Content uploaded by John Richard Kirwan
Author content
All content in this area was uploaded by John Richard Kirwan on Sep 10, 2015
Content may be subject to copyright.
The relationship between soft tissue swelling,
joint space narrowing and erosive damage
in hand X-rays of patients with rheumatoid
arthritis
J. Kirwan, M. Byron and I. Watt
1
Rheumatology Unit, University of Bristol Division of Medicine and
1
Department
of Clinical Radiology, United Bristol Healthcare NHS Trust, Bristol Royal
In®rmary, Bristol, UK
Abstract
Objectives. To test the hypotheses that the progression of joint space narrowing behaves
differently from the progression of erosions and that clinically and radiologically assessed
soft tissue swelling relates more to diffuse cartilage loss than to erosive damage.
Methods. Radiographs and clinical data were obtained from 28 patients in a prospective,
multicentre, randomized, placebo-controlled trial of prednisolone 7.5 mg daily over 2 yr.
Radiographic scoring included the Larsen score, joint space narrowing and soft tissue swelling.
Clinical joint in¯ammation in the hands was assessed every 3 months and cumulative synovitis
score over the period of study was then calculated for each joint. The placebo-treated patients
and the prednisolone-treated patients were analysed separately. The Larsen scores were
compared after log transformation wtransformed score = log
10
(original score + 1)x. Changes in
Larsen scores and joint space narrowing scores were compared with the cumulative presence
of clinical synovitis and radiological soft tissue swelling using the correlation coef®cient.
Results. There was a difference in the rate of progression in the Larsen score between
placebo- and prednisolone-treated patients, but there was no signi®cant difference in the rate
of joint space loss. In placebo-treated patients, measures of synovitis correlated more strongly
with progression of joint space narrowing than with changes in the Larsen score. In
prednisolone-treated patients there was no correlation between clinical synovitis and change
in Larsen score (r = 0.029) and only a slight and non-signi®cant correlation with joint space
narrowing (r = 0.127). Radiographic evidence of soft tissue swelling remained correlated with
joint space narrowing (r = 0.279, P < 0.001) but was not correlated with change in Larsen score
(r = 2 0.113, P < 0.001 for difference between correlations). The correlation between Larsen
score progression and joint space narrowing seen in the non-treated patients was completely
abolished in the glucocorticoid-treated group (r = 2 0.003).
Conclusions. The progression of joint space narrowing behaves differently from the
progression of erosions. Prednisolone slows (or even stops) the progression of erosions (as
assessed by the Larsen score) while making no difference to the progression of cartilage loss
(as assessed by joint space narrowing). The results also suggest that synovitis, whether measured
clinically or radiologically, is more closely related to diffuse cartilage loss than to erosion
progression. Any link between synovitis and erosions is abolished by glucocorticoid therapy
while the link between synovitis and cartilage loss is not, pointing to at least two different
mechanisms for these observed radiological features.
K
EY WORDS: Synovitis, Diffuse cartilage loss, Erosions, Glucocorticoids, Pathology.
There is a broad correlation between the clinical and
radiological severities of rheumatoid arthritis (RA) w1x.
However, evidence is accumulating to suggest that dif-
ferent pathological processes are involved in the clinical
manifestations of synovitis and the progressive erosive
radiological damage seen in RA w2±7x. Recent clinical
w2±5x studies describe disassociation between clinical
synovitis, the serum acute-phase response and radio-
logical progression, whilst others w8, 9x do report
Correspondence to: J. Kirwan, University of Bristol Division of
Medicine, Bristol Royal In®rmary, Bristol BS2 8HW, UK.
Submitted 12 November 1999; revised version accepted 2 October
2000.
Rheumatology 2001;40:297±301
ß 2001 British Society for Rheumatology297
correlation between joint swelling and radiological
damage. The latter studies examined joints in groups,
a process which in¯ates apparent relationships. In a
study that analysed data in relation to individual joints
w10x, the correlation between persistent synovitis and
erosive progression was weak (r = 0.248, explained vari-
ance 6%). Furthermore, imaging studies have identi®ed
different types of synovial pathology within the same
joint w11x, and histological studies have shown that
erosions are more likely to be associated with in®ltration
by macrophages than lymphocytes w6x. One explanation
for these observations is that the clinical signs and
symptoms of in¯ammation are caused by synovial patho-
logical processes different from those that cause the
erosive joint damage seen on radiographs. However,
persistent, sustained intra-articular in¯ammation may be
responsible for the diffuse cartilage loss w10, 12x manifest
as radiographic joint space narrowing. We therefore
hypothesized that (i) the progression of joint space nar-
rowing might behave differently from the progression of
erosions, and (ii) that clinically and radiologically asses-
sed soft tissue swelling will relate more to diffuse car-
tilage loss than to erosive damage. The aim of this study
was to use hand radiographs taken from a randomized
controlled trial of prednisolone in RA to test these
hypotheses.
Methods
Radiographs and clinical data were obtained from a pro-
spective, multicentre, randomized, placebo-controlled
trial of prednisolone 7.5 mg daily that has been fully
reported elsewhere w2x. That study included 128 patients
aged 18±69 yr with RA of < 2 yr duration and currently
active disease. The 28 patients entered through one
centre who had the required hand radiographs available
were included in the present study. Clinical and radio-
graphic data were available at baseline, 1 yr and 2 yr of
follow-up for the following joints: four proximal inter-
phalangeal (PIP) joints, ®ve metacarpophalangeal (MCP)
joints and the thumb interphalangeal joint.
Radiographs
Radiographic scoringwasundertaken by twoindependent
methods. All available radiographs were viewed jointly by
the same experienced radiologist (IW) and rheumatologist
(JRK), using the same viewing conditions w2x. To ensure
similar conditions for assessing the radiographs from any
one patient, and to avoid the possibility of bias that might
develop over the several sittings required to read and
score the radiographs, their presentation was in blocks
of 30. Each block included 0, 1 and 2 yr ®lms in random
order from 10 randomly selected patients. All identifying
markings were covered. Each joint was scored by the
method of Larsen w13x, which grades the degree of joint
damage on a scale from 0 (radiologically normal joint) to
5 (maximum degree of joint destruction) with reference to
a standard atlas of radiographs. Assessments were made
by consensus between the observers and were recorded
on coding sheets. The change in Larsen score between
baseline and 2 yr was calculated for each joint.
As a separate exercise undertaken with radiographs
marked and randomized in a similar manner, two experi-
enced observers (IW and MB) scored each joint of both
hands for soft tissue swelling (none = 0, minimal = 1,
de®nite = 2, marked = 3, not scorable = X), following a
prede®ned sequence. This was then repeated for each
joint in the same sequence, identifying joint space nar-
rowing (none = 0, minimal = 1, de®nite = 2, no remain-
ing joint space = 3, fused = 4, not scorable = X). The
cumulative soft tissue swelling score over the period of
study (i.e. sum total of the score for the three X-rays) was
then calculated for each joint. The change in joint space
narrowing was also calculated for each joint.
Clinical data
Joint in¯ammation was assessed every 3 months by the
method of Thompson et al. w14x , in which joints are
recorded as being in¯amed if both soft tissue swelling
and tenderness are present simultaneously. This method
has been validated within and between observers w14x,
correlates well with the acute phase response w14x,and
has been used in other multicentre studies w15x. After
publication of the original study report w2x, and as
a separate exercise, the data relating to the hands (for
the PIP joints, the MCP joints and the thumb inter-
phalangeal joint) were taken from the original study
record forms and entered into a new database. Joints
were recorded as either showing synovitis (score = 1) or
not doing so (score = 0) on each occasion of examina-
tion. The cumulative synovitis score over the period of
study (i.e. the sum total of the scores for the nine visits)
was then calculated for each joint.
Comparisons
The placebo-treated patients and the prednisolone-
treated patients were analysed separately. The Larsen
scores were compared after log transformation wtrans-
formed score = log
10
(original score + 1)x. The sum of
the joint space narrowing scores for all joints taken
together was used for the total joint space narrowing
score for each patient on each occasion. Means and 95%
con®dence intervals (CI) were calculated and the groups
compared using the unpaired t-test. Changes in Larsen
scores and joint space narrowing scores were also
compared with the cumulative presence of clinical
synovitis and radiological soft tissue swelling using the
correlation coef®cient.
Results
Twenty-®ve of the 28 patients had full clinical and
radiological data for the analysis undertaken here.
Eleven had received prednisolone and 14 had received
placebo. There were no signi®cant differences in baseline
data between the groups with regard to articular index
(treated group mean 201 versus untreated group mean
204, P = 0.942), plasma viscosity (1.85 versus 1.79,
P = 0.405), HAQ (health assessment questionnaire)
298 J. Kirwan et al.
scores (1.21 versus 1.40, P = 0.466), log Larsen score
(0.39 versus 0.42, P = 0.896), log joint space narrowing
(0.32 versus 0.41, P = 0.551) and percentage of patients
who were non-erosive (73.1 versus 73.3).
The cumulative clinical synovitis and radiological soft
tissue swelling scores for both groups are shown in
Table 1, together with the mean joint space narrowing
and Larsen scores for years 0 and 2. Although patients
treated with prednisolone had a lower cumulative clinical
synovitis score (15.8 versus 23.3, P = 0.045), the more
striking signi®cant difference between the patient groups
was in the progression of Larsen scores (Fig. 1). There
was a clear difference in the rate of progression in Larsen
scores between the prednisolone- and placebo-treated
groups (P = 0.896 and 0.037 for 0 and 2 yr respectively),
in agreement with the full study report w2x. (In this subset
of patients there was a small but statistically non-
signi®cant reduction in the Larsen score for those treated
with prednisolone). Joint space narrowing scores are
shown in Fig. 2. Joint space was lost in both treatment
groups, with no signi®cant differences between them
(P = 0.466 and 0.254 for 0 and 2 yr respectively).
The correlation coef®cients comparing cumulative
scores for radiographic soft tissue swelling (radiographic
synovitis), clinical soft tissue swelling (clinical synovitis),
radiographic joint space narrowing (cartilage loss) and
Larsen score (erosion progression) are shown in Table 2
for patients who did not receive glucocorticoid treatment
and in Table 3 for those who did. For placebo-treated
patients (Table 2), clinical and radiographic assessments
of soft tissue swelling were signi®cantly correlated
(r = 0.472, P < 0.001). Clinical synovitis correlated with
change in Larsen score (r = 0.281, P < 0.01) but more
strongly with changes in joint space narrowing
(r = 0.348, P < 0.001). Radiographic soft tissue swelling
did not correlate with change in Larsen score but was
weakly related to change in joint space narrowing
(r = 0.192, P < 0.01). In these patients, progression of
Larsen score and change in joint space narrowing were
correlated (r = 0.322, P < 0.001), but less so than the
correlation between the two (clinical and radiographic)
measures of soft tissue swelling (r = 0.472, P = 0.066 for
the difference between correlations).
In patients treated with glucocorticoids (Table 3),
cumulative clinical and radiographic measures of soft
tissue swelling remained correlated with each other
(r = 0.427, P < 0.001). There was no correlation between
clinical synovitis and change in Larsen score (r = 0.029)
and only a slight and non-signi®cant correlation with
joint space narrowing (r = 0.127). Radiographic evid-
ence of soft tissue swelling remained correlated with
joint space narrowing (r = 0.279, P < 0.001) but was
not correlated with change in Larsen score (r = 2 0.113;
P < 0.001 for the difference between correlations). The
correlation between Larsen score progression and
joint space narrowing seen in the non-treated patients
was completely abolished in the glucocorticoid-treated
group (r = 2 0.003).
TABLE 1. Mean (S.D.) clinical and radiographic scores for placebo- and prednisolone-treated patients
Placebo Prednisolone P (t-test)
Number of patients 14 11
Cumulative clinical synovitis score 23.3 (12.0) 15.8 (10.2) 0.045
Cumulative radiographic soft tissue swelling score 17.6 (12.3) 11.2 (7.3) 0.326
Mean joint space narrowing, year 0 3.27 (5.62) 2.00 (2.71) 0.466
Mean joint space narrowing, year 2 7.67 (9.23) 4.31 (5.09) 0.254
Mean of Larsen scores after log transformation, year 0 1.61 (2.91) 1.46 (1.69) 0.896
Mean of Larsen scores after log transformation, year 2 4.22 (3.11) 1.16 (2.13) 0.037
FIG. 1. Mean proportionate change in Larsen scores. Vertical
bars show 95% CI.
FIG. 2. Mean proportionate change in joint space narrowing.
Vertical bars show 95% CI.
299Synovitis and progressive erosions in RA
Because some of the data had mildly skewed dis-
tributions, log transformation wnew value = log
10
(old
value + 1)x was applied to all the data and the analysis
was repeated. Minor differences were noted in the cor-
relation coef®cients, all adding greater support to the
conclusions.
Discussion
This study examined the relationship between clinical
and radiographic measures of synovitis and the two
main destructive lesions of RA, erosions and cartilage
loss. The results are based on clinical and radiological
follow-up in a randomized, controlled trial. They sup-
port the hypothesis that the progression of joint space
narrowing might behave differently from the progres-
sion of erosions. This study shows that prednisolone
slows (or even stops) the progression of erosions (as
assessed by Larsen scores) while making no difference
to the progression of cartilage loss (as assessed by joint
space narrowing). The results also suggest that synovitis,
whether measured clinically or radiologically, is more
closely related to diffuse cartilage loss than to erosion
progression. Any link between synovitis and erosions is
abolished by glucocorticoid therapy while the link
between synovitis and cartilage loss is not, pointing to
at least two different mechanisms for these observed
radiological features.
Several precautions were taken to avoid systematic
bias in these results. Firstly, clinical synovitis was
assessed as part of a double-blind, randomized, con-
trolled clinical trial. It was not linked in any way to
the later radiological assessment of soft tissue swelling.
Because the clinical scoring system has been validated
as a whole w14x, it may be less reliable when applied to
individual joints. Therefore only the presence or absence
of clinical synovitis (rather than any system of grading
or weighting) was used in this study. Secondly, the
progression of erosions, as assessed by the Larsen score,
was performed with radiographs masked, viewed and
scored in random order, and assessed at a time when the
hypothesis being tested here had not been formulated.
Thirdly, the assessment of joint space narrowing was
similarly masked and randomized, and was performed
over 2 yr later than the Larsen score assessments and the
clinical evaluations.
It may be argued that a bias could have arisen in
the results because radiological assessment of synovial
swelling and joint space narrowing were undertaken at
the same viewing session. However, radiographs were
assessed in random order and it was the cumulative
occurrence of soft tissue swelling over each patient's
sequence of radiographs that was compared with the
change in joint space narrowing over the 2 yr of obser-
vation. It is thus unlikely that this method of reading
radiographs caused any systematic effect.
The observations from this limited study are clearly
consistent with the hypotheses that were tested, but
similar ®ndings in other groups of patients should be
sought before these hypotheses can be considered
proven. Some published studies of the histology w5, 6x
and magnetic resonance imaging w11x of rheumatoid
synovium suggest that there may be areas of different
pathology and that these may relate differently to the
progression of erosions.
Our current understanding of the pathology of joint
in¯ammation and damage centres on the role of T
lymphocytes, macrophages and synoviocytes. T cells
appear to be central to the initiation of early disease
processes, and to the potentiation of in¯ammation, as
recently reviewed by Weyland w16x. The production of
high synovial ¯uid concentrations of cytokines and
matrix metalloproteinases could cause generalized car-
tilage loss. However, as proposed by Gay and colleagues
in 1993 w17x, synovial hyperplasia and macrophage
in®ltration may offer a separate and localized patholo-
gical mechanism for cartilage and bone destruction.
How invasive, ®broblast-like synoviocytes might cause
TABLE 2. Correlation between clinical and radiographic features in placebo-treated patients
Cumulative radiographic
soft tissue swelling score Change in Larsen score
Change in joint
space narrowing
Cumulative clinical synovitis score 0.472*** 0.281*** 0.342***
Cumulative radiographic soft tissue swelling score 0.140 0.192**
Change in Larsen score 0.322***
**P<0.01, ***P<0.001.
TABLE 3. Correlation between clinical and radiographic features in prednisolone-treated patients
Cumulative radiographic
soft tissue swelling score Change in Larsen score
Change in joint
space narrowing
Cumulative clinical synovitis score 0.427*** 0.029 0.127
Cumulative radiographic soft tissue swelling score 20.113 0.279***
Change in Larsen score 20.003
***P<0.001.
300 J. Kirwan et al.
joint damage was reviewed by Firestein w18x. Animal
models of joint tissue destruction can proceed without
concurrent T-cell involvement. Such localized destruct-
ive lesions may represent the clinical occurrence of
erosions. Thus, plausible cellular mechanisms have been
elucidated which might ®t with clinical observations.
In particular, the ®nding that differences in synovial
histology in biopsies from knee joints correlate with
overall radiographic progression in patients is relevant
w5, 6x. An abundance of T cells was related to little
progression of erosions, while an abundance of macro-
phages occurred in patients who subsequently had
erosive progression w5, 6x. It is therefore feasible that
these different pathological mechanisms of in¯ammation
and localized tissue destruction might relate separately
to the clinical signs of joint in¯ammation and erosion
development, and that glucocorticoid treatment could
affect them differently.
Two further studies of prednisolone in RA, as yet
published only in abstract form, support a differential
effect of glucocorticoids on conventional radiographic
features w19±21x. Improved understanding of the role
of new imaging techniques such as magnetic resonance
imaging will provide substantially greater sensitivity in
future investigations of disease mechanisms in patients.
References
1. Scott DL, Spector TD, Pullar T, McConkey B. What
should we hope to achieve when treating rheumatoid
arthritis? Ann Rheum Dis 1989;48:256±61.
2. Van Leeuwen MA, van der Heijde DM, Rijswijk MH et al.
Inter-relationships of outcome measures and process
variables in early rheumatoid arthritisÐa comparison of
radiological damage, physical disability, joint counts and
acute phase reactants. J Rheumatol 1994;21:425±9.
3. Kirwan JR and the ARC Low Dose Glucocorticoid Study
Group. The effect of glucocorticoids on joint destruction
in rheumatoid arthritis. N Engl J Med 1995;333:142±6.
4. Hickling P, Jacoby RK, Kirwan JR and The Arthritis and
Rheumatism Council Low-Dose Glucocorticoid Study
Group. Joint destruction after glucocorticoids are with-
drawn in early rheumatoid arthritis. Br J Rheumatol
1998;37:930±6.
5. Mulherin D, Fitzgerald O, Bresnihan B. Clinical improve-
ment and radiological deterioration in rheumatoid arth-
ritis:evidencethatthepathogenesisof synovialin¯ammation
and articular erosion may differ. Br J Rheumatol 1996;
35:1263±8.
6. Mulherin D, FitzGerald O, Bresnihan B. Synovial tissue
macrophage populations and articular damage in rheuma-
toid arthritis. Arthritis Rheum 1996;39:115±24.
7. McGonagle D, Gibbon W, O'Connor P et al. A pre-
liminary study of ultrasound aspiration of bone erosion in
early rheumatoid arthritis. Rheumatology 1999;38:329±31.
8. van der Heide A, Remme CA, Hofman DM, Jacobs JWG,
Bijlsma JWJ. Prediction of progression of radiologic
damage in newly diagnosed rheumatoid arthritis. Arthritis
Rheum 1995;38:1466±73.
9. Hassell AB, Plant MJ, Clarke S et al. Small joint synovitis
in rheumatoid arthritis: should it be assessed separately?
Br J Rheumatol 1995;34:51±5.
10. Kirwan JR. The relationship between synovitis and
erosions in rheumatoid arthritis. Br J Rheumatol 1997;
36:225±8.
11. Jevtic V, Rozman B, Watt I, Presetnik M. Use of contrast
enhanced MRI in the assessment of therapeutic response
to a disease-modifying and anti-rheumatic drug. Br J
Rheumatol 1995;34:956±9.
12. Fell HD, Jubb RW. The effect of synovial tissue on the
breakdown of articular cartilage in organ culture. Arthritis
Rheum 1978;20:1359±72.
13. Larsen A, Dale K, Eek M. Radiographic evaluation of
rheumatoid arthritis and related conditions by standard
reference ®lms. Acta Radiol 1977;18:481±91.
14. Thompson PW, Silman AJ, Kirwan JR, Currey HLF.
Articular indices of joint in¯ammation in rheumatoid
arthritis: correlation with the acute phase response.
Arthritis Rheum 1987;30:618±23.
15. Prevoo MLL, van Riel PLCM, van't Hof MA et al.
Validity and reliability of joint indices. A longitudinal
study in patients with recent onset rheumatoid arthritis.
Br J Rheumatol 1993;32:589±94.
16. Weyland CM. New insights into the pathogenesis of
rheumatoid arthritis. Rheumatology 2000;39(Suppl. 1):
3±8.
17. Gay S, Gay RE, Koopman WJ. Molecular and cellular
mechanisms of joint destruction in rheumatoid arthritis:
two cellular mechanisms explain joint destruction?
Ann Rheum Dis 1993;52:S39±S47.
18. Firestein GS. Invasive ®broblast-like synoviocytes in
rheumatoid arthritis. Arthritis Rheum 1996;39:1781±90.
19. Wassenberg S, Rau R, Ratingen HZ. Low dose
prednisolone therapy (LDPT) retards radiographically
detectable destruction in early rheumatoid arthritis.
Arthritis Rheum 1999;42:S243.
20. van Everdingen AA, Jacobs JWG, van Reesema, DR
Bijlsma JWJ. Low dose glucocorticoids in early RA inhibit
radiological joint damage. Arthritis Rheum 1999;42:S271.
21. Zeidler H, Rau R, Stinfeld P for the Low Dose
Prednisolone Therapy (LDPT) Study Group. Ef®cacy
and safety of low dose prednisolone in early rheumatoid
arthritis (RA). Arthritis Rheum 1999;42:S271.
301Synovitis and progressive erosions in RA
