Objective To investigate the reliability of ultrasonographic assessment of osteophytes and explore the concordance of osteophytes detected by ultrasound, MRI, conventional radiography (CR) and clinical joint examination in patients with hand osteoarthritis (HOA).
Methods The study included 127 HOA patients (116 women, mean age 68.6 years (SD 5.8)) with ultrasound, CR and clinical examination of both hands and MRI of dominant hand. Osteophytes were assessed by all imaging modalities on 0–3 scales, whereas clinical bony enlargement was assessed as absent/present. An ultrasound atlas of ostephytes was developed, and the intra and inter-reader reliability of scoring ultrasound osteophytes on still images using the atlas as reference was examined. The reliability for ultrasound readings was examined with κ and percentage exact agreement (PEA) and percentage close agreement (PCA), and the sensitivity, specificity and PEA/PCA of ultrasound was calculated in comparison with MRI, CR and clinical examination.
Results Ultrasound had high sensitivity (0.83) and specificity (0.75) in detecting osteophytes compared with MRI, with excellent PCA (96.1%). Moderate/large osteophytes (grade 2–3) were demonstrated more often by ultrasound (n=401) than by MRI (n=288) in 851 interphalangeal joints. Ultrasound detected more osteophytes (53.2%) than CR (30.0%) and clinical examination (36.9%). Intra and inter-reader reliability of ultrasound was excellent (PEA >88%, PCA 100% and weighted kappa >0.91).
Conclusion Ultrasound can reliably assess osteophytes in patients with HOA. Good agreement was found between osteophytes detected by ultrasound and MRI, while ultrasound was more sensitive than CR and clinical examination, which could be due to a multiplanar joint demonstration by ultrasound.
Statistics from Altmetric.com
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.
Hand osteoarthritis (HOA) is a frequent joint disease, especially in older people.1 It has a potentially substantial impact on health-related quality of life,2 and the impact of HOA is expected to increase further with an ageing population. Conventional radiography (CR) is the standard imaging modality in the assessment of structural abnormalities in HOA,3 ,4 but ultrasonography is recognised as a useful imaging modality for the detection of synovitis and osteophytes in osteoarthritis studies.5,–,7 Recently published research priorities by the Outcome Measures in Rheumatology (OMERACT) Ultrasound Task Force include validation of sonographic scoring systems for osteoarthritis.8 Up to now there is no consensus on an ultrasound scoring system for the assessment of finger joints in HOA,9 but a proposed preliminary scoring system includes evaluation of osteophytes, grey scale synovitis and power Doppler signals.10
Osteophyte formation is a key feature of osteoarthritis,11 and sonographic and radiographic detected osteophytes in finger joints have been found to be related to pain12 as well as progression of radiographic signs of HOA.13 Furthermore, ultrasound has been shown to be more sensitive than CR in the detection of finger osteophytes,12 ,14,–,16 but there is little knowledge regarding the sensitivity of ultrasound compared with MRI. So far the only study addressing this issue included a limited number of patients (n=14) and the reproducibility of the MRI scoring was poor.17
High reliability was recently shown for the scoring of grey scale synovitis and power Doppler in patients with rheumatoid arthritis (RA) by the use of an ultrasound atlas as reference,18 and MRI features (including osteophytes) could reliably be assessed in HOA with the use of a recently proposed MRI atlas.19 We assumed that an imaging atlas for osteophytes in HOA would also be important for improved reliability of this joint feature.
The objectives of the present study on HOA patients were to examine the concordance of osteophytes in finger joints detected by ultrasound compared with MRI, CR and clinical joint examination, and second, to develop an ultrasound atlas including a semiquantitative scoring of osteophytes for examination of the reliability of sonographic scoring.
One hundred and twenty-seven patients from a follow-up study of the Oslo Hand Osteoarthritis cohort2 were assessed by ultrasound in addition to MRI, CR and clinical examination. All patients had either radiographic HOA and/or fulfilled the clinical criteria for HOA of the American College of Rheumatology.20 The patients gave written consent according to the Declaration of Helsinki, and the study was approved by the local ethics committee (the Regional Committee for Medical and Health Research Ethics, South-East).
One trainee (AM) and one experienced sonographer (HBH) performed the ultrasound assessments together and reached consensus on each scoring. A linear array transducer was used (5–13 MHz, Siemens Antares, Sonoline; Siemens Medical Solutions, Mountain View, California, USA) with fixed settings of the machine. To ensure standardisation, the same ultrasound machine without software upgrading was used throughout the study. The sonographers were blinded to the results of the other assessments.
Osteophytes were defined as cortical protrusions,21 and the presence of osteophytes was scored semiquantitatively as: 0, none; 1, minor; 2, moderate; 3, major size of osteophytes.10 ,18 The proximal and distal bones were assessed together for each joint and the largest osteophyte was scored independently of the number, size and location of other osteophytes in the same joint. The sonographers scored osteophytes in the following 15 joints bilaterally (standard scanning projections)22: carpometacarpal (CMC) 1 (radio-palmar), metacarpophalangeal (MCP) 1–5 (dorsal), (proximal) interphalangeal ((P)IP) 1–5 (dorsal) and distal interphalangeal (DIP) 2–5 (dorsal). Each joint was scanned longitudinally from the radial to the ulnar side, and transverse scanning was performed if there was uncertainty about the presence of pathology.
Ultrasound atlas of osteophytes
The sonographers collected still images from ultrasound examinations and AM developed a novel ultrasonographic atlas of osteophytes including representative example images from each of the different finger joint groups (figure 1, supplementary figure S1, available online only).
Magnetic resonance imaging
The dominant hand of 107 patients was examined using a high-field (1.0T) extremity MRI unit (ONI; GE Healthcare, Waukesha, Wisconsin, USA). The acquisition and scoring of MRI features have previously been described.19 In short, the second to fifth (P)IP and DIP joints were assessed for osteophytes using the Oslo Hand Osteoarthritis MRI score (grade 0–3). In contrast to the ultrasound examination, the scoring of MRI included both dorsal and palmar side of each joint. Proximal and distal bone was assessed separately and the highest score was used in the present calculations (ie, a joint with an osteophyte grade 3 distally and grade 1 proximally was interpreted as a joint with osteophyte grade 3). High intra and inter-reader reliability has been shown for scoring of osteophytes with use of this atlas (intraclass correlation coefficient >0.88), and MRI-defined osteophytes have also shown good validity against CR.19 ,23
Bilateral CR (posterior–anterior view) was performed on 117 of the patients. IKH examined the x-rays bilaterally for osteophytes in 15 joints according to the Osteoarthritis Research Society International (OARSI) atlas (grade 0–3): CMC 1, MCP 1–5, (P)IP 1–5 and DIP 2–5.24 High intra and inter-reader reliability has been shown for this scoring (intraclass correlation coefficient >0.89, weighted kappa (κw) >0.71).23
Clinical examination and questionnaires
One experienced rheumatologist (BSC), blinded to the imaging results, performed clinical examinations of all finger joints (CMC 1, MCP 1–5, (P)IP 1–5 and DIP 2–5) in both hands, and the presence of bony enlargements was assessed in each joint (giving the bony enlargement joint count). The patients also completed the following patient-reported outcomes: Australian/Canadian osteoarthritis hand index (AUSCAN),25,–,27 the functional index for hand osteoarthritis (FIHOA),28 ,29 and the arthritis impact measurement scales 2 (AIMS-2) hand/finger subscale.30 The AUSCAN consists of three subscales assessing hand pain, morning stiffness and physical function, respectively, while the FIHOA and AIMS-2 include items concerning physical function only. The subscale scores were calculated by summation of the assigned values.
Reliability of ultrasound
For the reliability exercise of osteophyte scoring by ultrasound, one of the authors (IKH) randomly selected 150 anonymous images of osteophytes from the database (10 images of each of the 15 individual finger joints) and rearranged the images for the second scoring. AM and HBH separately performed two scorings (1 week apart) of the osteophytes (0–3 scale) with the use of the ultrasound osteophyte atlas as reference. They were blinded to previous scoring results, and the calculations were performed by IKH.
All statistical analyses were performed by the use of SPSS statistics 17.0. The intra and inter-reader reliability of osteophyte assessments were calculated by the use of κw,31 percentage exact agreement (PEA) and percentage close agreement (PCA, defined as a difference of ±1 between the readings/readers). A κw value greater than 0.8 indicates excellent reliability.32 The agreement of osteophytes assessment at the individual joint level between ultrasound and MRI, CR as well as clinical bony enlargement was examined by PEA and PCA. We also calculated the sensitivity and specificity of ultrasound in the detection of osteophytes using MRI, CR or clinical examination as references, respectively. Comparisons of joint findings on the group level between two modalities were performed by the use of the Wilcoxon signed-rank test on dichotomised scores (‘osteophytes grade 0’ and ‘osteophytes grade 1–3’). The cut-off level for statistical significance was set at p<0.05. Comparison of three modalities was conducted by the use of the Friedman test, with post-hoc analysis comparing two and two modalities with the Wilcoxon signed-rank tests and Bonferroni correction (comparing three modalities gave an adjusted significance level set at p<0.017).
Demographic variables and clinical data from the cohort are presented in table 1. A total of 127 patients was examined clinically and by ultrasound, including the assessment of 3771 joints (39 joints were excluded because of prosthesis or ankylosis). CR of hands was available in 117 of 127 patients (10 patients were missing because of practical problems at the Department of Radiology) with 3405 scored joints (five patients mistakenly had CR of their dominant hand only, and 30 joints were excluded because of prosthesis or ankylosis). MRI was performed in 107 of 127 patients (20 patients missing because of practical problems or contraindications for MRI), with scoring of 852 joints (four joints were not evaluated due to limited field of view). The patients missing CR (n=10) and MRI (n=20) were not significantly different from the others regarding the number of osteophytes detected by ultrasound (p=0.61 and 0.56, respectively). A total of 3785 joints was examined clinically, including 14 joints with prosthesis or ankylosis.
Prevalence of osteophytes in different modalities
The prevalence of osteophytes detected by the different modalities (ie, score ≥1 for the imaging methods) is shown in table 2. Osteophytes were identified in 1989 of the 3771 joints examined by ultrasound (52.7%), and all patients in this cohort had osteophytes in at least four finger joints. The prevalence of osteophytes was similar in dominant and non-dominant hands, with a mean number of affected joints of 8.2 (SD 2.7) and 7.8 (SD 3.0), respectively (p=0.35).
In the 107 patients with both ultrasound and MRI of the second to fifth (P)IP and DIP joints in the dominant hand (n=851 joints), osteophytes were detected in 74.5% of the joints by ultrasound and 86.5% by MRI. The median number of interphalangeal joints of the second to fifth finger with the presence of osteophytes was 6 (IQR 4–8) by ultrasound and 7 (IQR 6–8) by MRI (p<0.001).
Fifteen joints in each hand were bilaterally examined by ultrasound, CR and clinical examination in 117 patients. On an individual joint level, osteophytes were detected more often by ultrasound (53.2%) than by CR (30.0%), while clinical bony enlargement was found in 36.9% of the joints. On the patient level, the median number of joints with osteophytes/bony enlargement detected by ultrasound, CR and clinical examination was 16 (IQR 12–20), 8 (IQR 4–14) and 11 (IQR 7–15), respectively (p<0.001). Post-hoc analysis demonstrated the number of affected joints by clinical examination to be significantly lower than the number of joints detected by ultrasound (p<0.001) and significantly higher than CR (p<0.001).
Concordance between ultrasound and imaging/clinical examinations
Ultrasound had high sensitivity and specificity with MRI as reference, and close agreement was excellent between the two methods (table 3). Severe pathology (grade 2–3) was detected more often by ultrasound (n=401) than by MRI (n=288) (p<0.001). Ultrasound detected osteophytes in all joints in which MRI had scored 2–3, except for two joints in which the osteophytes were found by MRI in the lateral part of the joints. Among 129 joints in which osteophytes were detected by MRI but not ultrasound, 127 of the joints were scored as grade 1 on MRI.
Among joints with no radiographic changes (n=2376), ultrasound demonstrated osteophytes in 832 joints. However, the majority of these (n=559) was considered small by ultrasound (grade 1). Conversely, CR detected osteophytes in 41 joints without identified ultrasound pathology, and 37 of these joints had scored 1 by CR.
In 76.6% of the joints, ultrasonographic and clinical assessment agreed on the presence or absence of osteophytes/bony enlargement. Ultrasound detected osteophytes in 736 joints without clinical bony enlargement, but 484 of these were scored 1 by ultrasound. Only 144 joints had clinical bony enlargement and no ultrasound-detected osteophytes.
Ultrasonographic atlas of osteophytes
An atlas was developed with representative ultrasound images of osteophyte scoring in finger joints. Four joint regions (CMC1, MCP, (P)IP and DIP) with four score categories (0–3) and five example images for each category (except MCP grade 3, with only three examples) resulted in an atlas with 78 images (see figure S1, available online only). An example of each score is illustrated in figure 1.
Two sonographers used the novel ultrasound atlas as reference for the reliability exercise (table 4), with excellent intra and inter-reader reliability for both readers and scoring sessions (PEA >88%, PCA 100% and κw >0.91).
This is the largest study until now with comparison of osteophytes examined by ultrasound and two other imaging modalities (MRI and CR) as well as clinical joint examination in HOA. Good agreement was found between osteophytes detected by ultrasound and MRI, whereas ultrasound was more sensitive than CR and clinical examination, which could be due to a multiplanar joint demonstration by ultrasound and MRI. To increase the reliability of scoring, a novel ultrasound atlas of osteophyte scores was introduced, and the use of this atlas resulted in excellent intra and inter-reader reliability of scoring ultrasound osteophytes.
To our knowledge, only one previous study with 14 patients only has compared the validity of ultrasound with MRI as reference, showing high agreement between the two modalities in the assessment of osteophytes.17 In contrast to our study, osteophytes were detected less frequently by MRI than ultrasound. However, MRI was performed without the use of standardised scoring methods and the inter-observer reliability was poor.
In our experience, a challenge when scoring osteophytes by ultrasound is to distinguish between scores 0 and 1. Furthermore, it was also difficult on fat-suppressed T1-weighted MRI scans to distinguish between small osteophytes (grade 1) at the dorsal side of the joint and the extensor tendon. This may have led to an overestimation of osteophytes on MRI, and could explain why small osteophytes were detected by MRI in 127 joints without abnormal ultrasound findings. Another explanation for MRI osteophytes missed by ultrasound could be that the MRI scoring was performed on the whole joint circumference, while ultrasound was performed on the dorsal part of the joints only.
In our study, ultrasound detected more osteophytes in finger joints than CR, which is in line with several recently published studies.12 ,14,–,16 This is most likely due to the multiplanar imaging capability of ultrasound, and the fact that posteroanterior radiographs are only able to detect osteophytes tangentially to the beam and therefore mainly on the lateral and medial side of the joints. Therefore, osteophytes located on the dorsal or palmar side of the joints are less easily detected by CR. Furthermore, a more strict definition of osteophyte on CR (ie, the OARSI atlas only shows examples of clear definite osteophytes) may have contributed to a lower number of affected joints. Slightly more joints had clinical bony enlargement (36.9%) than joints with osteophytes found on CR (30.0%). Osteophytes and bony enlargements are not necessarily the same feature,33 but our observation indicates that a thorough clinical assessment alone can provide useful information,3 even though pathology in some joints was underestimated.
Consistent with previous studies,1 ,34 we found no evidence of a higher prevalence of osteophytes in the dominant hand, which suggests that HOA is not only caused by mechanical stress. Neurogenic, hormonal, metabolic or genetic factors may play a role.35,–,39
By the use of the novel ultrasound atlas as reference, high intra and inter-reader reliability of osteophyte scoring was currently found. Our reliability exercise included scoring of still images, while performing ultrasound examination on patients would have been the optimal test of reliability. The sparse previous literature on the reliability of ultrasound in detecting osteophytes in finger joints of patients has demonstrated acceptable to high agreement between readers.10 ,17
With the introduction of new imaging modalities, scoring systems may translate into improved evaluation of morphological changes. Haugen et al19 recently published the first MRI scoring system to assess individual HOA features, including both structural and inflammatory features. High reliability was found for most of the features, including osteophytes (as presented in this study). Furthermore, there are several radiographic scoring methods for HOA, such as the Kellgren–Lawrence grading scale40 ,41 and the OARSI atlas,24 which have all demonstrated high reliability and sensitivity to change.42 ,43 There is no similar agreement for ultrasound examination of HOA, but a group of experts in ultrasound assessment and osteoarthritis proposed a semiquantitative scoring system to include an evaluation of osteophytes, grey scale synovitis and power Doppler signals.10 A comprehensive ultrasound atlas of synovitis and power Doppler scoring is published,18 and in combination with the present ultrasound atlas of osteophytes, the two atlases will provide sonographers with tools for standardised ultrasound assessment of HOA.
Other relevant features in HOA, such as erosions, joint space narrowing and cartilage thickness, were not included in the proposed scoring system due to concerns about reliable definitions, acquisition and current available ultrasound technology.10 Furthermore, to make a feasible examination of each joint, the proximal and distal osteophytes were assessed together, because there are no studies indicating the importance of detailed examination on the location of osteophytes. However, this topic should be examined in future studies.
Definition of the gold standard is commonly a challenge in validation studies. The major strengths of the present study are the high number of patients examined with several imaging modalities and clinical examination, with readings of images performed independently and blinded for results, as well as reliability testing of ultrasound assessments of osteophytes. However, a limitation could be that the ultrasound examinations were only performed on the dorsal side of each finger joint. This may have underestimated the number of joints with ultrasonographic osteophytes, but an MRI study showed that osteophytes were present predominantly on the dorsal proximal side of the joint.44 Therefore, the current finding of high concordance between MRI and ultrasound suggests that it may be sufficient to perform ultrasound only on the dorsal side of the finger joints. Finally, the high female/male ratio may limit the generalisability of the findings.
In summary, examination of osteophytes in finger joints of a large HOA cohort demonstrated good agreement between ultrasound and MRI, while ultrasound was more sensitive than CR and clinical examination. The present study introduced a new ultrasound atlas of osteophyte scores (see figure 1 and supplementary figure S1, available online only), and the reliability exercise using this atlas resulted in high accuracy within and between readers. The current results support the use of ultrasound for the detection of osteophytes in patients with HOA and suggest that the ultrasound atlas will be a helpful tool in future HOA studies.
The authors would like to thank the participants of the Oslo Hand Osteoarthritis cohort and the staff at the Department of Rheumatology at Diakonhjemmet Hospital in Oslo, Norway, for helping to perform this study.
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:
- Web Only 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.
Funding Funding was provided by Abbot.
Competing interests None.
Patient consent Obtained.
Ethics approval Ethics approval was received from the Regional Ethical Committee (Norway).
Provenance and peer review Not commissioned; externally peer reviewed.