Diagnostic performance of ultrasonography, bursography and standing magnetic resonance to detect navicular intrabursal pathology in horses with foot pain

Navicular intrabursal pathology is one of the most common causes of forelimb lameness in horses. An accurate diagnosis of which structure is involved is essential for therapy and prognosis. The study aimed to compare the diagnostic performance of ultrasonography, bursography, and standing magnetic resonance (MRI) imaging as diagnostic tools of the proximal navicular intrabursal pathology using bursoscopy as the gold standard. A total of 21 feet of 17 horses met the inclusion criteria. The feet underwent ultrasonography, followed by MRI, and thereafter bursography and bursoscopy. The images were interpreted blindly to document the presence or absence of the podotrochlear lesions. The lesions of interest were dorsal fibrillation and splits of the deep digital flexor tendon (DDFT), adhesions, and fibrocartilage defect of the navicular bone. The accuracy, specificity, sensitivity, positive predictive values, and negative predictive values were calculated. Furthermore, the kappa test was used to calculate the inter-observer agreement. The bursoscopy revealed lesions in 95% (20/21) of the feet. Eighteen dorsal fibrillation and eleven split lesions in the DDFT, nineteen fibrocartilage defects, and eight adhesions were identified. MRI showed accuracy of 95% (20/21) for dorsal fibrillation and 86% (9/21) for split lesions in the DDFT, whereas the accuracy for fibrocartilage defects was 60% (14/21). Adhesion formation with MRI was predicted correctly in only 57% (3/21) of the feet. Bursography showed with MRI similar accuracy regarding the adhesions. Moreover, bursography had an accuracy for fibrocartilage defects of 67% (20/21), and DDFT lesions were detected in 86% (18/21) of the feet. The ultrasound examination predicted DDFT lesions in 67% (14/21) of the feet, whereas its accuracy for adhesion lesions was 33% (10/21). (95% CI). The inter-observer agreement was the lowest for ultrasound compared to MRI and bursography. A combination of ultrasonography and bursography can be an alternative to standing MRI when the latter is not an option or as a complementary test.

Methods
2.1. Data collection

This prospective blinded study was carried out between 2016 and 2020 on client-owned adult sport non-racing horses referred with unilateral or bilateral forefoot lameness after obtaining the owners’ approval. The horses were included in the study if the lameness was located in the hoof region via diagnostic analgesia, and 2) radiography, ultrasonography, and standing MRI of the foot were carried out, 3) bursography combined with bursa analgesia was performed within two weeks after the previous imaging modalities, and 4) bursoscopy was carried out within three weeks after the bursography. The exclusion criteria were negative podotrochlear bursa analgesia performed prior to the MRI or ultrasound of the foot. The inclusion was decided upon by an associated Diplomate of the European College of Veterinary Diagnostic Imaging (KG) and a German board-certified veterinary radiologist (MM). The sample size was based on convenience sampling, i.e., the number of animals meeting inclusion criteria during the study period. The diagnostic work-up was performed as part of the patient’s standard clinical evaluation. Therefore, no institutional animal care and use approval was required.

2.2 Clinical examination 

The clinical examination was performed by an experienced German board-certified equine surgeon (GM). All horses were examined clinically on hard and soft surfaces, on the straight line, and on circles, either in the clinic facilities or in the stable. After gaits were evaluated,  flexion tests and hoof tester examination, the palmar digital nerve analgesia was applied just proximal to the ungular cartilage using 1 mL of mepivacaine with epinephrine, followed by a compression bandage [17]. The analgesia was considered positive if the lameness resolved or substantially improved after 5 min. 

2.3 Radiographic and sonographic examination

Standard radiographs, including lateromedial, dorsopalmar, and dorso45°proximal-palmarodistal radiographs, were acquired after proper foot preparation. Subsequently, ultrasound examination of the podotrochlear apparatus was also performed on the same day by the same operator (GM) as described previously [9,18]. Briefly, the palmar area of the pastern was prepared for sonography. The images were acquired transversely and longitudinally using a 7.5MHz micro-convex transducer (EDGE, Fujifilm Sonosite, Inc., Bothell, WA, USA) in a comparable manner with the foot in a non-weight-bearing position. The frequency was adjusted to achieve better image quality, and the focus was set at the level of the DDFT. The angle contrast ultrasound technique (ACUST) [19] was used to increase the conspicuity of the DDFT lobes and surrounding soft tissues to enhance evaluation of the dorsal border. Moreover, ACUST is the only possible way to image the DDFT in the suprasapodeme region due to the hoof configuration and subsequent oblique insonation angle. The transcuneal approach was not realised, since only the proximal bursa was the area of interest.

2.4 Standing MRI material and protocol

The foot MRI was acquired with a 0.27T MRI scanner (Hallmarq Veterinary Imaging Ltd., Guildford, UK) in a standing position under sedation. Images were obtained using gradient recall echo, fast spin echo, and short tau inversion recovery (STIR) sequences in sagittal, transverse, and dorsal planes (Table 1). The transverse images were angled perpendicular to the DDFT in the supra- and infrasesamoidean region. In addition to the standard basic protocol, an additional transverse proton density and high-resolution T2 FSE sequences were acquired, depending on the findings observed at the time of the acquisition by the radiologist performing the examination. 

2.5 Bursography

The bursography was performed simultaneously with intrathecal diagnostic analgesia using the lateral approach as described previously [20]. Briefly, a 3 mL mixture of 1:1 positive contrast agent and local anaesthetic (1.5 mL mepivacaine, Intra-Epicaine 20mg/mL, Dechra Veterinary Products Deutschland GmbH, Aulendorf, Germany, and 1.5 mL Solutrast, 250 mgI/mL Iopamedol, Bracco Imaging Deutschland GmbH, Konstanz, Germany) was injected intrathecally under aseptic conditions. Immediately after the intrathecal injection, lateromedial and palmaro-45°-proximal-palmarodistal oblique radiographs of the navicular region were acquired. After 5 min, the results of the analgesia were evaluated and considered positive if a significant improvement in or abolishment of the lameness was observed.

2.6 Bursoscopy

 All surgeries were performed by the same operator (GM) using a standard transthecal approach [15], and treatments were administered as needed. Briefly, after positioning the animal in dorsal recumbency and after routine surgical preparation, the arthroscopic and instrumental portals were created just proximal to the ungular cartilage. The T-ligament was sharply dissected, and the bursa was inspected. Using motorized blades, arthroscopic scissors, and an arthroscopic punch, the lesions were debrided. During the surgical debridement, the arthroscopic and instrumental portals were often exchanged. Eventually, after copiously flushing, the portals were closed, and a sterile bandage was applied. 

2.7 Image analysis

Diagnostic and surgical findings were reviewed by independent, experienced observers who were blinded to the history, clinical examination, or tentative diagnosis of these horses. The ultrasound, bursography, and MRI images were anonymized in DICOM format to enable the operator to adjust the contrast windows level and width as required. The bursoscopy images and clips were provided in JPEG and MP4 format, respectively. 

The bursography was assessed by a board-certified European College of Veterinary Diagnostic Imaging radiologist (ZJ). The MRI and ultrasound findings were interpreted by an American College of Veterinary Radiology-certified radiologist (NW). Since the same operator had to evaluate the modalities of the same patient, the ultrasonography images were evaluated six months after the MRI to avoid possible bias. The bursoscopy images and videos were assessed by a member of the board of the European College of Veterinary Surgeons (BB).  

To assess interobserver agreement of all three diagnostic modalities, the interpretation of all findings was performed independently and blinded by an experienced radiologist (MM, German board-certified veterinary radiologist) and one board-certified radiologist (NW, ZJ). The assessment protocols were collected and adapted from literature [6,7,10,11,13] by the first author and approved by the observers. The proximal navicular bursa was reviewed for the following lesions: 1) fibrocartilage defects of the proximal aspect of the navicular bone flexor surface (bursography, MRI), 2) adhesions/synovial masses (ultrasound, bursography, MRI), 3) split of DDFT (ultrasound, bursography, MRI), and 4) dorsal border lesion/tendinopathy of DDFT (ultrasound, bursography, MRI). In case of DDFT tendinopathy and fibrocartilage defects, the location was also reviewed (lateral/medial). The lesions were reported as present or absent, grading the severity according to the worst lesion in that both lobes were affected, and their position (medial vs lateral vs both). Lesions not evaluated with bursoscopy were purposely excluded from the study.

The fibrocartilage was defined as normal if it appeared as a uniform radiolucent area 1 to 2 mm in thickness covering the flexor surface of the navicular bone in bursography. In MRI, the fibrocartilage was considered abnormal if irregularity of the flexor surface was evident without a compact defect associated with subjective thinning and change in signal of the palmar fibrocartilage. 

Adhesion was defined in bursography as focal interruption of the contrast column between the navicular bone and the dorsal margins of the DDFT. It was the authors’ decision to use the term adhesion as a synonym for both synovial masses and adhesions [21].  An adhesion was diagnosed in MRI and ultrasound when there was a discontinuity in the navicular bursa fluid between two structures lining the proximal pouches, with or without extraneous tissue present.  During the surgery, adhesions were identified as an abnormal band of tissue connecting two structures. 

A split of the DDFT was diagnosed in bursography if there was a linear accumulation of contrast agent inside the dorsal area of the DDFT lobe. In MRI, the split was defined as narrow linear areas of increased signal intensity on T1 sequences at the suprasesamoidean portion of DDFT. In ultrasound, the DDFT lesion was defined as a heterogeneous hypoechogenic line extending through the DDFT lobe. 

A dorsal fibrillation of the DDFT was defined in bursography as loss of the smooth margin of the palmar surface of the bursa. In MRI, this lesion was characterised as irregularity of the dorsal margin of the DDF tendon visible on T1, PD, T2, and STIR FSE sequence, whereas in ultrasound examination, it was defined as interruption of the dorsal margin of the DDFT with or without isoechogenic structures originating from DDFT and protruding into the navicular bursa.  

2.8. Statistical assessment   

Statistical analyses were performed by the first author in collaboration with an external medical statistician using SPSS V.23 (SPSS Software, IBM Munich Center GmbH, Munich, Germany). The statistical analyses included calculating the sensitivity, specificity, and accuracy, and the positive and negative predictive values for each modality using bursoscopy as the gold standard with a 95% confidence interval. Weighted kappa values for each lesion’s grade with 95% confidence intervals were calculated to assess inter-observer agreement in MRI, bursography, and ultrasound as follows: Kappa <0  as “no agreement”; Kappa = 0.0 to 0.20 as “poor agreement”; Kappa = 0.21 to 0.40 as “fair agreement”; Kappa = 0.41 to 0.60 as “moderate agreement”; Kappa = 0.61 to 0.80 as “substantial agreement”; Kappa = 0.81 to 1.00 as “almost perfect agreement” [22]. The prevalence rate of each lesion was used as reported previously [7].