Knee osteoarthritis (OA) is accompanied by non-destructive synovial proliferation, joint effusions and popliteal cysts.1 2 Several arthroscopy studies have shown that synovitis is common in painful knee OA.3^–5 Plain x rays play an important role in the diagnosis and follow-up analysis of OA, while the relationship between radiograpic findings and patients’ symptoms remain unclear.

Moreover, alterations appear late, and signs of synovial disease are indirect.6 So, the ability to assess the degree of inflammation (synovitis) by plain x ray is very limited.

MRI reveals the entire spectrum of OA-related abnormalities in the knee joint and allows the assessment of soft tissue structures, cartilage and bone lesions.1 7^–10 Musculoskeletal ultrasonography (MUS) has been shown to depict early changes of synovium and articular cartilage in patients with joint diseases.11^–13 Quantitative MRI- and ultrasound (US) measurements of effusion, synovial membrane and cartilage thickness were correlated in knee OA.14

Recently, it could be shown that inflammatory processes can be measured by contrast medium-enhanced-MUS (CE-MUS) in combination with Power Doppler (PD)15^–20 and contrast medium-enhanced MRI (CE-MRI).19 21 Several working groups have used these methods in different settings and different purposes in arthritis or OA.

Other studies were not complete in regards to CE-MUS with PD and comparison with MRI: in some studies, the knee or finger joints rheumatoid arthritis patients were investigated without using MRI as a reference standard.15^–18 22 Another study used MRI in patients with knee OA without using PD or CE-MUS.23 Multiple joints of patients with different rheumatic diseases have been investigated using Levovist and MRI, but MRI examination was only performed in some patients.19

To the best of our knowledge, no study so far has investigated the efficacy of CE-MUS as a new method of digital synovial vascularisation quantification and PDS in detecting synovial processes in knee OA compared with CE-MRI. Further objectives of our study were the diagnostic value of B-mode and PDS compared with CE-US with regard to assessing synovial processed in knee OA. Furthermore, a new second-class ultrasound contrast medium sulfur hexafluoride (SonoVue®, Bracco Diagnostics, Netherlands) has been used.

PATIENTS AND METHODS

Patients

We evaluated 41 patients (26 female) with a mean age of 65 years (SD 54–78) between February and May 2004. All patients had a knee OA of Kellgren & Lawrence24 grade II or III, and the diagnosis of knee OA had to be confirmed at least 6 months prior to study start based on the 1986 ACR diagnostic criteria.25 The affected mean knee joint pain under rest and during activity as assessed by the visual analogue scale (0–100) had to be ⩾40. In addition, a control group of 6 healthy subjects (2 women) with a mean age of 36 years (30–46) was recruited.

The severity of knee joint pain was calculated as an average of pain under rest and during activity and scored according to the cutoffs, which can be seen in table 1. The study was approved by the local Ethics committee. Informed consent was obtained from all patients admitted to the study.

Musculoskeletal sonography

All 41 patients underwent CE-MUS. MUS was performed by experienced rheumatologists according to the EULAR recommendations26 by using a 12–5-MHz linear transducer (ESAOTE, Technos MPX). For CE-MUS, an 8–4-MHz linear transducer was applied. Interpretation of ultrasound images was done in consensus reading by two experienced rheumatologists (MB, AS).

The second generation contrast agent SonoVue® was applied intravenously. Contrast medium enhancement was assessed for the superior recess and then evaluated by the so-called time/intensity analysis (see below).

In MUS, effusion is characterised by an anechoic or hypoechoic area of the suprapatellar pouch which is displaceable and characterised by a lack of PDS signal.27 Synovial hypertrophy was defined as an hypoechoic area which is not displaceable and poorly compressible.27

Effusion and synovial hypertrophy were graded according to the cutoffs which can be seen in table 1. Each knee was evaluated with (PDS) with the following PD setting parameters: Doppler frequency 6.3 MHz, pulse repetition frequency 500 Hz, gain 103 and wall filter low. PDS was graded on a semiquantitative scale from 0 to 3 (0: no PDS; 1: mild PDS, single color marked vessel; 2: medium PDS; 3: strong PDS, large confluent marked vessel >50% of examined area). The same was done for contrast medium enhancement measured as slope values (0: no contrast medium enhancement; 1: mild; 2: medium; 3: strong contrast medium enhancement).

In addition, the presence of synovitis and degree of effusion in the lateral knee recess were measured. The examination of the lateral knee recess was done with parallel scanning of the lateral knee recess in flexed knee position (with less than 90° flexion).

Time/intensity analysis

All MUS examinations of the knee joint were obtained by a dedicated technology (CnTI™, ESAOTE) operating at a low mechanical index (MI = 0.089–0.118; most examinations have been performed with an MI of 0.098), after an intravenous bolus injection of 4.8 ml of SonoVue®. The time/intensity analysis has already been used in different indications.28^–30 For more details and an illustration of time/intensity analysis, please refer to fig 1.

Magnetic resonance imaging

Thirty-six out of 41 patients and all 6 healthy volunteers underwent magnetic resonance imaging (MRI). Five patients did not undergo MRI examination due to a high body mass index, anxiety or technical problems. Assessment of the knee joint by MR imaging was performed on a permanent magnet 0.2 T C-Scan imager (Esaote Bioimaging, Genoa, Italy), using a dedicated dual phased array coil. The patients underwent MRI on the same day as the ultrasound examination. The MR imaging protocol comprised a sagittal short tau inversion recovery sequence, a coronal T1-weighted spin echo sequence, an axial T2-weighted turbo spin echo sequence and a sagittal T2-weighted multi echo (spinecho and proton density) sequence, all using a slice thickness of 5.0 mm. This was followed by a three-dimensional T1-weighted gradient-recalled echo (GRE) sequence in sagittal section orientation with 0.8 mm slice thickness followed by a dynamic series of 13 fast sagittal T1-weighted spin echo sequences containing 3 slices each with a slice thickness of 8 mm (duration per sequence 9 s). After the first dynamic sequence, an intravenous injection of gadolinium diethylenetriaminepentaacetic acid (Gd-DTPA; Magnevist, Schering AG, Berlin, Germany) at a dosage of 0.2 mmol/kg body weight was administered.31 These fast dynamic sequences were followed by a second T1-weighted sagittal GRE sequence with 0.8-mm slice thickness. The postcontrast GRE sequences were used for three-dimensional reconstruction in transverse and coronal slice orientations.

In the MRI, the joint effusion in the superior and lateral recess and contrast medium enhancement (slope measurement) were evaluated.21 32 33

The MRI studies were performed and interpreted by two radiologists experienced in musculoskeletal MRI imaging in consensus reading.

Joint effusion was defined as a high signal intensity on T2 sequences and low signal intensity on T1-weighted sequences that is not enhanced after contrast media injection. The extent of an effusion was assessed on the axial and sagittal T2 weighted axial images, as well as in the postcontrast T1 images by measuring the maximum anterior–posterior diameter of the suprapatellar recess in the sagittal images and the maximum side-to-side diameter of the lateral recess in the axial images.

Effusion and contrast medium enhancement were scored and graded according to the cutoffs, which can be seen in table 1.

Statistics

Non-parametric statistical methods using the statistical package SPSS version 12.0 were applied with a level of significance of 0.05 (Mann–Whitney test). The correlation between the change in sonographic and MRI findings and clinical data (visual analogue scale (VAS)) of knee OA patients was analysed by Spearman’s rank correlation test (r_s). A kappa analysis was performed.

RESULTS

US and MRI parameters

The mean values including SD, the grading of US and MRI parameters and the cutoffs can be seen in table 1. Examples for US and MRI images are shown in fig 2. Parameters which are not shown in table 1 are as follows.

US parameters

The mean value for effusion in the lateral recess was 5.1 mm (SD 3.0). Taking a cutoff of ⩾2 mm, an effusion in the lateral recess was found in 78.1% of patients. The values in the lateral recess were not graded. The mean value for effusion in the lateral recess in MRI was 5.6 mm (SD 3.6). Taking a cutoff of ⩾2 mm, an effusion in the lateral recess was observed 81%.

MRI parameters

Contrast enhancement in MRI was evaluated in 33 out of 36 patients. In 3 patients, assessment of MRI contrast enhancement was not possible due to technical problems and movement artefacts of the dynamic sequences regarding the MRI. Notably, US slope values and MRI slope values cannot be compared with each other directly due to different scaling.

Comparison between patients and healthy controls

The 6 healthy controls showed no pain under rest or during activity according to VAS and were tested to be grade 0 for all mentioned US examinations. There was a highly significant difference between the 6 healthy controls and the 41 patients in regards of all VAS parameters (p = 0.000), effusion in the superior and lateral recess in US (p = 0.003 and 0.001, respectively), effusion in MRI in the superior and lateral recess (p = 0.003 and 0.002, respectively), CE in US and MRI (p = 0.000 and 0.001, respectively). Also, in regards of synovial hypertrophy in the superior and lateral recess as measured by US, there was a significant difference between the 6 healthy controls and the 41 patients (p = 0.012 and 0.031, respectively).

Clinical assessment of patients’ data for VAS with other parameters of disease activity

VAS data correlated moderately with CE-MUS (r = 0.386; p = 0.018) as well as with PD in the superior recess (r = 0.366; p = 0.020). Otherwise, data acquired by the VAS did not correlate with other parameters like effusion in the superior or lateral recess, contrast medium enhancement in MRI and synovial hypertrophy in US. Further correlation analysis revealed that contrast medium enhancement between US and MRI showed a good correlation (r = 0.634; p<0.001).

Relationship between synovial hypertrophy and effusion in the superior and lateral recess found by US

About 10% of patients (4/41) had no significant effusion in the superior recess but effusion in the lateral recess (mean 4.3 mm, SD 1.3), while 2 out of 41 patients had effusion in the superior recess (mean 5.8 mm, SD 0.2) without having effusion in the lateral recess. In 17.1% of patients (7/41), there was no synovial hypertrophy in the superior recess but there was synovial hypertrophy in the lateral recess (mean 3.1 mm, SD 1.8). Conversely, 11 patients had synovial hypertrophy in the superior recess (mean 3.4 mm, SD 1.3) without synovial hypertrophy in the lateral recess.

The correlation between joint effusion and synovial hypertrophy in the superior recess was 0.431 (p = 0.005), and the correlation between joint effusion and synovial hypertrophy in the lateral recess was 0.415 (p = 0.01).

Kappa analysis

Kappa analysis showed a value of 0.481 between US and MRI with regard to the effusion in the superior recess. Thirty-six cases could be assessed with regard to detecting effusion in the superior recess by US and MRI at the same time.

Kappa analysis of other parameters showed no significant findings between US and MRI for the lateral recess or contrast medium enhancement.

Kappa analysis furthermore showed a value of 0.53 between effusion in the superior recess and PD signal in the superior recess without any other significant findings for the other parameters.

Sensitivity between US and MRI

If MRI was taken as the reference standard, there was a sensitivity of 72% for assessing effusion in the superior recess and 81% for assessing effusion in the lateral recess.

Using the cutoffs which have been explained above and in table 1, US and MRI showed positive findings in B-mode, PDS, CE-US and CE-MRI, as can be seen in table 2.

DISCUSSION

This study evaluated a new method of digital synovial vascularisation quantification by using contrast-enhanced MUS in the detection of synovitis in patients with knee osteoarthritis in comparison with contrast-enhanced MRI and healthy subjects.

Comparison between patients and healthy controls

As shown by significant differences between healthy controls and our 41 knee OA patients for all US and MRI parameters, we assessed significant findings by US and MRI.

VAS data

Our findings that knee OA pain was significantly higher during activity than under rest and that pain data acquired on VAS hardly correlated with other US parameters are consistent with the findings by Tarhan and Unlu,23 who found similar VAS data in 58 patients with symptomatic knee OA and by D’Agostino et al,34 who assessed the prevalence of inflammation in 600 patients with painful knee OA and found that in at least half of the subjects, US inflammation was not detected, and no correlation was found between US inflammatory signs and pain intensity during recent physical activity.

A possible explanation for this could be that considerable effusion in the superior or lateral recess might to some extent act as a buffer and so lead to decreased pain sensation in the patients. Thus, assessment of synovitis by the VAS alone is not sufficient. There are new data showing that although it was possible to determine some clinical and radiological predictors of OA inflammation, it was not possible to determine useful clinical rules for an individual patient. These results suggest that imaging (ultrasound or magnetic resonance imaging) remains the most useful tool for the detection of synovial inflammation.34 35

US B-mode findings: Effusion and synovial hypertrophy in the superior and lateral recess

The mean value for effusion in the superior recess of 6.1 mm is higher than that found by others.34 Synovial hypertrophy in the superior and lateral recess was found in 59%, which is similar to the results by Walther et al,22 who found 46% of patients with OA had hypertrophic synovial tissue by US examination.

The mean synovial thickness in the superior recess of 1.9 mm is similar to that found in other studies (2.1 mm).34 Synovial thickness might be used as an objective method to monitor treatment response in inflammatory knee joint diseases.11

As shown, about 10% of our patients had significant effusion in the lateral recess without having effusion in the superior recess. This emphasises the importance of assessing effusion and synovial hypertrophy in the lateral knee recess, even in the absence of findings in the superior recess.

PD sonography (PDS) findings

PDS in our study revealed positive results in 63% of patients. PDS was able to detect inflammatory processes somewhat more often than B-mode ultrasonography. The usefulness of PD in evaluating the degree of inflammation in knees of patients with rheumatoid arthritis (RA) has been shown in RA patients.15 16 PDS is known to be sensitive to change and a reliable diagnostic method for qualitative grading of the vascularity of the synovial tissue.22

MRI findings on effusion

MRI detected effusion in the superior recess in 61%. Hill et al1 stated that in patients with symptomatic knee OA, effusions were highly prevalent on MRI, with 37% having a small effusion, which is very similar to what we found in our study (grade 1 effusion in our study in 36% of cases). Moderate effusions were seen by Hill et al1 in 36%, in contrast to 22% in our study. Fernandez-Madrid et al23 demonstrated that joint fluid was detected in 60% of OA patients on MRI.10 Tarhan and Unlu23 found 85% knee joint effusion on MRI.

Contrast-medium enhancement in US and MRI

Using the contrast-medium SonoVue® in US for the superior recess, 95% of patients showed positive results in contrast to 82% of patients in MRI. CE-US was not performed for the lateral recess in this study. There are studies suggesting that it is possible to split the total amount of 4.8 ml of SonoVue® into smaller dosages,36 which allows a valid assessment of several joint regions in one investigation. Further studies are needed to assess the diagnostic value of CE-US in assessing the lateral recess. Other authors evaluated the usefulness of CE-PDS in patients with rheumatic diseases, mostly using the contrast medium Levovist.15^–19

Kappa analysis/comparison between US and MRI

Kappa analysis showed a value of 0.481 between US and MRI with regard to effusion in the superior recess. Taking the 36 patients with knee OA who had all the US and MRI examinations, there was an agreement in 27 cases and a disagreement in 9 cases. Effusion in the superior recess was detected in 58% by US and in 61% by MRI. Effusion in the lateral recess was found in 78% by US and in 83% by MRI. Thus, MRI was more sensitive in detecting effusion in the lateral recess compared with US.

Sensitivity of imaging techniques

As shown in table 2 objective inflammatory findings could be found by US as well as MRI. PDS showed positive results in 63%, and so was more sensitive than B-mode techniques (effusion in superior recess in B-mode in 58%, synovial hypertrophy in 59%). By using the contrast medium SonoVue® and time/intensity analysis, pathological findings were observed by us in 95%. Thus, contrast-enhanced US was more sensitive than PDS. The use of an intravenous contrast medium may increase sensitivity in the evaluation of the synovial vascularisation. Minimal increments in synovial perfusion could be detected by CE-MUS. Time/intensity analysis by US proved to be a valid method to assess inflammatory processes in OA of the knee. CE-MUS was more sensitive than CE-MRI in analysing the synovial vascularisation in knee OA. MRI and US showed a similar efficacy in detecting effusion in the superior or lateral recess (see table 2), irrespective of whether contrast medium was used or not.

In 6 patients, CE-US showed positive results, while CE-MRI showed no signs for inflammation (grade 0). Four of these 6 patients had grade 1 findings in CE-US, and 2 patients had grade 2 findings in CE-US. As CE-MRI was used as the reference method, the CE-US findings in these 6 patients could be considered falsely positive. However, those 2 patients with grade 2 findings in CE-US also had positive findings in PD of the superior (grade 1 each) and lateral recess (grade 2 each) as well as positive findings in effusion in the superior recess in US (grade 1 or 2) and MRI (grade 2 and 3). In contrast, those 4 patients with grade 1 findings in CE-US had grade 0 in effusion in the superior recess (in both US and MRI) and PDS (superior and lateral recess). Thus, grade 1 findings in CE-US could be considered “borderline” findings. Another question is whether CE-MRI as the reference method has limited diagnostic sensitivity.

Furthermore, it has to be kept in mind that there were limitations of MRI in 5 patients who were not able to undergo MRI examination due to obesity or medical reasons.

Performing CE-US with SonoVue® instead of standard grey-scale US and Power Doppler US caused additional costs of €65 and took about 5–10 min longer per joint examination, which is acceptable in our opinion, given the greatly increased sensitivity of CE-US compared with standard grey-scale US. There are studies suggesting that a lower amount of SonoVue® might be sufficient to assess synovitis in rheumatoid arthritis,36 though this has not yet been fully evaluated. In regards of side effects and contraindications such as severe cardiovascular disease (NYHA III or IV), patients certainly have to be selected carefully. In our patients, however, no adverse events could be observed. Furthermore, CE-US with SonoVue® has also been used safely for examining patients with rheumatoid arthritis.37 38 Moreover, a retrospective analysis in over 20 000 investigations has shown that SonoVue® is safe.39

In conclusion, for the first time contrast-enhanced MUS and magnetic resonance imaging have been used to assess the synovitis process in patients with knee OA. US and MRI showed a good correlation in measuring effusion and contrast medium enhancement. This may serve as a model for the assessment of other inflammatory diseases, such as rheumatoid arthritis, and also for testing new antirheumatic drugs in the future.