| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Diagnostic Imaging, Research Institute, The Hospital for Sick
Children, 555 University Ave., Toronto, ON M5G 1X8, Canada.
2 Department of Medical Imaging, University of Toronto, Toronto, ON,
Canada.
3 The Brain-Body Institute, St. Joseph's Healthcare, Hamilton, ON, Canada.
4 Departments of Radiology and Medical Physics and Applied Radiation Sciences,
McMaster University, Hamilton, ON, Canada.
5 Department of Public Health, Family and Community Medicine, The Hospital for
Sick Children, Toronto, ON, Canada.
6 Department of Pathology and Laboratory Medicine, Mount Sinai Hospital,
Toronto, ON, Canada.
Received July 28, 2004;
accepted after revision March 8, 2005.
Address correspondence to A. S. Doria.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. Dynamic contrast-enhanced MRI was performed in nine arthritic and three nonarthritic knees of juvenile rabbits. A two-compartment pharmacokinetic model of signal intensity-time data was implemented to generate parametric maps of signal slope, maximal percentage of signal change, capillary permeability, leakage space volume, and time-to-peak. MRI values were compared with clinical, laboratory, and histologic markers for evaluation of synovial changes during the progression of arthritis.
RESULTS. Parametric maps of capillary permeability and signal slope depicted significant differences between arthritic and nonarthritic knees. Arthritic knees showed increased capillary permeability (p = 0.006) and signal slope (p = 0.01) with time after onset of disease as opposed to nonarthritic knees (permeability, p = 0.65; slope, p = 0.56). Significant correlations were found between temporal changes in capillary permeability (p = 0.002), signal slope (p = 0.003), and serum concentrations of amyloid A. No relationship was noted between any MRI parameters and histologic scores. The discriminative power of MRI indexes varied according to the stage of arthritis: time-to-peak was most accurate for differentiation of presence versus absence of arthritis in early arthritis (day 1, p = 0.0002), and signal slope was most accurate in midterm arthritis (day 14, p = 0.001).
CONCLUSION. In vivo capillary permeability and signal slope have distinctive dynamic MRI properties. The accuracy of MRI parameters for diagnostic evaluation of experimental arthritis differs according to the stage of disease.
Keywords: animal studies • contrast media • dynamic MRI • pediatric imaging • musculoskeletal imaging • rabbit
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MRI has been used experimentally to characterize and quantify microvasculature, providing information about tissue microvessel structure and function [4, 5]. On dynamic contrast-enhanced MRI (DCE-MRI), small molecular contrast material administered IV quickly equilibrates between blood and most compartments of the extracellular space. Analysis of the synovium enhancement curves can estimate the physiologic properties of the synovial microvasculature including perfusion, blood/plasma volume, and transendothelial permeability of the contrast agent using T1-weighted images obtained over the course of several minutes. Although previous studies have established the capability of DCE-MRI to distinguish malignant from benign tissues in many tumors [6, 7], limited data are available in the literature about the ability of DCE-MRI to assess the degree of synovial inflammation in rheumatoid arthritis as related to time points from disease onset.
The rabbit antigen-induced arthritis model used in this study resembles human rheumatoid arthritis in several aspects, including histopathology and response to therapeutic agents [8]. The animals used in the current study were skeletally immature animals to mimic changes in growing joints of children with arthritis.
Because the antiangiogenic compounds entering clinical trials are aimed at different points in the angiogenesis cascade, the choice of optimal DCE-MRI parameters for diagnosing physiologic changes at different time points becomes essential. The goals of this study were to determine whether measurements of DCE-MRI parameters in antigen-induced arthritis are different in arthritic (albumin-injected) and control (noninjected) knees over time; to define the MRI parameter that best discriminates between presence and absence of knee arthritis at different time points in progression of disease in an animal model; and to evaluate whether measurements of MRI parameters correlate with clinical, laboratory, and histologic synovium markers.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Animal Model
Arthritis was induced by intraarticular injection of an antigen (bovine
serum albumin, Sigma Chemical Co.) into one of the knees of nine presensitized
juvenile (10-week-old, 2.2-2.8 kg) male New Zealand white rabbits
(experimental day 0) as described previously
[9]. Only animals with a
positive skin test (indurated erythema > 1.0 cm) were eligible for
induction of arthritis. Injected knees were randomly selected among the
rabbits. Three rabbits did not undergo any injection (control group). Only one
knee underwent dynamic MRI scanning, and the same joint was scanned during all
time points. Animals were sacrificed after the final DCE-MRI scan at day 28
after induction of arthritis. Two of the nine rabbits in the study died
unexpectedly from unrelated disease, one at day 7 and the other at day 21
after induction of arthritis. MRI data from these rabbits were included in the
statistical analysis, taking into consideration adjustments for time points.
None of the animals' joints developed septic arthritis during the experiment.
This was confirmed by negative cultures of the synovial tissue obtained
immediately after euthanasia.
MRI Data Acquisition
MRI was performed on a 1.5-T Signa LX MRI unit (GE Healthcare) with a
3-inch surface coil applying the equilibrium method
[10]. Before imaging, a
25-gauge catheter was inserted into the rabbit's ear vein. A T1 map, modified
for spiral k-space acquisition, was obtained before administration of the
contrast agent [11,
12]. For dynamic scanning, the
anesthetized animals underwent a sagittal T1-weighted half-echo 2D spoiled
gradient-recalled single-slice sequence (TR/TE, 8.5/3; phase-encoding steps,
128; thickness, 3 mm; field of view, 7 cm; bandwidth, 31.25 kHz; matrix, 256
x 128; number of excitations, 2; flip angle, 30°; scanning time, 4
min 32 sec) during a bolus injection of 0.3 mmol/L/kg of gadopentetate
dimeglumine (Magnevist, Berlex Canada). The contrast agent was injected 10 sec
after initiation of image acquisition at a rate of 1 mL/sec using an automated
MRI-compatible power injector. A total of 250 images were acquired during the
DCE-MRI scan, with a temporal resolution of 1.088 sec per image. A single
sagittal slice was chosen from a coronal localizer through the central portion
of the knee joint.
This imaging protocol was repeated at six consecutive time points: days 0, 1, 7, 14, 21, and 28 after induction of arthritis. Images were subsequently transferred offline to a Pentium III (Intel Corp.) PC for processing.
MRI Data Analysis
Contrast enhancement was quantified using a two-compartment pharmacokinetic
model of signal intensity-time data
[13]. Investigation of
undesirable movement effects and unsuccessful injection of gadolinium chelate
that could have potentially occurred during imaging acquisition was performed
by cinematic depiction of gray-scale images in a movie loop. If any movement
effect was detected, that particular examination was excluded from the
analysis.
Color-encoded parametric images derived from the pharmacokinetic model were
displayed and used to define 2-cm2 regions of interest (ROIs)
(Fig. 1). A single ROI was
placed on the suprapatellar region of the rabbits' knees. This region
undergoes significant swelling in early arthritis. Enhancing synovial tissues
of the suprapatellar bursa of the knee were manually segmented from
contrast-enhanced images by a single operator. This operator was blinded to
all clinical information and therefore to whether a given knee belonged to a
rabbit from the control group or the study group. Parametric maps at the pixel
level, implementing the Tofts' model
[14], and a least-squares
minimization procedure, using a Marquardt-Levenberg algorithm
[15], were calculated using
Matlab, version 6.5 (The Mathworks). The analysis produced parametric maps of
signal slope (dS/dt = incremental rate of enhancement after the arrival of the
contrast bolus), maximal percentage of signal change (
S = peak
concentration of contrast agent in both the intravascular and
extravascular-extracellular space), capillary permeability to the contrast
agent (KPS
), volume of the extravascular leakage space
(VE), and time-to-peak (TTP) as defined elsewhere
[14].
|
Laboratory Measurements
Measurements of the serum concentration of amyloid A were obtained at each
time point with a solid-phase sandwich enzyme-linked immunosorbent assay
(ELISA) as previously described
[16]. We used commercial kits
(Phase Serum Amyloid A Assay, Tridelta Development).
Histologic Methods and Grading
After the animals were euthanized, synovial sections were removed from the
suprapatellar, infrapatellar, lateral, and posterior recesses of the knee;
fixed in buffered formalin; and processed and stained with H and E for
histologic evaluation. Histologic sections were selected based on their
relative anatomic location within the complete data set. Evaluation of
histologic sections was performed using an optical microscope and an ocular
micrometer with 40-power objective and 10-power ocular lenses. The grading
system included six histologic parameters of synovial inflammation, of which
three represented acute synovitis (polymorphonuclear cell infiltration,
synovial surface integrity, and vascular proliferation) and three were
evidence of chronic synovitis (synovial cell hyperplasia, lymphocyte cell
infiltration, and fibrosis). This grading system was devised in our
institution (Mount Sinai Hospital) based on a scale reported elsewhere
[17].
Each of the histologic features was scored on a scale of none (0), mild (1), moderate (2), or severe (3). Synovial samples of suprapatellar, infrapatellar, lateral, and posterior recesses of the knee were scored separately according to the aforementioned scoring system. A single score was given for each location of that knee. The scores for the entire joint, however, were obtained on the basis of the most severe findings noted on the synovial sections available for each knee. Synovial vascularity was graded by means of a semiquantitative score that ranged from 0 (none) to 3 (highly vascularized). Because vascularity was separated out as a feature of the acute synovitis score, acute synovitis and vascularity scores were not independent of each other. A final acute score was obtained by adding up the three scores pertaining to acute inflammation, which yielded a maximum score of 9. Similarly, a final chronic score was also obtained by summation of the three scores pertaining to chronic inflammation, yielding a score out of 9. Therefore the overall score incorporating scores of acute and chronic inflammation ranged from 0 to 18.
Statistical Analysis
We compared baseline clinical measurements of arthritic and nonarthritic
knees and the rabbits' weights recorded before and after injection using a
two-tailed paired Student's t test. Results of this study account for
the repeated-measures nature of the data. Measurements of the five MRI
parameters were compared with clinical, laboratory, and histologic markers
over time by repeated-measures analysis of variance and the statistical
significance of their associations and trends was computed using Pearson's
correlation coefficients and linear regression, respectively. The areas under
the receiver operating characteristic curves (AUCs) were compared by using a
nonparametric method as described elsewhere
[18]. For different MRI
parameters, the corresponding AUCs were compared by chi-square tests. The
injection or noninjection of albumin to knee joints set cutoff values for
creation of AUCs. A p value of less than 0.05 was considered to
indicate a statistically significant difference. All statistical analyses were
performed using SAS, version 8.2 (SAS Institute).
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The baseline clinical characteristics of rabbits from different knee categories (albumin injected or noninjected) were similar with regard to animal weight (p = 0.59), serum levels of amyloid A (p = 0.92), and clinical measurements (AP diameter, p = 0.91; LL diameter, p = 0.28).
All rabbit knees injected with antigen developed clinical signs of arthritis in the early course of disease and had a histologic diagnosis of arthritis at day 28 after injection of the antigen. The AP diameters of arthritic knees were significantly different before injection (mean ± SD, 34.03 ± 3.99 mm) compared with day 1 after antigen injection (40 ± 4.07 mm) (p = 0.006). The AP diameters of the noninjected knees were not different before injection (33.83 ± 3.82 mm) or on day 1 after antigen injection (35.48 ± 2.49 mm) (p = 0.69). Although the longitudinal design of this study did not enable a direct clinical-histologic correlation in early arthritis, results from a pilot experiment (unpublished data) showed that increases in the AP diameters of knees and in the serum concentration of amyloid A in the early course of arthritis correlated well (> 0.7) with histologic scores of acute synovitis. This pilot experiment compared clinical and laboratory data with histologic synovial scores at different time points.
DCE-MRI contrast-enhancement curves showed that the only MRI parameters
that changed significantly with time in albumin-injected knees were dS/dt
(p = 0.01) and KPS (p = 0.006). No
significant differences were observed in noninjected knees
(Table 1).
|
In arthritic knees, experimental maxima for dS/dt, KPS,
serum amyloid A, and clinical measurements were obtained in early arthritis
(day 1 after induction of arthritis) with a secondary peak noted in midterm
arthritis (day 14) (Table 1).
The mean clinical and laboratory measurements varied significantly across time
in albumin-injected knees (AP diameter, p = 0.01; serum amyloid A,
p = 0.01 by repeated-measures analysis of variance). Although two
peaks have also been noted for serum concentration of amyloid A at days 1 and
14 of arthritis, measurements obtained in rabbits with an albumin-injected
knee were significantly higher than measurements obtained in animals with
noninjected knees only in early arthritis (day 1 after induction, p =
0.0004; day 14, p = 0.35 by two-tailed paired Student's t
test).
Moderate but significant positive correlations were noted with both dS/dt
(r = 0.62, p = 0.003)
(Fig. 2A) and
KPS (r = 0.63, p = 0.002)
(Fig. 2B) when regressed with
serum temporal concentrations of amyloid A. Borderline negative correlations
were observed for TTP (r = –0.42, p = 0.049). No
significant correlations were found between clinical measurements (AP or LL
diameters) and any of the MRI parameters. Further, no relationship was noted
between any of the MRI parameters and synovial histologic scores.
|
|
(Az, 0.944) were
significantly larger than the AUCs of other MRI parameters (p =
0.001) (Fig. 3B), indicating
that these DCE-MRI parameters possess the greatest sensitivity and specificity
for differentiation between the presence or absence of arthritis of knee
joints. Whereas the TTP index held the highest accuracy for discrimination of
the inflammatory status of the joint in early arthritis, it showed the lowest
accuracy for this purpose in midterm arthritis (day 14).
|
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
). This phenomenon may be explained
by transendothelial macromolecular extravasation of vesiculovacuolar
organelles containing gadolinium, which are known structural components of
microvascular hyperpermeability in inflamed tissues
[19-21].
Although normal synovial tissues may also lead to increased enhancement,
arthritic tissues hold the distinctive characteristic of vascular leakiness,
which is more pronounced in abnormal tissues because of the presence of
vesiculovacuolar organelles
[22]. These abnormal tissues
tend to enhance early with a rapid and large increase in signal intensity
compared with that of the normal surrounding tissues
[23].
Although the dS/dt index can be influenced by exogenous factors (cardiac
output, contrast agent, and so on)
[10], which makes it less
reliable, in our study this parameter helped to distinguish between normal and
rheumatoid synovium [24]. The
differential KPS shown between arthritic and nonarthritic
joints is consistent with results obtained by others in studies comparing MRI
patterns of angiogenic-related musculoskeletal processes (e.g., fibrosarcomas)
with normal tissue [25].
The findings of this study show that the time point of maximum signal enhancement for permeability and slope in albumin-injected knees, day 1 after induction of arthritis, is coincident with the peak serum amyloid A concentration and with the maximum AP and LL joint diameters. In our study, a second, smaller peak in the serum concentration of amyloid A was noted on day 14 after induction of arthritis. The presence of a small, second rise in the serum concentration of erythrocyte sedimentation rate (ESR) in midterm arthritis is suggested, but is not conclusive, in a study conducted by Hunneyball et al. [26] in a rabbit model of arthritis.
Serum amyloid A proved to be more sensitive than ESR to quantify the
presence of inflammatory activity in our animal model in a previous pilot
study (unpublished data). As a result, we hypothesized that the amplitude of
changes is even higher for serum amyloid A than for C-reactive protein (CRP).
Our data also show that there is a positive correlation between dS/dt,
KPS, and serum amyloid A concentration measurements. Similar
correlations have also been established between dS/dt and cervical cancer
recurrence [10].
Regions of high vascular permeability would be characterized by low blood flow and hypoxia reflecting clinical and laboratory (ESR) inflammatory-angiogenic conditions of the joints [27]. However, few previous studies have convincingly shown a relationship between the dS/dt and clinically derived inflammatory activity in rheumatoid arthritis [3]. This is due to challenges in obtaining effectively timed MRI data such that a calculable time/effect course of disease after therapy can be depicted.
We hypothesized that the amplitude of laboratory findings of inflammation in early arthritis would be significantly higher than the degree of periarticular soft-tissue swelling, which would result in a significant correlation between laboratory and MRI measurements but not between clinical and MRI measurements. Nevertheless, although the degree of joint soft-tissue swelling in acute arthritis (day 1 after induction of arthritis) was minimal, it enabled differentiation from measurements obtained before induction of arthritis. This hypothesis is supported by the study by Hunneyball et al. [26] that showed a higher magnitude of inflammatory changes detected by laboratory measurements (CRP, 120 x increase in postinduction measurements compared with preinduction measurements) than by joint width measurements (8 x increase).
In contrast with results from previous studies that showed the rate of synovial membrane enhancement correlates with histologic features of acute inflammation [28, 29], we did not observe any relationship between the MRI parametric curves and the degree of inflammation on histology. There is controversy in the literature regarding the potential associations that may exist between histology and DCE-MRI findings in angiogenic-related pathologic processes. Whereas some studies have shown a positive correlation between tumor enhancement and microvascular density on histomorphometric assessment [30-32], other studies have shown no correlation [33-35]. Although we have used a semiquantitative scoring system rather than a histomorphometric procedure to quantify synovial vascularity, we postulate that microvascular density cannot be assumed to be the only determinant factor of tissue enhancement.
Histomorphometric parameters may reflect anatomic information at a static single time point and may not provide sufficient information on the dynamic microcirculation function of the tissue [10]. Moreover, the design of this study favored the imaging assessment since the same animal was evaluated over time, which tended to reduce the interrabbit variability of measurements. Nevertheless, histologic sections obtained at 28 days of arthritis would not be expected to correlate with DCE-MRI parameters acquired in the acute stage of arthritis when angiogenesis and inflammation are at their peak.
Our study shows that the diagnostic performance of MRI parameters varies across the time course of arthritis. We showed that the maximal signal-enhancement change in the synovium, which intuitively could be considered as a diagnostic criterion for presence of synovial inflammation, is not the best pharmacokinetic feature to differentiate arthritic and nonarthritic synovial tissue. This result is in keeping with information obtained in studies of breast tumors [36]. Furthermore, significant differences in the accuracy of TTP and dS/dt parameters used to discriminate between presence and absence of arthritis were detected. TTP allowed the most accurate discrimination between presence and absence of early arthritis and presented with a graded deterioration in diagnostic performance throughout disease progression. However, dS/dt was most discriminative of the inflammatory status of the joint in midterm disease (day 14 after arthritis induction).
Although our rate of procedure failure is within acceptable limits compared with rates reported by experienced personnel in the literature (5-7%) [37], our study has limitations. The time-intensity curves were directly influenced by the size and shape of the ROIs, and this latter parameter was dependent on the positioning of the joint, which could not be perfectly reproduced in the six examinations for each animal. The area of the ROI, however, was kept constant for the analysis of each examination. One characteristic of our rabbit model is that although soft-tissue swelling is noted in the acute phase of arthritis [38, 39], it does not produce significant joint effusion in the synovial recesses of the knee [40]. As a result, the ROIs drawn in this study encompassed the perisynovial tissues of the knee based on the anatomic location of the suprapatellar bursa, but we were unable to satisfactorily visualize the synovium membrane that overlies the bursas of the knee while drawing ROIs.
This process of averaging measurements for synovium and perisynovial soft tissues may have underestimated our results. More significant differences are expected to be noted in experiments where joint effusion is prominent, which would facilitate the isolation of the synovium from the adjacent tissues for definition of ROIs. This would certainly decrease the effect of averaging on MRI measurements. Note is made that the use of a micromolecular contrast medium to distinguish tissues with regard to their permeability to the contrast agent enables the contrast agent to be readily transported across the vascular wall, leaking out of normal and out of abnormal vessel walls. As a result, what we have measured in this study were dynamic MRI properties in periarticular soft tissues of the knee.
Although the results of this study were obtained from data extraction of average enhancement curves and ignored the heterogeneity of synovial vascular characteristics, recently Hayes et al. [41] showed that the average kinetic parameter estimate is a close approximation of parameter values obtained from individual pixels. Finally, the histopathologic-MRI correlations were performed by matching partial-mount histologic specimens selected through a standardized anatomic protocol, which may have represented only samples of the synovium, with MRI-derived maps of single-sliced MRI sections. Thus, a direct comparison between geometrically oriented ROIs of the MRI scans and the corresponding histologic sections is not warranted.
In conclusion, the findings of this study indicate that microvascular
permeability (KPS) and signal-enhancement slope (dS/dt) are
distinctive DCE-MRI properties with clinical impact that merits further
assessment. Our results show that the accuracy of MRI parameters for
diagnostic evaluation of experimental arthritis differs according to the stage
of disease. Our results also suggest that TTP may best discriminate arthritic
from nonarthritic joints in early disease. However, if a patient has midterm
arthritis, dS/dt should be evaluated. Although values of KPS
are different in inflamed and normal joints in early experimental arthritis,
this parameter may not be accurate enough to discriminate the inflammatory
status of the joint.
Dynamic MRI may allow rational selection of timing for treatment, which would reduce the duration of scanning—a factor that is especially important for the pediatric population—and could be beneficial for planning therapeutic strategies. In addition, with the proper imaging protocol, this technique may be helpful in assessing therapeutic efficacy, which would contribute to the development of new antiangiogenic drugs for treatment of rheumatoid arthritis. Further investigation, however, is required to evaluate the therapeutic effect of different categories of antiangiogenic drugs on the patterns of dynamic MRI contrast enhancement and to determine whether the results of this experimental study are directly translatable to patients with rheumatoid arthritis.
Acknowledgments
We thank Geraldine Kent for help with preparation of albumin solution,
Maria Mendes for literature search and discussion of synovial histologic
scores, Anguo Zhong for collaboration as an animal care technologist in part
of the study, Marvin Estrada and Vicky L. Hannam for animal injections, and
Niels Celeghin for analysis of data.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
monoclonal antibody therapy. Rheumatol Int1997; 16:249
-255[CrossRef][Medline]
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  E. Miller, E. Uleryk, and A. S. Doria Evidence-Based Outcomes of Studies Addressing Diagnostic Accuracy of MRI of Juvenile Idiopathic Arthritis Am. J. Roentgenol., May 1, 2009; 192(5): 1209 - 1218. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
