CANINE ELBOW dysplasia (CED), a common developmental disorder of the cubital joint of dogs, is comprised fragmented medial coronoid process (FMCP), ununited anconeal process (UAP), osteochondrosis (OC), and elbow incongruity, alone or in combination. Although the epidemiology, etiopathogenesis, diagnosis, and treatment of CED have been extensively investigated, the ideal diagnostic approach for accurately and consistently determining the presence and extent of CED has not been established. Our objective was to review the literature on diagnostic imaging for CED, synthesize current concepts and develop a framework for optimal diagnostic algorithms to establish early diagnosis, grading of pathologic changes, and staging progression of CED to direct treatment options and determine prognosis.
Radiography has been as the standard-of-care imaging modality for diagnosis, grading, and registry of CED. For accurate and complete radiographic assessment of elbow disorders, 4 projections should be evaluated: standing-angle (also termed neutral or extended) mediolateral, flexed mediolateral, craniocaudal, and craniolateral–15°–caudomedial oblique.1–3
In large breed dogs, the anconeal process of the ulna is a separate center of ossification. The physis associated with the anconeal ossification center is seen radiographically in the immature dogs until 20–22 weeks of age.4,5 If the physis remains radiographically visible beyond this time, it is considered ununited. An UAP is best diagnosed on a flexed mediolateral projection and appears radiographically as a discrete or irregularly marginated radiolucent gap between the anconeal process and proximal ulna (Fig 1). A flexed mediolateral projection avoids superimposition of the medial humeral epicondyle and the anconeal process allowing clear observation of the process. The physis associated with the medial humeral epicondyle is superimposed over the anconeal process on extended mediolateral projections, and because it does not normally close radiographically until after closure of the anconeal physis, a superimposed normal medial humeral epicondylar physis may be confused for UAP if the flexed mediolateral projection is not performed (Fig 2).4 Although radiographic changes associated with UAP are not consistently noted on other projections, all 4 recommended projections should be obtained and evaluated because of the potential for concurrent disorders.6
Elbow OC lesions occur almost exclusively on the weight-bearing surface of the distal medial humeral condyle.1,2,4,5,7 OC is observed as a radiolucency, an irregularity, flattening, or defect in the subchondral bone of the articular margins of the humeral condyle. Most often, there is associated sclerosis of the subchondral bone surrounding the radiolucency. OC lesions can be identified on lateral (neutral and flexed) and craniocaudal projections, but are generally best identified on craniocaudal or cranial 15° lateral–caudomedial oblique projections (Fig 3).1,2,4,5,7“Kissing lesions”, thought to result from erosive changes in articular cartilage and subchondral bone associated with FMCP, also occur on medial aspect of the humeral condyle and appear as subchondral sclerosis with or without associated lucency or concavity of the articular margin of the condyle (Fig 4). In some cases, kissing lesions can be distinguished from OC lesions by presence of subchondral lucency or sclerosis of a surface of the radius or ulna, adjacent to the humeral condylar OC lesion.4,6,7
Whereas UAP and OC are typically diagnosed definitively by comprehensive radiographic assessment, presence and severity of FMCP and elbow incongruity can be difficult to diagnose with certainty using radiography.1,7–12 FMCP is often considered a “rule out” diagnosis made when there is radiographic evidence of elbow osteoarthritis without definitive radiographic evidence of UAP, OC, trauma, or incongruity.1,3,7 A radiographically distinct osteochondral fragment is rarely identified because of the 2-dimensional nature of radiographic projections of the complex 3-dimensional anatomy of the canine elbow joint and the wide spectrum of pathology involving cartilage and/or bone that occurs with FMCP.7 Thus, radiographic findings associated with FMCP should be carefully investigated on multiple projections. Radiographic findings associated with elbow dysplasia and osteoarthritis secondary to FMCP include proximal anconeal osteophytosis, proximal radial osteophytosis, and subchondral sclerosis of the semilunar notch and medial coronoid process of the ulna on flexed mediolateral and craniocaudal projections (Fig 5).1–7 The radiographic appearance of the medial coronoid silhouette can be useful determining the likelihood of FMCP. The normal medial coronoid process is observed radiographically as a sharply marginated triangular-shaped area of subchondral bone with its silhouette superimposed over the radial head and the joint surface on the extended medial lateral projection (Fig 6A). On the craniocaudal projection, the medial coronoid process is a distinct, triangular process, extending from the proximomedial aspect of the ulna (Fig 6B). In medial coronoid disease, radiographic changes involving the medial coronoid process can include flattening, rounding, proliferation, distinct fragmentation, or an ill-defined margin on 1 or more projections (Fig 7). Use of a distomedial–proximolateral oblique projection of the elbow enhances identification of medial coronoid abnormalities and fragmentation.13 Because of these limitations, radiographic diagnosis of FMCP requires careful, comprehensive assessment of craniocaudal, mediolateral, and oblique projections for all of the potential radiographic abnormalities described above when CED is suspected. However, it is important to recognize that FMCP and other elbow pathology cannot be definitively ruled out based on an absence of radiographic changes.
Despite use of a variety of in vitro and in vivo techniques, radiographic determination of elbow incongruence remains challenging.9,10,12,14–17 Radiographic sensitivity for detection of moderate to severe incongruency (>2 mm) is high, regardless of beam angle15; however, other studies report lower sensitivity because of the complex anatomy of the elbow and limitations associated with identifying a 3-dimensional structure on a 2-dimensional image.9,17 Although Wind18 stated that incongruency could not be affected by positioning others disagree, concluding that radiography was not sensitive enough to evaluate the elbow effectively because of superimposed structures and the influence of positioning.9,17 Mason et al. reported low sensitivity among board certified radiologists when evaluating multiple radiographic projections when asked if the joint was normal or abnormal and evaluate for radio-ulnar incongruence.9 The shape of the trochlear notch is also affected by positioning, therefore caution should be taken when evaluating this finding.17
Elbow registries have been formed to reduce the incidence of elbow dysplasia in the canine population.19,20 Based on radiographic interpretation, elbows are identified as normal or dysplastic with presence of arthrosis, and/or the presence of 1 or more of the following changes: UAP, OC, malformation or FMCP or incongruity.
We performed 2 prospective studies assessing forelimb lameness in dogs21,22 in an attempt to correlate radiographic assessment of CED with clinical signs. FCMP was the most common diagnosis in both studies, and sensitivity, specificity, and likelihood ratios, indicated that radiography was clinically useful for diagnosis of elbow pathology. Because radiography is widely available, efficient, cost effective, and does not typically require general anesthesia, comprehensive radiographic assessment will likely continue to be a valuable component of the diagnostic algorithm for CED. However, the complexity of the elbow joint, substantial variation in radiographic appearances, and an inability to directly assess articular cartilage pathology, make reliance on radiography as a sole diagnostic modality incomplete for optimal clinical progress. Thus other modalities including nuclear scintigraphy, computed tomography (CT), magnetic resonance imaging (MRI), and/or arthroscopy may be necessary for definitive diagnosis.
For CED, scintigraphy has been used for localizing the origin of thoracic limb lameness and/or facilitating detection of early or subtle pathologic changes in the elbow joint23–25 before anatomic changes are evident on radiographs. To date, soft tissue phase imaging has been less for elbow disorders than in other joints,23–25 which may reflect the relatively small amount of soft tissue surrounding the joint, technical limitations, or a true lack of soft tissue pathology in CED. During the bone phase, radiopharmaceuticals are incorporated with hydroxyapatite in bone preferentially localizing where there is increased bone production and vascularity. Because the elbow joint is small and CED often involves the entire joint, scintigraphic changes are typically observed as an overall increase in radiopharmaceutical uptake that is localized to the elbow joint (Fig 8). Focal increased uptake may occur in the proximal aspect of the ulna with UAP, distal humeral condyle for OC, or base of the semilunar notch with FMCP.23 Scintigraphy has high sensitivity for detection of presence or absence of elbow joint disease and for lesion localization to the joint, but no diagnostic benefit over survey radiographs for definitive diagnosis of the specific type of elbow disease.23
99mTechnetium phosphonates are typically used for scintigraphy of joint tissues because of their short half-life, availability, and relatively low cost. Scintigraphic imaging requires a license for handling radiopharmaceuticals, appropriate radiation isolation facilities, personnel trained in radiation handling and safety, and documentation of radioisotope usage. After radiopharmaceutical administration, the patient and any biologic waste must be isolated, the time varying depending on country and state regulations. Consequently, because of these requirements and regulations, nuclear scintigraphy is often limited to academic institutions and large private referral centers, diminishing its availability and application.26 Other disadvantages include the relatively low specificity and relatively poor image resolution compared with other techniques.
In dogs, ultrasonography is selectively used for imaging musculoskeletal soft tissues (tendons, ligaments, menisci, muscles) and less commonly for bone and articular cartilage.27–30 Ultrasound imaging bone and cartilage is limited by depth of penetration and ability to clearly distinguish tissue architecture because of the high acoustic impedance of bone, which is more dense and less compressible than soft tissues. At the soft tissue–bone interface, most sound is reflected leaving relatively little to create an image of the deeper tissues, resulting in a sharp hyperechoic line representing the normal, intact cortical or subchondral bone with distal acoustic shadowing.27 Osteophytes or abnormal bone can be recognized ultrasonographically as irregular, hyperechoic lines, extending from the cortical margins of the bone (Fig 9). Fractures or open physes are recognized by discontinuities in the hyperechoic line of a normal bone margin. Thinning of the hyperechoic line is indicative of a destructive process of bone.
In dogs and cats, ultrasonography yields more clinically useful diagnostic information in larger diarthrodial joints (shoulder, stifle) compared with a smaller joint like the elbow27–30; however, the ultrasonographic appearance of the canine elbow has been reported.27–30 In the normal canine elbow, the medial coronoid process appears as a sharply marginated process along the medial aspect of the joint (Fig 10). With FMCP, the surface of the medial coronoid process is often irregular with proliferation or distinct fragmentation (Fig 11). When medial coronoid process pathology results from incomplete or abnormal endochondral ossification, the medial coronoid process may have the echogenicity of fibrous or soft tissue, instead of normal bone with distal acoustic shadowing (Fig 12). The anconeal process can be identified with sagittal and transverse imaging of the caudal aspect of the joint. UAP can be diagnosed by noting an irregularity or “break” in the cortical bone margin on these projections.29,30 Elbow OC lesions are difficult to image with ultrasound because of their location in the joint in conjunction with the anatomical complexity of overlying cortical bone surfaces and the associated distal acoustic shadowing. To our knowledge, elbow joint incongruity has not been evaluated by ultrasound and would likely be difficult for these same reasons.
Diagnostic use of CT for musculoskeletal imaging has increased.31–33 For CED, the multislice cross-sectional imaging of CT alleviates problems of superimposition improving examination of tissue anatomy and architecture. An ability to adjust the window and level of Houndsfield units (HU) allows simultaneous quality imaging of bone and soft tissue. Three-dimensional image reconstruction is an added enhancement providing additional information on anatomic relationship and extent of changes. Disadvantages include a requirement for general anesthesia, cost of equipment purchase, use, and maintenance, and exposure to ionizing radiation.
Multiple studies have examined the value of CT for diagnostic imaging of the canine elbow.1,8,10–12,31–35 Recommendations for performing elbow CT in dogs include scanning from the point of the olecranon to 2 cm distal to the radial head. Scan thickness should be 1 or 2 mm with an overlapping slice index of 0.5 or 1 mm for reconstructing imaging planes, after initial scan. When evaluating elbow scans, window widths of 1500 and 3500 HU and a window level of 500 HU should be used (Fig 13). CT allows excellent delineation and differentiation of the medial and lateral coronoid processes, the medial and lateral aspects of the humeral condyle, the radial incisure, and radial head. Transverse, sagittal, and dorsal imaging planes at both 1500 and 3500 HU are best for identifying the radial incisure, trochlear defects, subchondral bone (normal or sclerotic), fragments, and incongruity of the humeroulnar, humeroradial, and radioulnar joints. Transverse images at 3500 HU are considered optimal for identifying hypoattenuating lesions (lucencies) of the medial coronoid process.33 UAP and OC have been identified using CT, but because they are typically readily identified on radiographs, there have seemingly been no focused efforts on CT-imaging of these disorders in the canine elbow. With CT, UAP is best seen on a sagittal plane reformatted image as a hypoattenuating line (either partial or complete) between the anconeal process and the proximal aspect of the ulna. OC lesions are best identified on a sagittal or dorsal plane reformatted image as a lucency or flattening of the medial aspect of the humeral condyle with surrounding subchondral bone sclerosis.1,32
CT provides complete imaging of articular subchondral bone but not articular cartilage. Thus, with CT many subchondral bone changes associated with FMCP and OC including sclerosis, fissures, necrosis, cysts, and fragmentation are detectable. Similar to radiographs, the CT appearance of FMCP can be highly variable. Imaging planes and imaging windows/levels can contribute to the variability.33 Medial coronoid process abnormalities identified by CT include abnormal shape, sclerosis, osteophytosis, distinct or separate fragments, fissures or in situ fragments, lucency or hypoattenuation, and associated radial incisure irregularities or lucencies (Figs 14 and 15).1,32 The medial coronoid process ossifies by 22–24 weeks of age in dogs, so if the dog is >6 months and there is hypoattenuating tissue within the medial coronoid process on CT, the medial coronoid process is considered abnormal.
CT has been the most widely used diagnostic modality to assess incongruity of the humeroradioulnar joint and is reportedly affected by positioning.12,17 Radio-ulnar incongruency (RUI) was recently evaluated with the elbow in extension, a standing angle (135°), and in supination, or pronation.12 Pronation had an effect on RUI, resulting in elevation of the apex of the medial coronoid process with the opposite effect noted with supination. On joint extension (160°), there was a cranial translation of the ulna, increasing the space between the radius and ulna at the ulnar incisure.12 Reconstructed images from dorsal and sagittal planes are useful for accurately determining incongruity of the radius and ulna.1,34,35 The most reliable reconstruction plane is the mid-coronoid oblique plane, because it allows the most accurate measurement of radioulnar congruence.10,11
Magnetic Resonance Imaging
For CED, MRI allows imaging in multiple planes (transverse, dorsal, sagittal) without repositioning the patient or reformatting the image information (Fig 16). Furthermore, use of a multitude of sequence types allows better delineation of medullary bone, subchondral bone, soft tissue, cartilage and tissue interfaces.36–38 Whereas CT provides better resolution for imaging bone, MRI is more sensitive for subtle changes in bone architecture including bone marrow lesions (“bone bruising”) and is the only modality allowing tissue differentiation at the bone–cartilage interface.36–38
MRI also has limitations for imaging the canine elbow because of the relatively small size of the joint, complex articulations, and thin articular cartilage surfaces of the humerus, radius, and ulna.37,38 For these reasons, MR arthrography (MRA), using gadolinium-DTPA, has been recommended for imaging the canine elbow.39 MRA allows improved lesion identification and classification for FMCP and subchondral lesions.39 Gradient echo fast imaging with steady-state precession (GE FISP) sequence is most useful when compared with spin echo and fat saturation sequences.38 Use of a small surface coil configuration is also highly recommended for CED to get improved signal-to-noise ratio. All MRI planes, dorsal, sagittal, and axial/transverse, are potentially useful for diagnosis of elbow disorders.
To our knowledge, MR diagnosis of UAP has only been described in a single case,32 where UAP was seen as a hyperintense cleft between the proximal aspect of the ulna and the anconeal process on sagittal plane, 3-dimensional Fourier transform echo gradient fast imaging sequence. Similarly, MR characteristics of canine elbow OC lesions have not been well described. OC lesions would be best evaluated in sagittal and dorsal planes, and would be seen as thickening of the articular cartilage of the medial aspect of the humeral condyle, flattening of the condylar surface, and/or cartilage erosions/irregularities most often with associated subchondral bone lesions. Small articular cartilage lesions without subchondral changes may be difficult to definitively image because of an inability to clearly distinguish the thin apposing articular cartilage surfaces in this region.32,37–40 Sensitivity for detecting lesions on the medial humeral condyle with MR was 77% and 72% for radiography whereas for FMCP, accuracy of MR for detection of medial coronoid abnormalities was 95.5% compared with 77% for radiography.40 MR was 91% sensitive for detection of nonfragmented, nonmineralized or mineralized medial coronoid processes.40 Administration of gadolinium provided no additional benefit for detection of mineralized and unmineralized, nondisplaced coronoid processes. These authors stated that the detection may have been improved by allowing more time and manipulating the joint through a complete range of motion so that gadolinium could fill the spaces between the fragments (Fig 17).
Elbow incongruity using MR has been assessed examining the relationship between the humeral trochlea and ulnar notch at multiple locations, and differences between large-breed, small-breed, and chondrodysplastic breed dogs.36 In large-breed dogs, there was a smaller articular gap at the level of the anconeal process and a wider gap at the center of the trochlear notch. It was concluded that large-breed dogs may not have the ability to compensate for incongruity between the radius and ulna, resulting in the potential for failure of normal endochondral ossification of the anconeus and coronoid process.
Compared with disorders of the canine shoulder and stifle, elbow MRI is still in its infancy and not routinely performed. Correlations of canine elbow MR findings with arthroscopic, gross, and/or histologic examinations are needed to improve diagnosis and understanding of CED.
Comparison of Modalities
To our knowledge there has been no comprehensive direct comparison of the all imaging modalities for diagnosis of CED. Arthroscopy has been compared with CT and radiography for assessment of CED41,42 and RUI.12 Arthroscopy (sensitivity, 94%; specificity, 81.9%) is reportedly superior to CT (sensitivity, 85%; specificity, 45.8%) and radiography (sensitivity, 99.3%; specificity, 42.4%) for diagnosing RUI. Using surgical findings as the reference standard for FMCP, comparison of different radiographic techniques (plain film, xeroradiography, linear tomography, arthrography, CT) identified CT as most accurate (86.7%), sensitive (88.2%), and with the highest negative predictive value (84.6%).41 A combination of plain-film radiography with linear tomography improved accuracy, approaching that of CT alone.41
A recent comparison of CT with arthroscopy for assessment of FMCP showed that these procedures were complementary for MCP assessment. Identification of an MCP fragment on CT was significantly correlated with finding a fragment arthroscopically. Fragmentation of cartilage alone, as well as nondisplaced fragments, were often not detected by CT imaging, but were detected using probing on arthroscopic assessment. Microcracks and fissures diagnosed consistently with CT imaging were often not detected during routine arthroscopy but could be verified when burring was used based on CT data. Another important finding of these investigators was that incompletely mineralized osteophytes at the apex of the MCP that are associated with the joint capsule may mimic FMCP on CT.42 Direct comparison of MRI to CT and arthroscopy is needed to determine utility for early diagnosis and clinically relevant characterization of all components of CED.