Article Text

Extended report
Ultrasound colour Doppler is associated with synovial pathology in biopsies from hand joints in rheumatoid arthritis patients: a cross-sectional study
  1. Martin Andersen1,2,
  2. Karen Ellegaard1,
  3. Josephine B Hebsgaard3,
  4. Robin Christensen1,
  5. Søren Torp-Pedersen1,
  6. Peter H Kvist3,
  7. Niels Søe4,
  8. John Rømer3,
  9. Nina Vendel5,
  10. Else Marie Bartels1,
  11. Bente Danneskiold-Samsøe1,
  12. Henning Bliddal1
  1. 1Department of Rheumatology, The Parker Institute, Copenhagen University Hospital, Copenhagen, Denmark
  2. 2Department of Translational Immunology, Biopharmaceutical Research Unit, Måløv, Novo Nordisk, Denmark
  3. 3Department of Histology, Biopharmaceutical Research Unit, Måløv, Novo Nordisk, Denmark
  4. 4Department of Orthopaedics, Section of Hand Surgery, Gentofte University Hospital, Gentofte, Denmark
  5. 5Department of Anaesthesiology, Intensive and day-care Surgery, Gentofte University Hospital, Gentofte, Denmark
  1. Correspondence to Professor Henning Bliddal, Department of Rheumatology, The Parker Institute, Frederiksberg Hospital, Nordre Fasanvej 57, Frederiksberg, Copenhagen 2000, Denmark; hb{at}frh.regionh.dk

Abstract

Objectives Little is known regarding the association between ultrasound-determined pathological synovial blood flow and synovial pathology in rheumatoid arthritis (RA). We therefore examined the association between colour Doppler ultrasound imaging and synovitis assessed by histopathology and specific cell markers by immunohistochemistry in patients with RA.

Methods 81 synovial sites from wrist and finger joints from 29 RA patients were evaluated by ultrasound colour Doppler and subsequently biopsied by needle arthroscopy. The association between ultrasound colour fraction and an overall synovitis score and immunohistochemical staining for CD3, CD68, Ki67 and von Willebrand factor was investigated, including repeated samples from the same patients. The overall synovitis score (total 0–9) assessed synovial lining hyperplasia (0–3), stromal activation (0–3) and inflammatory infiltration (0–3). Data were clustered within patients, thus a linear mixed model was applied for the statistical tests. Parsimony in the statistical models was achieved omitting covariates from the model in the case of what was judged no statistical significance (p>0.1).

Results Doppler colour fraction showed an association with the overall synovitis score (approximated Spearman, approximately r=0.43, p=0.003). The density of all immunohistochemical stainings showed a significant association with Doppler colour fraction: von Willebrand factor (approximately r=0.44, p=0.01), CD68 (approximately r=0.53, p=0.02), Ki67 (approximately r=0.57, p=0.05) and CD3 (approximately r=0.57, p=0.0003).

Conclusions Colour Doppler activity is associated with the extent of inflammation present in the synovial biopsies from RA patients. However, synovial pathology was also seen in biopsies taken from Doppler negative sites.

  • Rheumatoid Arthritis
  • Ultrasonography
  • Synovitis

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction

Analysis of synovial biopsies has identified new biomarkers associated with disease hallmarks in rheumatoid arthritis (RA) and identified key mediators involved in inflammatory joint disease.1–4 Several methods have been applied for obtaining synovial biopsies, but the gold standard is performed by an arthroscopic procedure.5 Imaging by, for example, Doppler ultrasound (DUS) is a very sensitive method of general interest for the determination of increased flow in joints as a sign of inflammation.6 DUS has shown superiority detection of arthritis compared to clinical examination, and DUS has predictive value in arthritis regarding response to treatment and development of erosions.1 ,7–9 With DUS, perfusion is displayed as an overlay on the grey-scale image. Increasing perfusion will show increasing areas of the synovium with colour. This may be quantified as a percentage/the colour fraction (colour area per grey-scale area in the region of interest (ROI)). The pannus of RA is characterised by pronounced angiogenesis and massive cellular infiltration. These highly metabolic active areas may correspond to the areas of abnormal synovial blood flow detected by DUS. It remains to be clarified whether histological findings in biopsies from these specific areas correspond to the DUS obtained in the same area.

A few studies, conducted on relatively small populations (five to 16 RA patients), have shown a correlation between DUS and synovial vascularity in RA.10–12 However, only two studies considered synovial inflammation, and the results were conflicting.13 ,14 In former studies, biopsies have mainly been obtained during replacement of knee joints at end-stage disease and may not be representative of the typical changes with RA. Knee joints with long-standing arthritis often have concomitant osteoarthritis, which may by itself show Doppler activity15 ,16 without histological changes corresponding to inflammatory disease.9 ,12 ,15 Being non-invasive, DUS would be of advantage in the diagnostic evaluation of disease activity, prognosis and response to treatment.

Further validation and exploration of the translatory aspects regarding DUS activity and synovial inflammation are necessary. The aim of our study was therefore to examine the association between DUS obtained with highly sensitive ultrasound equipment and synovial pathology in hand joints from 29 RA patients. DUS activity was calculated as the ratio of DUS colour/synovial grey scale in the ROI. For the histological examinations of the synovial tissue we used a renowned synovitis score and applied immunohistochemical staining of cell subsets involved in RA pathology (T cells and macrophages) and evaluated the extent of cellular division by the cellular division marker, Ki67. We chose the vessel marker, von Willebrand factor (vWF), for the evaluation of synovial vascularisation.

Methods

Patient recruitment

Patients with RA17 and synovial hypertrophy determined by ultrasound grey scale were eligible for the study. Patients under 18 years of age, allergy to local anaesthetics, anticoagulant treatment, prednisolone dose above 7.5 mg/day, or bad skin at the site of interest, were excluded. Otherwise no patient selection occurred. The enrolment period dated from March 2010 through to January 2012. Patient recruitment, clinical and ultrasonic examination were performed at the outpatient clinic of the Department of Rheumatology, Frederiksberg Hospital, Denmark. The needle-arthroscopic procedure was carried out at the Department of Hand Surgery, Gentofte Hospital. Synovial biopsies were analysed at the histology section, Biopharmaceutical Research Unit, Novo Nordisk, Måløv. Micro-arthroscopy was performed within 24 h after clinical and ultrasonographic examination. The joints (three of less per patient) that seemed most inflamed clinically and by DUS were selected for biopsy. The study was approved by the local ethics committee (no. H-4-2009-117), and signed informed consent was obtained from each patient.

Clinical and ultrasound outcome measures

Tender and swollen joints were assessed by an experienced rheumatologist (HB) applying a semiquantitative scale ranging from 0 to 3 (0=none, 1=mild, 2=moderate 3=high pain/swelling). Ultrasound evaluation was performed by an experienced ultrasound specialist (KE, intrareader reliability 0.95).18 Colour DUS was used to assess the vascularisation of the synovial tissue. The ROI was defined as the synovial tissue. Maximum colour fraction (CFmax) in the ROI was selected as a marker of synovial inflammation. Colour Doppler mode was chosen as a marker of synovial inflammation because the sensitivity of detecting synovial blood flow on the GE Loqic E9 (Milwaukee, Wisconsin, USA) was higher compared to the power Doppler mode.19 Ultrasound examinations were performed as previously described using a 14 MHz centre frequency linear array matrix A. This was preset for the ultrasound examinations and installed and remained unchanged throughout the whole study period. The Doppler preset was adjusted for maximum sensitivity for low flow (pulse repetition frequency of 0.4 kHz, lowest wall filter on 45 Hz and 7.5 MHz Doppler frequency) with Doppler gain just below noise level.20 In all patients the wrists, metacarpal phalangeal (MCP) joints and proximal interphalangeal (PIP) joints were examined with ultrasound to identify the joints with most pronounced involvement. All examinations were performed from standardised dorsal and dorso-lateral positions. This was effected by having specific anatomical landmarks in the ultrasound image in the various positions.21

The anatomical landmarks are listed in the supplementary files, appendix 1 (available online only). Ventral scanning positions were omitted because biopsy material was solely taken from the dorsal part of the joint. Following identification of the anatomical landmarks in the grey-scale image, the Doppler was activated and, while keeping the landmarks in the image, the transducer was adjusted until the scan plane with the most Doppler activity was identified. The transducer was held in this position for a couple of heart cycles, whereupon the image was frozen. With the cine-loop function, the frames with maximum and minimum Doppler activity, corresponding to the systole and diastole, were stored in DICOM format. The digitally stored images were transferred to a processing program (ImagePro). The synovial Doppler activity was calculated as the ratio of the systolic DUS pixel count per unit of grey-scale pixel count, defined as CFmax.22 In the wrist joint the synovial tissue in the radiocarpal and mid-carpal was evaluated separately, if possible.

Needle-arthroscopic biopsy procedure

To ensure optimised conditions for obtaining a biopsy corresponding with the DUS position, the surgeon (NS) was not blinded for DUS findings, disease history or medical status. A small joint arthroscope (1.9 mm with a 30° angle of vision (Karl-Storz, Tuttlingen, Germany) was used for all joints, and the procedure was carried out under local anaesthesia in the operating room of the orthopaedic department. A method using two portals was used in order to obtain biopsies under direct visual control using a 2 mm forceps. The wrist portals were established laterally for the extensor tendons. For PIP and MCP joints the landmarks for the portals were the small concavities dorso-ulnar and dorso-radial for the central slip in the PIP joint and through the sagital band in the MCP joint. With these portal locations it was possible to obtain a good joint visualisation and sufficient distance between the arthroscope, forceps and the joint structures.

Histological and immunohistochemical assessments

Synovial histopathology was determined by a score developed by Krenn et al21 for grading chronic synovitis. The score ranges from 0 to 9 and consists of three components, each scored from 0 to 3: hyperplasia of the synovial lining (0–3), activation of synovial stroma (0–3), and inflammatory infiltrates (0–3). Scores ranging 0–1 correspond to no synovitis, 2–3 slight synovitis, 4–6 moderate synovitis and 7–9 strong synovitis. The density of vWF, CD3, CD68 and Ki67-positive cells was calculated as the fraction of the immunoreactive area in the whole synovial section area.

Biopsies were fixed in 4% formalin overnight at 4°C, paraffin embedded, and cut into 3 µm sections. Standard H&E staining was performed for Krenn analysis. Immunohistochemical stainings were performed using monoclonal rabbit anti-human-CD3 (0.3 µg/ml, RM-9107-S1; Thermo Scientific), mouse anti-human-CD68 (0.1 µg/ml, M 0814; DAKO), mouse anti-human-Ki67 (0.5 µg/ml, M7240; DAKO) and polyclonal rabbit anti-human-vWF (0.03 µg/ml, A 0082; DAKO). Microwave heat-induced antigen retrieval was performed in TEG buffer (pH 9) for 15 min. Endogenous peroxidase and biotin activity were blocked by 0.5% H2O2 and avidin-biotin blocking kit (DAKO), respectively. Non-specific antibody binding was blocked with pre-incubation solution (7% normal donkey, 3% human serum, 3% bovine serum albumin, 3% skim milk and 0.1% Tween in Tris-buffered saline) for 2 h at room temperature (RT) 20°C. Primary antibodies were diluted in pre-incubation solution (skimmed milk reduced to 0.5%) and applied to sections overnight at 4°C. With the exception of the CD3 staining, biotin-conjugated secondary antibodies were diluted in pre-incubation solution (skimmed milk reduced to 0.5%) and applied to sections for 1 h at RT. Antibody binding was detected with Vectastain ABC complexes (Vector). For CD3, Envision-horseradish peroxidase-conjugated secondary antibody was applied to sections for 30 min at RT. In all stainings, antibody binding was visualised with diaminobenzidine, and cell nuclei were counterstained with haematoxylin. Finally, sections were dehydrated in alcohol and xylene and mounted with Pertex. As negative controls, the primary antibodies were replaced by species-matched unspecific IgG isotype controls.

All immunostained sections were scanned at 40×, using the Nanozoomer 2.0 HT system (Hamamatsu, Glostrup, Denmark). Automated whole-slide image analysis was performed on the scanned images with the Visiopharm Integrator System (V.4.2.2.0; Visiopharm, Hørsholm, Denmark). All tissue slides were analysed by measuring the density of CD3, CD68, Ki67 and vWF immunoreactivity, and results were given as the percentage immunoreactive tissue area calculated as the number of immunoreactive pixels/total number of pixels in the ROI. A detailed description of the calibration of the automated analysis can be found in the supplementary files (available online only).

All stainings including HE were performed within 10 serial sections chosen at random by two experienced histologists (JBH, PHK) who were blinded to patient characteristics and colour DUS findings. Sections without a defined synovial lining, or with folded areas after staining, were omitted from the analysis. The synovitis score was calculated by PHK (intrareader reliability 0.84).

Statistics

Data were clustered within patients, thus a linear mixed model was applied for the statistical test in order to prevent double-counting errors due to inflated SE. Parsimony in the statistical models was achieved by omitting covariates from the model in the case of what was judged no statistical significance (p>0.1). Although the statistical inference was based on hierarchical models, it was decided that the Spearman rho estimate was important to the overall visual interpretation. These empirical correlations ignoring the double-counting are presented throughout as approximated Spearman coefficients—without inferring statistically from the inflated sample size. The statistical inference (eg, a p value) of these approximated r-coefficients is based on a sample size equal to the number of patients rather than the number of synovial sites (ie, in this study n=29). For model optimisation purposes the square root of CFmax and log 10 transformation was performed on Ki67 and CD3 densities. Statistical analyses were calculated by the use of SAS V.9.2 software. MA and RC (statistical analysis and data registration) had access to all data.

Results

Patient characteristics

Thirty-three patients were referred from the outpatient clinic to the programme. Three RA patients left the study before the arthroscopic procedure; two withdrew consent and one was excluded by the surgeon due to anticoagulant treatment. One patient was excluded from the data analysis due to a change in diagnosis (psoriatic arthritis) during review of the files.

A total of 29 RA patients fulfilled the inclusion criteria and were enrolled in the study (figure 1).

Figure 1

Patient flow. Patients with rheumatoid arthritis were included if they exhibited synovial hypertrophy determined by ultrasound grey scale in the joints of the hands. biopsies, total number of biopsies analysed; joints, number of individual joints biopsied. *Turned down by surgeon due to anticoagulant treatment.

The patients were primarily rheuma-factor positive women with long-standing disease duration and moderate disease activity on average. The patient population consisted of approximately one quarter treated with disease-modifying antirheumatic drug (DMARD) monotherapy (28%), DMARD combination therapy (21%), biological DMARD (28%) and no DMARD (24%), respectively. Four patients received prednisolone 7.5 mg or less per day in addition to the DMARD treatment.

Demographics are given in table 1.

Table 1

Demographics and clinical characteristics

Ultrasound, synovial biopsies and synovial histopathology

The colour fraction showed a median of 12%, with an IQR of 1–26%, corresponding to moderate synovitis. The min–max values ranged from 0% to 57%, reflecting no synovitis to high-grade synovitis. Eighty-one synovial biopsies were analysed in the study. As each patient might have biopsies from up to three joints, a total of 20, 10 and four patients had biopsies performed from wrists, MCP and PIP joints, respectively, resulting in 58 wrist, 15 MCP and eight PIP biopsies. The Krenn score ranged from 1 to 8, corresponding to a range of no synovitis to strong synovitis. Moderate synovitis (4–6) was most predominant and was seen in 40 biopsies (49% of all biopsies). Three biopsies (4%) showed no synovitis, 22 biopsies exhibited slight synovitis (27%), and strong synovitis was detected in six biopsies (7%). The results are summarised in table 2. A table with additional information on biopsy characteristics can be found in the supplementary files (available online only).

Table 2

Colour Doppler ultrasound and histology results

The association between DUS measured as the square root of the maximum colour fraction (√CFmax) and the synovitis score was borderline to be significant in the unadjusted model (p=0.09). Following adjustment for non-significant covariates, the p value decreased to p=0.003 (approximated r=0.43). All immunohistochemical markers showed significant association with √CFmax in the unadjusted model apart from vWF (p=0.096), which decreased to p=0.01(approximated r=0.44) after adjustment according to the procedure described above.

Table 3 gives an overview of the statistics, and figures 2 and 3 show scatter plots depicting CFmax and the histological synovitis score and the immunoreactive areas for the vWF, CD68, CD3 and Ki67 stainings in the synovial section.

Table 3

RA histology versus √CFmax stepwise covariate elimination

Figure 2

Scatter plot of the association between the maximum colour fraction (CFmax) (x-axis) and the Krenn score. Squareroot CFmax was used for the statistical calculations. Approximated Spearman, r=0.44.

Figure 3

Scatter plot of the association between colour Doppler ultra sound (DUS) and cellular markers in the biopsy. Colour DUS is depicted as the maximum colour fraction (CFmax) at the x-axis, whereas the percentage of biopsy area positive for the histological marker is depicted on the y-axis. Squareroot (√)CFmax was used for the statistical calculations. (A) Association between √CFmax and von Willebrand factor (vWF). y-axis depicts vWF density (percentage) in the biopsy. Approximated Spearman, r=0.44. (B) Association between √CFmax and CD68. y-axis depicts CD68 percentage in the biopsy. Approximated Spearman, r=0.53. (C) Association between √CFmax and Ki67. y-axis depicts Ki67 percentage in the biopsy. Approximated Spearman, r=0.53. (D) Association between √CFmax and CD3-positive cells. y-axis depicts CD3 percentage in the biopsy. Approximated Spearman, r=0.57.

Discussion

In this study of hand joint synovitis in RA, DUS was associated with the overall pathology in the synovium, synovial vascularity as indicated by vWF, and the presence of well-known cell subsets involved in the immune reactions of this disease. Our findings support previous observations of DUS as a sensitive indicator of local disease activity. The established associations between DUS and synovial pathology were robust because the selection process for synovial sections for histological analysis was arbitrary for each histological variable. However, using the histological picture as baseline with regard to blood flow, DUS showed false negative results. This is in accordance with a previous study,13 in which the absence of DUS did not rule out synovitis. Takase et al showed a higher correlation between power DUS and synovial pathology in RA patients undergoing knee arthroplasty than observed here, possibly due to the set-up in which the power DUS score was compared with the most inflamed synovial section present in the biopsy.14 Our results were based on arbitrarily selected synovial tissue sections and automated calculation of densities of histological markers of the whole biopsy section, offering an objective and unbiased method to determine synovial pathology.

Too much pressure with the ultrasound probe, resulting in negative Doppler, was avoided by our very experienced investigator, who ensured a recognisable layer of gel beneath the probe. Day-to-day changes in synovial microcirculation have been described and may thus have occurred within the 23–24 h lapse between DUS and arthroscopy.24

Variations were pronounced between synovial pathology and the colour DUS score, even though the surgeon was guided by the colour DUS description during the biopsy procedure. Less histological variability would probably be achievable if the site of biopsy was matched exactly to the scanned area. A multiple sampling procedure in which the histological cellular density could be reported as an average from various section depths of the biopsy would probably further decrease histological variability. A study showed that a section area of 2.5 mm2 from one depth was sufficient to keep the cellular variation below ±10% of the mean cell number in five out of nine RA patients.25 As our sections, on average, were 4.3 mm2 we assume that cellular variation within the biopsy did not play a major role in our setting. The automated procedure for calculating the cellular densities in the immunohistochemical stainings was based on pixel counts and not manual nucleus (cell count) counts, which may raise some challenges from background effects. However, our background was extremely low (data not shown), and our in-house validation data from skin psoriasis plaques showed very good correlation (r2=0.93 for CD3 and Ki67, data not shown). The factors regulating synovial blood flow and synovial cellularity may also exhibit different kinetics, resulting in a quicker change in synovial perfusion than synovial cellularity. This phenomenon may explain the ultrasound detected variation in the amount of tissue, that is, grey-scale changes, versus the vascularisation, that is, DUS, and resulting in Doppler negative biopsies with synovial histopathology, which was the case in 15% of the biopsies. This issue has not been solved; however, colour fraction is very well characterised and is a reasonable choice for the comparison between different joints as it reflects perfusion as millilitres/volume tissue.19

In conclusion, colour DUS was highly significantly related to the degree of inflammatory changes and can on a group basis be used for the determination of synovial activity. Taking the increasing accessibility of DUS in everyday practice into account, as well as the ongoing technological improvement in ultrasound equipment, this modality classifies a practical and feasible tool for evaluation of the inflammatory state in RA joints.

Acknowledgments

The authors would like to thank the study participants and B. Jørgensen for preparation of the synovial biopsies.

References

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:

Footnotes

  • Handling editor Tore K Kvien

  • Contributors All authors participated in the design of the study, interpretation of the data, writing of the article and approval of the final version.

  • Funding This study was supported by unrestricted grants from Novo Nordisk, the Danish Agency for Science, Technology and Innovation and the Oak Foundation.

  • Competing interests MA, JBH, PHK and JR are employees at Novo Nordisk. JBH, PHK and JR hold Novo Nordisk stocks. The remaining authors declare no competing interests.

  • Ethics approval The study was approved by the local ethics committee (no. H-4-2009-117).

  • Patient consent Obtained.

  • Provenance and peer review Not commissioned; externally peer reviewed.