Reliability of haemophilia early arthropathy detection with ultrasound (HEAD-US) in children: a comparative magnetic resonance imaging (MRI) study
Article Category: Research Article
Publicado en línea: 20 oct 2022
Páginas: 471 - 478
Recibido: 23 ago 2022
Aceptado: 04 sept 2022
DOI: https://doi.org/10.2478/raon-2022-0040
© 2022 Domen Plut, Barbara Faganel Kotnik, Luka Pusnik, Peter Slak, Ziga Snoj, Vladka Salapura, published by Sciendo
This work is licensed under the Creative Commons Attribution 4.0 International License.
Haemophilic arthropathy (HA) is caused by recurrent bleeding into joints and is characterized by synovial hypertrophy with hemosiderin deposition, cartilage destruction, and structural changes of subchondral bone. Long-term repeated hemarthroses lead to joint destruction and severe functional impairment.1 As the development of HA ordinarily begins during childhood, albeit with minimal changes, early detection is indispensable. Early recognition of subclinical arthropathy based on the imaging modalities is fundamental as it enables appropriate prophylactic treatment modification and prevents further disease progression.2,3 Among the imaging tools, magnetic resonance imaging (MRI) with its high spatial and contrast resolution is superior to the other modalities and enables the most precise assessment of early arthropathic changes.4 As MRI is a time-consuming modality with limited availability, a high cost, and requires sedation in young children, routine assessment of multiple joints with it is not feasible. Ultrasound (US) has been proven to be highly reliable in the assessment of early inflammatory and destructive joint changes in the adult haemophilic population5, 6, 7, however very few studies so far focused on the reliability of US in the paediatric haemophilic population. The US has even more advantages in the paediatric population: it is a safe technique without radiation, it enables a quick assessment of multiple joints, and sedation is not required even in young children. Thus, with possible detection of joint effusions, synovial hypertrophy, cartilage changes, and subchondral bone erosions it may be an invaluable tool for recognition of sub-clinical HA in children.8,9
We were interested if the changing appearance of the growing bone hinders the ultrasonographic evaluation of the pathologic processes caused by HA. Hitherto, a paucity of data has been published regarding US measurements of hyaline cartilage thickness in healthy children in comparison to MRI measurements. Consequently, published values are not standardised and require further research to distinguish unaffected growing bone from the pathologic processes caused by HA.10, 11, 12, 13, 14 The aim of our study was to assess the reliability of the US for evaluation of haemophilic arthropathy in children in comparison to the MRI.
The study was approved by the National Medical Ethics Committee (reference number 0120523/2015-8). The participants in this study were children, therefore informed consent for the participants was signed by their parents. The participants, however, gave their informed assent to the study. Research was conducted following the Helsinki Declaration.
The study included all children with severe haemophilia A in the country. The patients were recruited at the Slovenian National Haemophilia Comprehensive Care Centre at the University Medical Centre Ljubljana. The inclusion criteria were: diagnosis of severe haemophilia A, prophylactic treatment with factor concentrates, and age between 6 and 18 years. The age of 6 years as the low cut-off was chosen to avoid the need for anaesthesia for the MRI. The exclusion criteria were non-cooperation and contraindications for the MRI. Patient history (history of joint bleeds, haemophilia joint health score (HJHS), prophylaxis information) was retrieved from their medical records.
A ProSound F75 US scanner with a 13–5 MHz electronic linear-array transducer (Hitachi Aloka Medical, Ltd. Tokyo, Japan) was used to perform the US examinations. US was performed by an experienced paediatric radiologist (7 years of subspecialty experience). The assessment of each joint was made using the HEAD-US protocol and scoring method. This standardised method includes bilateral systematic evaluation of the elbows, knees, and ankles in defined positions for the detection of hypertrophied synovium and osteochondral damage. The results for each joint are expressed on a 9-point scale (0–8; 0 corresponds to the best joint condition, while 8 corresponds to the worst joint condition).8 The total scanning time per patient for all joints combined was approximately 20 minutes. A series of images and clips from all examinations for each patient were additionally independently reviewed and scored by another paediatric radiologist (2 years of subspecialty experience) to determine the inter-rater reliability. Both US reviewers were blinded to the results of MRI examinations.
MRI was performed on a 3T Achieva unit (Philips Healthcare, Eindhoven, The Netherlands). Phased array coils were used for the imaging of each joint. The protocol included 3D T2*-weighted water selective gradient echo sequence (FOV, 160×160×108mm; voxel size, 0.58×0.58×0.50mm; flip angle: 15°; TE 9.2/6.1ms; TR 26ms), and 3D proton density (PD) weighted turbo spin echo sequence (FOV, 160×160×161mm; voxel size: 0.52×0.52×0.52mm; TE 33ms; TR 1000ms). The total scanning time for each joint was approximately 15 minutes. In each patient, all joints were scanned in a single session for a total examination time around 1.5 hours. The MRI examinations were scored according to the International Prophylaxis Study Group (IPSG) MRI scale. The IPSG score includes evaluation of joint effusion, hypertrophied synovium with hemosiderin deposition, and osteochondral damage. The IPSG score uses an 18-point scale (0–17; 0 corresponds to the best joint condition with no disease present, while 17 corresponds to the worst joint condition with progressive arthropathy).15 The presence of HA was defined as IPSG score > 0. The scoring was performed independently by two experienced musculoskeletal radiologists (19 and 4 years of subspecialty experience) who were blinded regarding the results of the US examinations.
Descriptive statistics were obtained to describe the characteristics of the study group. Pearson correlation coefficient (r) was used to analyse the correlation between US and MRI examinations. Correlation was considered poor if r was < 0.3, fair if r was < 0.6, substantial if r was < 0.8, and excellent if r was > 0.8.16 The correlation results were graphically illustrated. The inter-rater reliability of HEAD-US and IPSG MRI scoring system for the total scores was made with Lin’s concordance correlation coefficient (CCC) and for all the sub-scores using Cohen’s kappa statistics (with quadratic weights). The results of US were referenced to the results of MRI in order to obtain measures of diagnostic accuracy (specificity, sensitivity, positive predictive value, and negative predictive value). Statistical analysis was performed using IBM SPSS Statistics for Windows software, version 25 (IBM Corp.). A two-tailed P value < 0.05 was considered to indicate statistical significance.
The study group included a total of 10 patients (age range 6 to 17 years, mean age 11.5 years). In each patient, six joints (elbows, knees, and ankles bilaterally) were systematically examined first using the US, followed by MRI, according to the protocols. Altogether in the study we assessed 60 joints: 20 elbows, 20 knees, and 20 ankles.
All the patients included in the study have been receiving prophylactic treatment with clotting factor concentrates. The type of prophylaxis for the patients was primary or secondary. The patients were on three times per weekly regimen. A proportion of patients were on individual prophylactic regimens according to population-based pharmacokinetic tools. Two patients developed inhibitors to prophylactic treatment. In one patient the inhibitors were successfully eradicated by immune tolerance induction. Details on treatment and joint-bleeds history are shown in Table 1 along with other study group baseline characteristics.
Characteristics of subjects included in the study
| Age: mean; range (years) | 11.5; 6–17 | |||
| Age at the start of prophylaxis: mean; range (years) Primary prophylaxis – 5 patients: Secondary prophylaxis – 5 patients: | 3; 0.8–6.6 2.2; 0.83.8; 2.5 | –3.8 –6.6 | ||
| Duration of prophylaxis: mean; range (years) | 9.1; 3.2–14.7 | |||
| Haemophilia Joint Health Score (HJHS): mean; range | 0.9; 0–7 | |||
| Number of previous joint bleeds per patient: mean; range | 16.2; 0–83 | |||
| Number of previous joint bleeds per joint: mean; range | 2.5; 0–71 | |||
| Number of previous joint bleeds: | Elbows | Knees | Ankles | Overall |
| 0 (number of joints) | 14 | 7 | 9 | 30 |
| 1–4 (number of joints) | 5 | 12 | 8 | 25 |
| > 5 (number of joints) | 1 | 1 | 3 | 5 |
Descriptive statistics for US and MRI results are gathered in Table 2.
Descriptive statistics for US and MRI assessment scores
| Joints | Statistic | US | MRI |
|---|---|---|---|
| % of zeros | 85 | 75 | |
| Elbows | Median Mean | 00.35 | 01.05 |
| SD | 0.93 | 2.82 | |
| % of zeros | 100 | 100 | |
| Knees | Median Mean | ||
| SD | |||
| % of zeros | 65 | 70 | |
| Ankles | Median Mean | 00.8 | 01.6 |
| SD | 1.2 | 2.2 | |
| % of zeros | 83.3 | 81.7 | |
| Overall | Median Mean | 00.38 | 00.88 |
| SD | 0.92 | 2.16 |
MRI results were used as a reference standard for joint status. The results of the correlation between US and MRI for detection and evaluation of HA in children are summarized in Table 3. The correlation with MRI for each joint type is graphically depicted in Figure 1. The correlation with MRI for US for the total score was excellent for all joints (r = 0.849 for the elbows, r = 1 for knees, r = 0.842 for ankles). The correlation of scores for specific joint components showed fair, moderate, or excellent correlation for all joint components in all joints. The correlation was the lowest for the evaluation of cartilage and bone in the ankles (r = 0.546 and r = 0.478) and bone in the elbows (r = 0.499). Figures 2-5 show images from the study.
Figure 1
Concordance plot for depicting agreement between US and MRI scores for all three joints. Equal size of the fields denotes perfect agreement. The plots demonstrate overall an excellent agreement between the methods. It can also be observed, that in most cases of discordance, US slightly undervalued the progression of the joint disease.
The results of the correlation analysis
| US vs MRI | Pearson’s correlation coefficient (r) |
||
|---|---|---|---|
| Elbows | Knees | Ankles | |
| Total score | 0.849 | 1 | 0.842 |
| Synovium | 0.841 | 1 | 0.722 |
| Cartilage | 0.829 | 1 | 0.546 |
| Bone | 0.499 | 1 | 0.478 |
Note: all the reported correlations are statistically significant (p < 0.05).
Figure 2
Anterior transverse US images over the distal humeral epiphysis in a 7-years and 16-years old healthy boys. A wavy osteochondral surface consisting of the convex capitellum and the concave trochlea is shown. Note the age-dependent anatomic differences: subchondral bony surface in the younger child
The inter-rater reliability of interpretation was excellent for the US examinations of all joints. The Lin’s CCC values for the total scores ranged from 0.986 to 1.000. The inter-rater reliability for the MRI was also excellent for all joints with the CCC values for the total scores ranging from 0.957 to 0.993.
Our study included 49 joints (15 elbows, 20 knees, and 14 ankles) with no signs of HA on MRI (IPSG scores were 0) and 11 joints with HA (5 elbows and 6 ankles). The IPSG MRI score of joints with HA ranged from 1 to 8, mean was 4.8. Two joints that showed signs of HA on MRI (two elbows in the same patient with IPSG MRI scores 4 and 1) were scored 0 on the US. There was one false positive on the US, an ankle with a score of 1. The calculated measures of diagnostic accuracy for HEAD-US are presented in Table 4.
Measures of diagnostic accuracy for detection of haemophilic arthropathy by US (HEAD-US) in comparison to MRI (IPSG MRI score) as the reference standard
| Specificity | 81.8% |
| Sensitivity | 98% |
| Positive predictive value | 90% |
| Negative predictive value | 96% |
Our study aimed to evaluate the reliability of US (HEAD-US scanning protocol and scoring method) for the detection and evaluation of haemophilic arthropathy in children in comparison to MRI (ISPG MRI scoring scale). We evaluated the three most commonly affected joints (ankles, knees, and elbows) in all children with severe haemophilia A in our country (n = 10). Overall, we evaluated 60 joints. The results of the correlation analysis showed a very high correlation for the evaluation of haemophilic arthropathy between the US and MRI for all the joints (r = 0.849 for elbows; r = 1 for knees; r = 0.842 for ankles). Excellent inter-rater reliability for both the US and MRI in our study further supports the validity of both methods for haemophilia imaging in children.
These results show that the HEAD-US method is reliable in comparison to MRI for the detection and quantification of HA in children. High specificity and sensitivity (81.8% and 98%) confirm the method as a dependable tool for the recognition of the presence of HA, whereas high correlation proves the method is also reliable in the quantification of the disease progression. Our results indicate that US is reliably applicable for all evaluated joints (elbows, knees, and ankles), however the detailed analysis of the joint components (synovium, cartilage, bone) showed some important differences. The correlation between the methods was the lowest for the evaluation of the cartilage and bone changes of the ankles (r = 0.546 and 0.478) and bone changes of the elbows (r = 0.499) in comparison to other joint components (r > 0.7). The lower correlation for the evaluation of osteochondral changes in the ankles was due to the limited visualization of the central weight-bearing part of the osteochondral surface of the ankle joint by US, which is a more commonly affected area of the joint. Comparable results were observed in our previously performed study in the adult population.17 Similarly, the lower correlation for the bone changes in the elbows was due to inability of the US to detect centrally located subchondral cysts. Nevertheless, as noted above, the overall correlation between both methods for all the joint components for all the joints was still substantial. In our study in the adult population, we observed a lower correlation between the US and MRI for the detection and evaluation of synovial hypertrophy in the ankles (r = 0.561). In this study in the paediatric population however, this was not the case (r = 0.722). We believe this can be attributed to a generally better ability of US to differentiate soft tissues in children due to a higher tissue water content. As synovial hypertrophy is the earliest sign of HA, high reliability to evaluate this finding in all joints in children is important for the clinical application of the method.
Figure 3
PD weighted MRI of ankles in sagittal plane. Image
Figure 4
An example of good concordance between HEAD-US and MRI in a 7-years old child. US image of the femoral trochlea in the transverse plane is shown
Figure 5
An example of a lesion causing a discordance between the US and MRI. A T2* weighted MR image of an ankle of a 16-years old boy in the sagittal plane is shown. A small subchondral cyst covered with intact cortical bone and articular cartilage (white arrow) is shown. MRI demonstrates a defect which cannot be visualized by US.
In the published literature, the comparisons of the joint assessment between the US and MRI within the paediatric haemophilic population are scant. Only three studies included exclusively children within their study group. Doria
The findings of the currently presented study and our previously published study in the adult population17 made us reconsider our clinical practice. Due to the good availability of US machines and reliability of the US to detect even early HA in clinically asymptomatic joints, we incorporated the US into our regular clinical yearly follow-up of paediatric patients with haemophilia. All the children with severe haemophilia in our country have been included in the screening program, even children younger than 6 years old. During this time, we found early HA in clinically asymptomatic joints with no previously recorded bleeds in two children and consequently modified their prophylactic treatment regimen.
The study had some research design limitations. Although our study group included all children with severe haemophilia A in the country, due to the rarity of the disease, the overall number of patients was relatively low (n = 10). Furthermore, we couldn’t include the youngest children with haemophilia aged under 6 years due to the requirement of general anaesthesia to perform MRI in this group of children. Most joints we evaluated in our study were either healthy or had only early HA. This is because HA is a progressive chronic disease and all of the included patients had prophylactic treatment since the early youth, therefore more progressive disease forms were prevented. However, extensive studies evaluating the value of US in comparison with MRI in patients with progressive HA have been already performed in the adult population and, moreover, diagnosing early HA remains the challenge for today’s medicine.
Our study proved that US using the HEAD-US method performed by paediatric radiologists is a reliable tool for detection and quantification of haemophilic arthropathy in children in comparison to MRI. Due to its simplicity, availability, and reliability, HEAD-US is an invaluable tool in diagnostics and regular follow-up of children with haemophilia and can be safely included into the regular screening protocols of children with severe haemophilia where possible. Further studies are needed to answer some important questions regarding the use of HEAD-US in children with haemophilia: what is the ideal start age for the screening? How often should the screening be performed during the childhood, and who should perform the scanning (radiologist, clinician, physiotherapist)?