The study was designed as a prospective and blinded in vitro study, where native radiographs and CE radiographs were compared with gross pathology as gold standard.
Fourty-eight pairs of front feet were collected from an abattoir from October 2014 and February 2015. The feet were frozen at -18°C for a minimum of one week and a maximum of four weeks. All feet were thawed 24h before the beginning of the examination. For practical reasons, the examinations were performed in 4 separate sessions, including 10, 26, 22, and 38 feet, respectively. No anamnestic information on the horses was provided. Only feet that optically resembled those of warmblood horses were considered eligible for the study. All limbs were cleaned, the frogs trimmed, and the sulci packed with special wax (Moser®, Waiblingen, Germany). Subsequently, all limbs were individually labelled with a metal plate carrying a five-digit randomized (Microsoft Excel 2010, Microsoft, Redmont, USA) number.
First, the standard DPr-(65°)PaDiO (66kV and 3 mAs) and PaPr(50°)-PaDiO (66kV and 2,5 mAs) views were taken in a random order (Microsoft Excel 2010, Microsoft, Redmont, USA). Standard wooden blocks, skyline tunnel and bracket for the X-Ray tube were used (manufactures unknown) to obtain the pictures with distance of the examiner. Second, 3ml Iomeprol 300mg ID/ml (Imeron 300M© Bracco Imaging GmbH, Konstanz, Germany) were injected directly into the NB using a spinal needle (BD© Spinal Needle, 0,9 *90mm, Becton Dickinson GmbH, Karlsruhe, Germany) and a standard 5 ml syringe. The landmark for needle insertion was a point just proximal to the central sulcus of the frog, with the needle directed toward the apex of the frog and parallel to the ground surface of the hoof (Turner 2013). The success of injection was controlled with a latero-medial (LM) view. Afterwards, the additional DPr-PaDiO and PaPr-PaDiO views were obtained using the settings described above.
A standard portable x-ray tube (TR 90/30, Gierth®, Riesa, Germany) was used for all radiographic examinations, without using a grid. A Mark 3 (Sound-Eklin® Carlsbad, CA, USA) direct X-Ray detector obtained the pictures. The DICOM files were then sent to the clinics picture archiving and communication system (PACS, Synology© running open source PACS from dcm4che). The X-Rays were examined on an Apple Thunderbold® Display A 1407 (Apple®, Cupertino, CA), USA) with 27“display running Osirix MD® version 6.5 (Apple®, Cupertino, CA); both parts were FDA approved.
After acquiring the x-rays the feet were dissected. The joint capsule and the collateral ligaments of the distal-interphalangeal-joint (DIP) were cut in order to exarticulate the joint. Then, the DIL and the DDFT were cut as distally as possible. This cut was continued around the DDFT. Afterwards, a bundle of navicular bone and DDFT was isolated, so that the flexor surface and the opposite part of the DDFT were visible. The DIL and the distal insertion of the DDFT on P3 were not evaluable. If there was an adhesion between the navicular bone and the DDFT, it was carefully removed. From all steps of preparation high resolution photographs were taken. The pathological assessment was directly performed and documented.
Evaluation of the x-rays and gross pathology:
The researcher (Resident ACSVSMR) and his supervisors (Dipl. ACSVSMR and a head of an equine university clinic in Germany) developed a scoring system for the evaluation of the x-Rays and the gross pathology to make the examination more objective and suitable for statistical analyses. All of the following findings were graded on a 0–3 scale (score 0 = good; score 1 = mild; score 2 = moderate; score 3 = severe changes) except of a round radiolucent opacity. The latter was graded on a 0–2 scale (score 0 = not existent; score 1 = existent, but without clear demarcation; score 2 = obvious, with clear demarcation). In order to weight the most important and unequivocal pathological findings, factors were introduced and multiplied with the respective score (Table 1).
Table 1
Test criteria’s for the x-ray projections and the gross pathology
Name | criteria | from | to | factor |
projection | DPr-PaDiO | | | |
O1 | contour | 0 | 3 | 1 |
O2 | bone density | 0 | 3 | 1 |
O3 | bone structure | 0 | 3 | 1 |
O4 | evaluation of vascular channels | 0 | 3 | 1 |
O5 | round radio lucent zone | 0 | 2 | 4 |
projection | PaPr-PaDiO | | | |
S1 | border between cortex and medulla | 0 | 3 | 1 |
S2 | bone density | 0 | 3 | 1 |
S3 | bone structure | 0 | 3 | 1 |
S4 | round radio lucent zone | 0 | 1 | 4 |
S5 | flexor cortex of NB | 0 | 3 | 2 |
projection | DPr-PaDiO CE | | | |
KO1 | filling of navicular bursa | 0 | 3 | 1 |
KO2 | pooling of contrast media (radio dense) | 0 | 3 | 2 |
KO3 | lack of contrast media (radio lucent) | 0 | 3 | 2 |
KO4 | structue of NB | 0 | 3 | 1 |
projection | PaPr-PaDiO (CE) | | | |
SK1 | Borders from contrast medium | 0 | 3 | 2 |
SK2 | filling of navicular bursa | 0 | 3 | 1 |
projection | gross pathology | | | |
A10 | navicular bone (flexor cortex) surface | 0 | 3 | 1 |
A11 | NB cartilage thickness | 0 | 3 | 1 |
A12 | NB cartilage color | 0 | 3 | 1 |
A13 | NB cartilage appaerance | 0 | 3 | 1 |
A14 | DDFT surface | 0 | 3 | 1 |
A15 | DDFT adhesion | 0 | 3 | 1 |
A16 | DDFT color | 0 | 3 | 1 |
A17 | DDFT metaplasia | 0 | 3 | 1 |
In the DPr-PaDiO view the contour, density, structure, changes in the navicular channels, and the existence of a radiolucent zone (factor 4) were evaluated. In the PaPr-PaDiO view the border between the flexor cortex and medulla, density, structure, the flexor cortex (factor 2), and theexistence of a radiolucent zone (factor 4) were noticed. These radiological findings were chosen for interpretation according to recommendations by Dyson 2008 and Butler et al. 2017. Some minor additional findings were documented but excluded from statistical analyses.
To date, there is no information on the DPr-PaDiO view with contrast medium in the NB. So far, only one picture showing the anatomic dimension of the NB has been published in literature (Jann et al. 1991). Figure 1 shows a bursogram without any pathological findings. Due to the lack of information about a DPr-PaDiO CE view, we hypothesized that a tendon tear would lead to a long, straight, and radio dense line (Figs. 2 and 3), a defect in the flexor cortex would manifest as a round, radio dense area (factor 2) (Fig. 4–6), and an adhesion would be represented by a radiolucent zone (factor 2) (Figs. 5 and 6). In addition, bursitis would lead to an irregular filling of the NB. The structure and border of the navicular bone were also graded. The dorsal border of the NB was also in aim, but only noticed as regular (grade 0) or irregular (grade 1), but not used for the statistical analysis.
There is, however, some information about the PaPr-PaDiO CE view (Wuerfel and Hertsch 2005; Turner 2013). Accordingly, the regular filling of the NB and the border to the DDFT, as evidence of dorsal tears or fibrillation (Figs. 2 and 3), or radiodense zones within the navicular bone, as evidence for lesions in the flexor surface, were graded (Factor2). The reported three lines, providing detailed information about the flexor cortex, were only seen in 43 feet. Therefore, they were noticed but excluded from statistical analyses.
In gross pathology the appearance, color, thickness, and the presence of lesions of the flexor surface was graded. Furthermore, the dorsal part of the DDFT was examined including surface, color, adhesions, and changes in appearance (Blunden et al. 2006a, Blunden et al. 2006b). The surface of the cartilage of the DIP joint was also evaluated but excluded from statistical analyses.
Statistic
The obtained scores from x-ray examination and gross pathology were transferred to one table (Microsoft Excel 2010, Microsoft, Redmont, USA). The scores were then summed up and arranged in four groups (group 1 = no findings, group 2 = mild, group 3 = moderate, and group 4 = severe findings). This evaluation was done for each radiographic view and the gross pathology, respectively. The group definition and frequencies are provided in Table 2. Subsequently, results were exported to SPSS (Version 23, IBM, Armonk, USA). Fleiss' kappa was calculated to compare the agreement from x-ray findings and gross pathology groups. Spearman's rho (ρ) was used to assess the association between the sum of scoring of every x-ray projection and gross pathology. Spearman's rho (ρ) and a Kruskal-Wallis-Test were performed for the sum of classes of native and all pictures compared to the gross pathology groups. Finally, sensitivity and specificity were determined. Therefore, group 1 and 2 were considered negative for changes related to ND and group 3 and 4 positive for each x-ray view and the gross pathology.
Table 2
Group definition and frequencies for all x-ray projections and gross pathology
| DPr-PaDiO | PaPr-PaDiO | DPr-PaDiO CE | PaPr-PaDiO CE | gross pathology |
Group1 | 29 (30.2%) | 42 (43.7%) | 45 (46.9%) | 42 (44.2%) | 39 (40.6% |
from | 0 | 0 | 0 | 0 | 0 |
to | 2 | 3 | 2 | 1 | 3 |
Group2 | 54 (56.3%) | 35 (36.5%) | 28 (29.2%) | 19(20.0% | 28 (29.2%) |
from | 3 | 4 | 3 | 2 | 4 |
to | 5 | 7 | 6 | 3 | 10 |
Group3 | 7 (7.3%) | 9 (9.4% | 10 (10.4%) | 16 (16.9%) | 11 (11.46%) |
from | 6 | 8 | 7 | 4 | 11 |
to | 9 | 12 | 10 | 5 | 15 |
Group4 | 13 (13.5%) | 10 (10.4%) | 13 (13.5%) | 18 (18.9%) | 18 (18.75%) |
from | 10 | 13 | 13 | 6 | 16 |
N Sum | 96 | 96 | 96 | 95 | 96 |