Article Text

Download PDFPDF

Extended report
Presence of MRI-detected joint effusion and synovitis increases the risk of cartilage loss in knees without osteoarthritis at 30-month follow-up: the MOST study
  1. Frank W Roemer1,2,
  2. Ali Guermazi1,
  3. David T Felson3,
  4. Jingbo Niu3,
  5. Michael C Nevitt4,
  6. Michel D Crema1,5,6,
  7. John A Lynch4,
  8. Cora E Lewis7,
  9. James Torner8,
  10. Yuqing Zhang3
  1. 1Department of Radiology, Boston University Medical Center, Boston, Massachusetts, USA
  2. 2Department of Radiology, Klinikum Augsburg, Augsburg, Germany
  3. 3Clinical Epidemiology Research and Training Unit, Boston University School of Medicine, Boston, Massachusetts, USA
  4. 4Department of Epidemiology and Biostatistics, University of California, San Francisco, California, USA
  5. 5Institute of Diagnostic Imaging (IDI), Ribeirão Preto, Brazil
  6. 6Division of Radiology, University of São Paulo, Ribeirão Preto, Brazil
  7. 7Division of Preventive Medicine, University of Alabama, Birmingham, Alabama, USA
  8. 8Department of Epidemiology, University of Iowa, Iowa City, Iowa, USA
  1. Correspondence to Dr Frank Roemer, Department of Radiology, Boston University Medical Center, FGH Building, 3rd floor, 820 Harrison Avenue, Boston, MA 02118, USA; froemer{at}


Objective To evaluate if two different measures of synovial activation, baseline Hoffa synovitis and effusion synovitis, assessed by MRI, predict cartilage loss in the tibiofemoral joint at 30 months follow-up in subjects with neither cartilage damage nor tibiofemoral radiographic osteoarthritis of the knee.

Methods Non-contrast-enhanced MRI was performed using proton density-weighted fat-suppressed sequences in the axial and sagittal planes and a short tau inversion recovery sequence in the coronal plane. Hoffa synovitis, effusion synovitis and cartilage status were assessed semiquantitatively according to the WORMS scoring system. Included were knees that had neither radiographic osteoarthritis nor MRI-detected tibiofemoral cartilage damage at the baseline visit. The presence of Hoffa synovitis was defined as any grade ≥2 (range 0–3) and effusion synovitis as any grade ≥2 (range 0–3). Logistic regression was performed to examine the relation of the presence of either measure to the risk of cartilage loss at 30 months adjusting for other potential confounders.

Results Of 514 knees included in the analysis, the prevalence of Hoffa synovitis and effusion synovitis at the baseline visit was 8.4% and 10.3%, respectively. In the multivariable analysis, baseline effusion synovitis was associated with an increased risk of cartilage loss. No such association was observed for baseline Hoffa synovitis.

Conclusions Baseline effusion synovitis, but not Hoffa synovitis, predicted cartilage loss. The findings suggest that effusion synovitis, a reflection of inflammatory activity including joint effusion and synovitic thickening, may play a role in the future development of cartilage lesions in knees without osteoarthritis.

Statistics from

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.

During recent years it has been increasingly accepted that osteoarthritis is a complex disease involving multiple joint tissues including the synovium, the subchondral and cortical bone, the menisci, the ligaments and cartilage, leading eventually to joint failure.1,,4 Synovitis is an important feature in the disease process and is defined as inflammation of the synovium. It may manifest itself phenotypically as thickening of the synovial membrane or indirectly as joint effusion as the result of synovial activation.5 6 The exact pathophysiological mechanisms that lead to synovial inflammation remain largely unknown, but activated synovial macrophages seem to play a key role in the inflammatory processes leading to synovial activation.7 It has been suggested that synovitis in osteoarthritis is most likely a secondary phenomenon due to phagocytosis of intra-articular debris, the release of soluble cartilage matrix macromolecules, the presence of calcium pyrophosphate dehydrate, or calcium hydroxyapatite crystals.8,,10 It is accepted that synovitis and effusion are frequently present in osteoarthritis and correlate with pain and other clinical outcomes.11 12

It has also been suggested that synovitis in osteoarthritis predisposes to further structural progression.13 14 One study using arthroscopy as a reference standard demonstrated a positive correlation between the severity of synovitis and the degree of progression of cartilage lesions over time, particularly in the medial patellofemoral compartment.15

Synovial inflammation already occurs in the early stages of the osteoarthritis process.16,,19 However, as multiple tissues are involved in the disease initiation process it remains unclear if synovitis and effusion have to be regarded as secondary collateral phenomena or if these inflammatory processes play an important independent role in disease initiation and later progression. To overcome this dilemma it seems appealing to investigate joints without macroscopic cartilage damage before the occurrence of apparent radiographic features of osteoarthritis.17

Both synovial thickening and joint effusion may be assessed non-invasively by MRI.4 6 13 20 Ideally, synovitis in osteoarthritis is evaluated using contrast-enhanced imaging after the intravenous administration of gadolinium-based contrast agents.6 12 21 However, due to possible side effects and associated costs, contrast-enhanced MRI is usually not part of large epidemiological or clinical osteoarthritis studies. Instead, current methods to assess synovial activation on non-enhanced MRI in large studies use a surrogate of signal changes in Hoffa's fat pad (‘Hoffa synovitis’) as it had been shown in a histological correlation study that these signal alterations represent mild chronic synovitis, which seems to be a sensitive, albeit non-specific measure when using contrast-enhanced MRI as the reference.13 18 22 Furthermore, intra-articular fluid-equivalent signal is evaluated according to the extent of capsular distension.4 20 Recent evidence showed that this measure reflects a composite of true joint effusion (ie, joint fluid) and synovial thickening (‘effusion synovitis’).21 These MRI-based semiquantitative measures of synovitis are associated with pain severity, and similarly a change in synovitis and effusion is associated with a change in pain severity in both directions.12 13 23 High-grade synovitis assessed on non-enhanced MRI seems to play a role in structural deterioration.13

In order to analyse the possible link between early synovial activation and later structural deterioration we wished to assess if knees without tibiofemoral radiographic osteoarthritis and without cartilage damage but with evidence of MRI-detected Hoffa synovitis or effusion synovitis (composite score of joint fluid and synovial thickening) are at an increased risk of cartilage loss at 30 months follow-up when compared with knees without these baseline features of synovial activation.

Patients and methods

Study design and subjects

Subjects were participants in the Multicenter Osteoarthritis (MOST) Study, a prospective epidemiological study of 3026 persons aged 50–79 years, with a goal of identifying risk factors for incident and progressive knee osteoarthritis in a sample either with osteoarthritis or at high risk of developing disease. Participants at high risk included those who were overweight or obese, had knee pain, aching or stiffness on most of the last 30 days, or had a history of knee injury that made it difficult to walk for at least 1 week, or had had previous knee surgery.

Subjects were recruited from two US communities, Birmingham, Alabama and Iowa City, Iowa, through mass mailing of letters and study brochures, supplemented by media and community outreach campaigns. The study protocol was approved by the institutional review boards at the University of Iowa, University of Alabama, Birmingham, University of California, San Francisco and Boston University Medical Campus.

Subjects were excluded from MOST if they screened positive for rheumatoid arthritis,24 had ankylosing spondylitis, psoriatic arthritis, Reiter's syndrome, had renal insufficiency that required haemo or peritoneal dialysis, a history of cancer (except for non-melanoma skin cancer), had or planned to have bilateral knee replacement surgery, were unable to walk without assistance, or were planning to move out of the area in the next 3 years.25 Subjects were further excluded if a diagnosis of rheumatoid arthritis or related diseases was established between the baseline and the follow-up visit. Figure 1 gives a detailed overview of the inclusion process for the present analysis.

Figure 1

Flowchart of knee inclusion. ACL, anterior cruciate ligament; KL, Kellgren–Lawrence; OA, osteoarthritis; TF, tibiofemoral.


At baseline, all subjects underwent weight-bearing posteroanterior fixed flexion knee radiographs using the protocol by Peterfy et al26 and a plexiglass positioning frame (Synaflexer, Synarc, San Francisco, California, USA). A musculoskeletal radiologist and a rheumatologist experienced in reading study films, both blinded to case/control status and clinical data, graded all posteroanterior films according to the Kellgren–Lawrence (K/L) scale. Radiographic tibiofemoral osteoarthritis was considered present if the K/L grade was 2 or greater. If readers disagreed on the presence of radiographic osteoarthritis, the film readings were adjudicated by a panel of three readers.

At the baseline clinic visit long-limb films were acquired with a 14-in × 51-in cassette. Mechanical alignment was measured as the angle formed by the intersection of the femoral and tibial mechanical axes. The femoral mechanical axis is the line from the centre of the femoral head through the centre of the knee, and the tibial mechanical axis is drawn as a line from the centre of the ankle to the centre of the knee. Neutral alignment was defined as 179–181°, varus malalignment as 178° or less and valgus malalignment as 182° or greater.

MRI acquisition

MRI were obtained in both knees with a 1.0T dedicated MR system (OrthOne; ONI Medical Systems, Inc., Wilmington, Massachusetts, USA) with a circumferential extremity coil using fat-suppressed fast spin echo PDw sequences in two planes, sagittal (TR 4800 ms, TE 35 ms, 3 mm slice thickness, 0 mm interslice gap, 32 slices, 288×192 matrix, two excitations (NEX), 140 mm2 field of view (FOV), echo train length (ETL) 8) and axial (TR 4680 ms, TE 13 ms, 3 mm slice thickness, 0 mm interslice gap, 20 slices, 288×192 matrix, 2 NEX, 140 mm2 FOV, ETL 8), and a short TI inversion recovery sequence in the coronal plane (TR 6650 ms, TE 15 ms, TI 100 ms, 3 mm slice thickness, 0 mm interslice gap, 28 slices, 256×192 matrix, 2 NEX, 140 mm2 FOV, ETL 8).

MRI interpretation

Two musculoskeletal radiologists, blinded to clinical data, read bone marrow lesions (BML) and cartilage status according to the whole-organ MRI score (WORMS) method.4 Baseline and follow-up MRI were read paired and with the chronological order known to the readers. MRI features were assessed simultaneously.

Signal alterations in the infrapatellar and intercondylar regions of Hoffa's fat pad were scored from 0 to 3 as a surrogate for synovial thickening according to the literature as this feature is not part of the original WORMS system.11 13 18 We will refer to these scores as ‘Hoffa synovitis’ in the following sections, although acknowledging that these signal changes also include non-specific alterations not necessarily related to synovitis22 (figure 2A).

Figure 2

(A) Sagittal proton density-weighted magnetic resonance image. Grade 2 Hoffa synovitis is shown in the infrapatellar (arrow) and intercondylar (arrowhead) regions of Hoffa's fat pad. (B) Axial proton-density weighted magnetic resonance image. Grade 2 joint effusion synovitis is depicted as fluid-equivalent signal within the joint cavity (black arrowheads). Marked capsular distension is present. Note incidental finding of a metaphyseal enchondroma in the distal femur (white arrowheads).

WORMS uses a combined measure of joint effusion and synovitis based on the amount of intra-articular fluid-equivalent signal. This composite score is graded from 0 to 3 according to the estimated maximal distension of the synovial cavity and was applied in addition to the mentioned signal changes in Hoffa's fat pad (figure 2B).4 We will refer to this scoring measure as ‘effusion synovitis’ in the following sections to acknowledge both constituents of the composite score.

Cartilage signal and morphology were scored according to the WORMS system from 0 to 6 in 14 articular surface regions. In a modification of WORMS developed for longitudinal readings, a score of within-grade worsening for cartilage assessment was introduced to reflect any subtle within-grade progression that did not fulfil the criteria of a full-grade change. Any change including within-grade and full-grade or greater increases in cartilage damage in at least one of 10 tibiofemoral subregions was defined as cartilage loss.2 14 In addition, BML and meniscal damage were assessed according to the WORMS system at baseline. BML size was scored from 0 to 3 based on the extent of regional involvement. Meniscal status was graded from 0 to 4 in the anterior horn, the body segment and the posterior horn of the medial and lateral meniscus. Furthermore, meniscal extrusion of the medial and lateral meniscal body was scored on the coronal plane according to previous publications as this feature is not part of the WORMS score.27 28 The inter-reader reliability (weighted κ) for the readings of the different features was 0.62 for BML, 0.65 for both synovitis and joint effusion, 0.65 for meniscal extrusion, 0.78 for cartilage morphology and 0.80 for meniscal status.

Knees that showed typical radiological signs of traumatic bone contusions, osteonecrosis, fracture or malignant bone infiltration were excluded from the analysis. However, of all analysed MRI only one knee showed a subacute tibial depression fracture at follow-up and was excluded.

Analysis approach

In order to minimise possible misclassification errors of low-grade Hoffa synovitis and effusion synovitis versus normal imaging findings, Hoffa synovitis was defined as any grade 2 or greater and effusion synovitis as any grade 2 or greater (figures 1 and 2). Knees with scores of either 0 or 1 were the reference. We compared the frequencies of cartilage loss in knee plates according to the presence or absence of Hoffa synovitis and effusion synovitis using χ2 statistics. We examined the association between baseline Hoffa synovitis and the risk of tibiofemoral cartilage loss using the logistic regression model. In the multivariable regression model, we adjusted for age, gender, body mass index, knee malalignment and other structural lesions (ie, patellofemoral cartilage damage, meniscus damage, meniscal extrusion, BML and effusion synovitis). We used the same approach to examine the relation of effusion synovitis with the risk of tibiofemoral cartilage loss. In this analysis we also adjusted for Hoffa synovitis. In an additional analysis we included only knees without any cartilage damage in all 14 articular subregions including the patellofemoral joint. Furthermore, we explored whether a lower threshold of ‘definite’ Hoffa and effusion synovitis yielded comparable results (using grades 0 only as the reference). All statistical calculations were performed using SAS software (version 9.1 for Windows).


As shown in table 1, of 514 subjects (514 knees) included in the analysis, the average age at baseline was 60 years and 56% were women. The mean body mass index was 29.1 kg/m2. The majority (92.8%) of knees had a K/L grade of 0, and the remaining knees (7.2%) had a K/L grade of 1. Seventeen per cent of the knees had frequent pain symptoms, and more than 60% knees had either varus or valgus malalignment.

Table 1

Characteristics of study sample (514 subjects, one knee per subject)

Eighty-five (16.5%) subjects recalled a history of injury to the examined knee badly enough to limit walking for at least 2 days and 14 (2.7%) knees had surgery.

At the baseline visit the majority of knees had not or only exhibited low-grade (ie, grade 1) BML, meniscal damage or meniscal extrusion. Slightly more than 9% of knees showed Hoffa synovitis (grade >1) and 10.3% presented with effusion synovitis (grade >1; table 2).

Table 2

Prevalence of specific features on baseline MRI

Concerning anatomical distribution in regard to articular plates, cartilage loss occurred most frequently in the retropatellar subregions (17.5% of knees) and medial weight-bearing femur (15.0% of knees). Only 6.6% of knees exhibited cartilage loss in the lateral weight-bearing femur. Cartilage loss was more frequently observed in the lateral tibia in knees with Hoffa synovitis when compared with knees without signs of Hoffa synovitis. Furthermore, cartilage loss was more frequently observed in the medial and lateral weight-bearing femur in knees with effusion synovitis. The anatomical distribution of cartilage loss is presented in table 3. Of 137 knees with cartilage loss in any of the tibiofemoral subregions, 45 (32.9%) also had tibiofemoral joint space narrowing progression on the x-ray.

Table 3

Proportion of knees with cartilage loss according to articular plates

As shown in table 4, baseline Hoffa synovitis was not associated with an increased risk of cartilage loss at follow-up (adjusted OR 1.0, 95% CI 0.5 to 2.1; p=0.89). However, knees with baseline effusion synovitis had an increased risk of cartilage loss (adjusted OR 2.7, 95% CI 1.4 to 5.1; p=0.002).

Table 4

Longitudinal association between baseline effusion synovitis and cartilage status at 30 months follow-up

We performed an additional analysis by limiting the knees that had no cartilage damage in all 14 articular subregions, including the patellofemoral joint (n=85). Of them, only two knees showed any grade 2 Hoffa synovitis or effusion synovitis, 24 (28.2%) knees showed grade 1 Hoffa synovitis and 16 (18.8%) exhibited grade 1 effusion synovitis at baseline. During the follow-up period, only 13 (15.3%) knees developed any cartilage loss, thus prohibiting us from assessing the association between either Hoffa synovitis or effusion synovitis and the risk of cartilage loss.

Using grade 0 only as the reference group neither baseline Hoffa synovitis nor effusion synovitis was associated with an increased risk of cartilage loss at follow-up (adjusted OR 0.9, 95% CI 0.6 to 1.3; p=0.48 and adjusted OR 1.5, 95% CI 0.9 to 2.3; p=0.11).


In knees without radiographic osteoarthritis we identified baseline joint effusion and synovitis assessed by a composite MRI measure, which we termed ‘effusion synovitis’ as a predictor of future cartilage loss over a 30-month period. The imaging surrogate of signal alterations in Hoffa's fat pad that we termed ‘Hoffa synovitis’ was not predictive of later cartilage loss. We adjusted our results for possible confounders of future cartilage loss and thus could show that baseline effusion synovitis as assessed on non-enhanced MRI is an independent predictor of cartilage deterioration and might play an important role in disease initiation.

Osteoarthritis is usually classified as a non-inflammatory disease as the leucocyte count in osteoarthritis synovial fluid is typically below the threshold defining an ‘inflammatory’ disorder.29 However, synovial activation is commonly observed in osteoarthritis. Clinically, it manifests itself typically as ‘flares’ of pain.30 On imaging and arthroscopy increased synovial thickness, hypervascularisation and proliferation of hypertrophic and hyperaemic synovial villi are observed.6 13 15 18 21 Histology shows hypertrophy and hyperplasia with an increase in the number of synovial lining cells and infiltration of mononuclear cells, with activated T cells, B cells and macrophages being the major components of the synovial infiltrate.19 31 The mediators affecting the activation of synovial macrophages have not been clearly identified, but are thought to to be caused by breakdown products of cartilage or meniscal degradation.7 8 Our results neither support nor contradict this hypothesis. Although we have analysed knees without any MRI-detectable cartilage damage, early cartilage degradation before being visually apparent on MRI may have been present. The synovial activation we have observed in the form of synovitis and effusion might thus have been triggered by these breakdown products. On the other hand, synovitis and effusion might have direct effects on cartilage matrix integrity that triggers further cartilage breakdown including surface damage.7 Also stretching of the joint capsule as a result of effusion with consequent joint laxity leading eventually to microtrauma and possibly cartilage damage might play a role.

The aim of our study was not to analyse the cause of synovial activation itself, but rather to assess whether synovial activation, once present, is an independent risk factor for later progression, which clearly seems to be the case. This supports findings of an earlier analysis from the MOST Study in which synovitis and effusion predicted rapid cartilage loss in knees with only minor cartilage damage (albeit statistically only borderline significant).14

We observed that baseline Hoffa synovitis did not predict progression, but effusion synovitis did. These results have to be interpreted carefully in light of the manifestations of synovitis and effusion on non-enhanced MRI that are commonly employed in large osteoarthritis studies such as the MOST Study. Signal changes assessed in Hoffa's fat pad have been used for a long time to evaluate synovitis on non-enhanced MRI and have shown clinically relevant associations.11 13 18 23 We have previously shown that these signal changes are a sensitive, but non-specific, measure using contrast-enhanced MRI as a reference, which might result in over-interpretation of these MRI findings towards inflammatory activity, although a non-inflammatory cause might be the reason for these MRI findings.22 A recent comparative study assessing joint effusion on non-enhanced and contrast-enhanced MRI showed that the commonly applied measure of water-isointense signal within the joint cavity and the amount of capsular distension is a weak measure of true joint effusion. In fact, hyperintensity within the joint cavity has to be interpreted as a composite of synovial thickening and joint effusion, and thus the measurement is truly a reflection of both.21 Consequently, there might be reason to assume that the composite measurement of ‘effusion synovitis’ is potentially a better marker of synovial activation on non-enhanced MRI than the surrogate and non-specific measure of signal changes in Hoffa's fat pad. Ideally, our findings need to be verified in a large study employing contrast-enhanced MRI in order to resolve the question of whether synovitis or joint effusion, or both, predict structural progression. In a histological correlation study, the degree of synovial thickening on contrast-enhanced MRI correlated with qualitative macroscopic analysis and with microscopic features such as surface fibrin deposition, fibrosis, oedema, congestion and infiltration.32 Using contrast-enhanced imaging for assessment, MRI-detected synovitis seems to be present to a relevant extent in osteoarthritic knees despite the absence of joint effusion.21 The fact that we did not find positive associations for the alternative definition of the reference group using only grade 0 as the reference confirms the difficulty of defining a threshold between ‘normal’ or ‘physiological’ and ‘abnormal’ or ‘pathological’ using semiquantitative (and also quantitative) approaches. For this reason, it seems appropriate to use a more conservative definition for disease or pathology as we have in the present study, which is also in analogy to radiography-based studies that define K/L grades 0 and 1 as knees without radiographic osteoarthritis.25 33 34

One possible shortcoming of our study is the fact that we employed 1.0T extremity MRI, which has been questioned to yield inferior image quality when compared with 1.5T or 3T large-bore systems. These issues, to the extent they exist, seem not to affect the semiquantitative scoring of knee osteoarthritis. In a comparative exercise scoring knees of subjects that had received a 1.0T extremity MRI scan and a 1.5T large-bore examination of the same knee on the same day, we could show good agreement, sensitivity and specificity for all assessed features.35 We believe that the slice thickness of 3 mm employed in MOST MRI was adequate in order to detect discrete focal defects but also more widespread cartilage damage. In fact, standard intermediate-weighted sequences seem to be superiorly suited to pick up small focal cartilage defects when compared with high-resolution three-dimensional sequences such as FLASH (fast low angle shot), SPGR (spoiled gradient recalled), dual echo at steady state or similar, as TSE (turbo spin echo) sequences offer superior contrast between joint fluid and cartilage surface.36,,38

As one of the main aims of our study was to assess the change in cartilage over time, the MRIs were read not blinded to the time point. This might result in a slight tendency to read more change in comparison with a blinded reading. However, it has been shown that scoring without knowing the chronological sequence substantially decreases sensitivity in the detection of clinically relevant changes in comparison with scoring in chronological order, and that it does not introduce false-positive changes.39 40 These studies showed that blinding to the time point can lead to misclassification of the longitudinal change in a feature and that it may compromise the assessment of the relation of that feature and its outcome.41 However, we have to acknowledge that to date longitudinal osteoarthritis studies comparing semiquantitative MRI assessment blinded and non-blinded to chronological order are missing.

As there were only a few knees without any cartilage damage in both the tibiofemoral and patellofemoral joints, we included knees without tibiofemoral cartilage damage and without or with patellofemoral cartilage lesions. Adjustment was performed for the presence of patellofemoral cartilage damage as a possible confounder of subsequent cartilage loss.

We did not analyse clinical symptoms or functional parameters at baseline or follow-up as this was not the focus of our study and would have gone beyond the scope of this work. However, several studies have shown that synovitis seems to be strongly associated with pain and is clinically relevant.12 13 23 Subjects who developed an inflammatory disorder such as rheumatoid arthritis or a related disease between baseline and the follow-up visit were excluded from the analysis.

Despite having analysed knees without cartilage damage, the concomitant presence of subchondral BML and meniscal damage in approximately 20–30% of the analysed knees suggests that early features of osteoarthritis are apparent before the manifestation of cartilage loss.42

In summary, we found that baseline joint effusion and synovitis assessed on non-enhanced MRI in knees without tibiofemoral radiographic osteoarthritis but at risk of developing osteoarthritis are predictors of later cartilage loss. The two MRI measures applied to non-contrast-enhanced images showed different results and the composite measure of effusion and synovitis seemed to be predictive, whereas Hoffa's signal changes were not. This leads to the assumption that joint effusion might play an additional role in predicting cartilage loss. Contrast-enhanced MRI needs to verify these findings in the future. Our results support the importance of further research in the field of the targeted inhibition of synovial activation in early osteoarthritis in order eventually to avoid progressive cartilage degradation and functional impairment.


The authors would like to thank the participants and staff of the MOST Study at the clinical sites in Birmingham, Alabama and Iowa City, Iowa and at the Coordinating Center at UCSF, San Francisco, California.



  • Funding This study was supported by NIH grants from the National Institute of Aging to CEL (U01-AG-18947), JT (U01-AG-18832), MCN (U01-AG-19069) and DTF (U01-AG-18820) and NIH AR47785.

  • Competing interests AG has received consultancies, speaking fees and/or honoraria (less than US$10 000 each) from Facet Solutions, Genzyme and Stryker, and (more than US$10 000) from Merck Serono, and is the President of Boston Imaging Core Lab (BICL). He receives research grants from General Electric Healthcare. FWR is vice president and a shareholder of BICL. MDC is a shareholder of BICL. None of the other authors have declared any possible conflict of interest.

  • Patient consent Obtained.

  • Ethics approval This study was conducted with the approval of the University of Iowa, University of Alabama, Birmingham, University of California, San Francisco and Boston University Medical Campus.

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