an expect if these were unusual extensions of anterior
unankylosed lumbar spine was found by histological spondylitis. Secondly, if Type b lesions are not severe
examination to be a discal hernia. Since, according to forms of anterior spondylitis, it is difficult to see how
Schmorl and Junghanns (1959), discal herniation is they could be primarily inflammatory, as the part of
common in non-spondylitic spines studied post the annulus involved is avascular and, as noted above,
mortem, it must be considered that a discal hernia we find no histological evidence for a primary
may be a coincidental lesion. If, however, discal osteitis in AS. An alternative possibility is that a
herniation is commoner in AS, and this has not at Type b lesion is a traumatic collapse or avulsion of a
present been established, then there must be a pre- rim weakened by the osteoporosis, which we have
disposing cause. We would suggest that this is found in unankylosed lumbar spines.
osteoporosis which, in the case referred to above,
had produced focal bone loss of the subchondral EXTENSIVE LESIONS
plate, similar to that which we have found in non- In contrast to localized lesions, Type e lesions appear
spondylitic forms of osteoporosis. Since the lumbar to occur almost exclusively in spines with advanced
spine in question was not ankylosed by bone, it is ankylosis. Our histological findings in this type of
possible that in AS the lumbar spine (and possibly destructive lesion are similar to those reported by
other zones) is susceptible to osteoporosis from the Coste and others (1963) and Rivelis and Freiberger
early stages of the disease. It could, of course, be (1969), the most detailed reports available. These
argued that the discal herniation observed by us was authors, however, differed in their interpretation of
the late result of a small destructive inflammatory the histology, the former drawing attention to the
 Destructive lesions of vertebral bodies in ankylosing spondylitis 357

inflammatory cell component, whereas the latter immediately after a traumatic event describe fractures
interpreted the pathology as a pseudarthrosis. We with or without dislocation and physicians seeing the
agree with Coste and others (1963) and with Hacken- patients only after an interval of a few months describe
broch (1967) that the inflammatory cell infiltration and debate destructive lesions.
is of a minor order; but, since we find similar if less
prominent infiltration in un-united fractures in non- TREATMENT
spondylitics, the possibility cannot be excluded on The management of these lesions has received little
pathological grounds that the inflammatory cell attention in the literature and the comments that have
infiltration in destructive lesions is merely secondary been made are based more on the individual author's
to persistent and uncontrolled traumatic tissue views of the pathogenesis than on reported follow-up
damage at the disc-bone border. The associated bone of cases. Louyot and others (1963) mentioned im-
destruction, the presence of minimal amounts of provement after radiotherapy but there is little
callus, evidence of old haemorrhage, and neighbour- support for this from other sources. It is of interest
ing bone sclerosis may indeed all be encountered in that one of our patients was referred because the pain
un-united fractures. Thus the histological evidence was not relieved by radiotherapy. We have treated
on which a primary inflammatory aetiology is our patients by immobilization on the assumption
entirely based is questionable. that destructive lesions are essentially traumatic.
The most important single piece of evidence sup- This has been successful without exception, rapid
porting a traumatic aetiology is provided by the symptomatic relief being obtained in all cases. The
sequential radiographs of Kanefield and others (1969) beneficial effect of this policy of immobilization is of
which indicate that a destructive lesion may be the interest in view of the known deleterious effect of
late result of an uncomplicated undisplaced fracture immobility on the symptoms of active spondylitis.
through an ankylosed segment. None of our patients The degree of immobilization required depended on
was seen at the time of onset of the new symptoms so the extent and instability of the lesion and the mode
that we cannot confirm or refute this observation. of life of the individual: in some a change of occupa-
However, in two of our five patients with severe tion was sufficient, in others some form of a spinal
lesions, there were discontinuities in the correspond- support was required, and in the most severe cases
ing neural arch. These fractures could conceivably surgical fusion was necessary. In our limited follow
have been secondary to instability induced by a pre- up we have not so far observed recurrence of
existing inflammatory disc lesion. On the other hand, pain, and in one patient not treated surgically and
Schmorl and Junghanns (1959) stated that fractures with a follow up of 16 years healing with massive
of the neural arch were invariably accompanied by osteophyte formation has occurred.
damage to the disc which might not become apparent
radiologically until some time after the initial trauma. Summary
Since our patients with extensive lesions invariably
presented some months after the onset, it is possible (1) Seventeen patients with AS and various types of
that the fractures of the neural arch and the initiation destructive lesions of the vertebral bodies have been
of disc pathology were contemporaneous. studied.
Severe lesions seem to occur mainly, if not ex- (2) In the fifteen patients studied clinically, the
clusively, in patients with advanced spinal disease. lesion was characterized by a change in symptoms,
From observations on two spines without destructive the patient complaining of localized back pain
lesions (before and after maceration), we can say that exacerbated by movement and of some months'
in AS the extent of bony bridging in ankylosed regions duration. All were manual workers.
may be variable-a feature noted by Aufdermaur
(1957), and secondly that an individual segment may (3) When first seen the lesions presented radio-
remain mobile when blocks of vertebrae above and logically as either (a) a localized defect in the vertebral
below are ankylosed. Such mobile segments will be rim or cartilaginous end-plate, or (b) destruction of
subject to abnormal stress. It would appear, therefore, most of the disc-bone border. In one of the ten
that, during the natural evolution of spinal ankylosis, localized lesions and two of the five severe lesions,
a situation may arise in which isolated segments, there was a fracture of the corresponding neural arch.
mobile or ankylosed, are especially susceptible to Severe lesions were seen only in spines with ankylosis
trauma. If severe destructive lesions are simply the above and below the affected segment. Tomography
late result of spinal trauma, then radiographs taken was necessary to visualize some localized lesions.
at the time of onset ofsymptoms should show evidence (4) Lesions were studied histologically in six patients.
of isolated segmental instability with or without a Localized lesions in the cartilaginous end-plate in the
fracture and no evidence of a destructive lesion. It is, lumbar region proved to be discal herniae. Rim
indeed, a striking feature of the literature on the late lesions in the thoracic region resembled the early
complications of AS that surgeons seeing the patient stages of senile kyphosis. Focal osteoporotic defects of
358 Annals ofthe Rheumatic Diseases

the subchondral plate were observed in ankylosed aetiology could not be excluded histologically.
and unankylosed spines and were considered to
predispose the spine in Cases of AS to localized (5) Measures calculated to reduce spinal mobility,
lesions. Severe lesions had many of the histological including surgical fixation, were invariably followed
features of a pseudarthrosis and a simple traumatic by rapid symptomatic relief.
References
ANDERSSON, O. (1937) Nord. Med. 14, 2000 (Rontgenbilden vid spondylarthritis ankylopoetica)
AUFDERMAUR, M. (1957) 'The Morbid Anatomy of Ankylosing Spondylitis.' Documenta Rheumatologica Geigy,
No. 2
BALL, J. (1971) Ann. rheum. Dis., 30, 213 (Enthesopathy of rheumatoid and ankylosing spondylitis)
COSTE, F., DELBARRE, F., CAYLA, J., MASSIAS, P., AND BEASLAY, E. (1963) Presse med., 71, 1013 (Spondylites
destructives dans la spondylarthrite ankylosante)
DANEO, V., AND Di VrrrORIO, S. (1970) Rev. Rhum., 37, 155 (La spondylodiscite de la spondylarthrite ankylosante)
DIHLMANN, W. (1968) 'Spondylitis Ankylopoietica-die Bechterewsche Krankheit.' Thieme, Stuttgart
EDSTROM, G. (1940) Acta med. scand., 104, 396 (Is spondylarthritis ankylopoietica an independent disease or a
rheumatic syndrome?)
FORESTIER, J., JACQUELINE, F., AND ROTiES-QUtROL, J. (1956) 'Ankylosing Spondylitis-Clinical Observations,
Roentgenology, Pathological Anatomy, and Treatment,' trans. A. U. Desjardins. Thomas, Springfield, Ill.
GUEST, C. M., AND JACOBSON, H. G. (1951) Amer. J. Roentgenol., 65, 760 (Pelvic and extrapelvic osteopathy in
rheumatoid spondylitis. A clinical and roentgenographic study of ninety cases)
HACKENBROCH, M. H. (1967) Z. Orthop., 103, 23 (Umschriebene osteolytische Prozesse bei Spondylarthritis
ankylopoetica)
HANSEN, S. T., TAYLOR, T. K. F., HONET, J. C., AND LEWIS, F. R. (1967)J. Trauma, 7, 827 (Fracture-dislocations of
the ankylosed thoracic spine in rheumatoid spondylitis. Ankylosing spondylitis, Marie-Strumpell disease)
HICKLIN, J. A. (1968) Ann. phys. Med., 9, 206 (Erosive vertebral disease in ankylosing spondylitis)
JACQUELINE, F. (1956) J. Radiol. Electrol., 37, 887 (Troubles de la structure osseuse et lesions destructives au
cours de la spondylarthrite ankylosante)
(1965) Rhumatologie, 17, 223 (Destructions du rachis anterieur lombo-dorsal au cours de la
spondylarthrite ankylosante)
KANEFIELD, D. G., MULLINS, B. P., FREEHAFER, A. A., FUREY, J. G., HORENSTEIN, S., AND CHAMBERLIN, W. B.
(1969) J. Bone Jt Surg., 51A, 1369 (Destructive lesions of the spine in rheumatoid ankylosing spondylitis)
LOUYOT, P., GAUCHER, A., MATHIEU, J., AND MIQUEL, G. (1963) Rev. Rhum., 30, 263 (La spondylodiscite de la
spondylarthrite ankylosante)
PFLUJGER, W. (1959) Tuberkulosearzt, 13, 414 (Spondylarthritis ankylopoietica und Spondylitis tuberculosa, eine
seltene Kombination)
RIVELIS, M., AND FREIBERGER, R. H. (1969) Radiology, 93, 251 (Vertebral destruction at unfused segments in
late ankylosing spondylitis)
ROMANUS, R., AND YDEN, S. (1952) Acta orthop. scand., 22, 88 (Destructive and ossifying spondylitic changes in
rheumatoid ankylosing spondylitis (Pelvo-spondylitis ossificans))
SCHMORL, G., AND JUNGHANNS, H. (1959) "The Human spine in Health and Disease", trans. S. P. Wilk and L. S.
Goin from 'Die gesunde und die kranke Wirbelsaule in Rontgenbild und Klinik', 4th ed. (1957). Grune and
Stratton, New York and London
SCHULITZ, K.-P. (1968) Arch. Orthop. Unfall-Chir., 64, 116 (Destruktive Veranderungen an Wirbelkorpern bei
der Spondylarthritis ankylopoietica)
(1969) Med. Klin., 64, 593 (Spondylodiszitis bei Morbus Bechterew)
SEAMAN, W. B., AND WELLS, J. (1961) Amer. J. Roentgenol., 86, 241 (Destructive lesions of the vertebral bodies
in rheumatoid disease)
SERRE, H., SIMON, L., AND CLAUSTRE, J. (1965) J. Radiol. Electrol., 46, 538 (Les lesions destructrices des disques
et des corps vertebraux au cours de la spondylarthrite ankylosante)
STREDA, A. (1964) Radiol. Diagn. (Berl.), 5, 43 (Inflammatory destructive changes in the spinal column in
ankylosing spondylarthritis)
WHOLEY, M. H., PUGH, D. G., AND BICKEL, W. H. (1960) Radiology, 74, 54 (Localized destructive lesions in
rheumatoid spondylitis)
WILKINSON, M., AND BYWATERS, E. G. L. (1958) Ann. rheum. Dis., 17, 209 (Clinical features and course of
ankylosing spondylitis)