Pseudoerosions of Hands and Feet in Rheumatoid Arthritis: Anatomic Concepts and Redefinition

Rheumatoid arthritis is a chronic inflammatory disease characterized by the development of osseous and cartilaginous damage. The correct differentiation between a true erosion and other entities—then often called “pseudoerosions”—is essential to avoid misdiagnosing rheumatoid arthritis and to correctly interpret the progress of the disease. The aims of this systematic review were as follows: to create a definition and delineation of the term “pseudoerosion”, to point out morphological pitfalls in the interpretation of images, and to report on difficulties arising from choosing different imaging modalities. A systematic review on bone erosions in rheumatoid arthritis was performed based on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. The following search terms were applied in PubMed and Scopus: “rheumatoid arthritis”, “bone erosion”, “ultrasonography”, “radiography”, “computed tomography” and “magnetic resonance imaging”. Appropriate exclusion criteria were defined. The systematic review registration number is 138826. The search resulted ultimately in a final number of 25 papers. All indications for morphological pitfalls and difficulties utilizing imaging modalities were recorded and summarized. A pseudoerosion is more than just a negative definition of an erosion; it can be anatomic (e.g., a normal osseous concavity) or artefact-related (i.e., an artificial interruption of the calcified zones). It can be classified according to their configuration, shape, content, and can be described specifically with an anatomical term. “Calcified zone” is a term to describe the deep components of the subchondral, subligamentous and subtendinous bone, and may be applied for all non-cancellous borders of a bone, thus representing a third type of the bone matrix beside the cortical and the trabecular bone.


Introduction
Rheumatoid arthritis (RA) manifests with three types of structural joint damage: joint space narrowing, erosions, and capsular abnormalities in the form of synovial proliferation and subluxations [1][2][3][4]. The diagnosis of erosions and their quantification as part of radiographic scoring systems is an accepted surrogate biomarker of structural progression of arthritis [4,5]. Erosions in RA have been defined in consensus statements and in studies with high-resolution peripheral quantitative computed tomography (HRpqCT) as cortical defects, breaks, or other discontinuities with underlying trabecular bone loss and characteristic locations that can be identified with imaging [6][7][8][9][10][11][12].
The search with the defined terms resulted in a total of 1487 results. An additional number of 59 papers were added after reference-screening. The flow diagram of the literature review may be seen in Figure 1. Ultimately, only 25 papers reported specifically on false-positive results or erosion-like changes.
inflammatory diseases, clinical studies comparing therapeutic measurements in RA, studies comparing the sensitivity of imaging modalities without report of false positive diagnosed erosions or erosion-like lesions, surgical procedures or longitudinal studies without direct reference to this topic, case reports and conference papers. Additionally, all papers without full text availability were excluded from the analysis.
Data extraction was performed by using a standardized Excel (Microsoft Corporation, Redmont, WA, USA) data extraction form: first author, year of publication, country, study population, number of patients, imaging modality, joints evaluated, reported sensitivity of imaging modalities, reported false positive or false negative diagnosis of bone erosions, reported limitations in image interpretation with respect to anatomy, the differential diagnosis to other erosive diseases, artifacts and signal-tonoise reduction.

Results
The search with the defined terms resulted in a total of 1487 results. An additional number of 59 papers were added after reference-screening. The flow diagram of the literature review may be seen in Figure 1. Ultimately, only 25 papers reported specifically on false-positive results or erosion-like changes. Based on the information gathered in the remaining papers, the false-positive results were subdivided into anatomic pseudoerosions, if the explanation for the false-positive diagnosis was described as a morphological phenomenon, and into artifact-related pseudoerosions, if the explanation for the false-positive diagnosis was related to the respective imaging technique.

Anatomic Pseudoerosions
Anatomic pseudoerosions, i.e., normal concavities of a bone with a potential for misinterpreting them as arthritis-related erosions, were described in twelve original papers and reviews and may be classified into four types according to their anatomic form and configuration (Table 1): (1) a groove or notch or its incomplete form, i.e., a jutty, (2) a sulcus as part of an osteofibrous channel, (3) a subcapital neck on long bones, or (4) a nutritional channel or a zonal roughness [3,11,41,[51][52][53][54][55][56][57]. According to their shape, they may be grouped into (1) shallow or broad concavities and (2) subchondral cysts, if en-face displayed on an image and occasionally with a small opening to the joint space, or (3) channel-like structures (Figure 2a,c) [3,54,55]. The anatomic location of pseudoerosions is predominantly at the carpal bones, the MCP-and the MTP-joints. Almost always they are linked to a ligament insertion (Figure 2b), a mucosal fold fixation or the hood of a tendon sheath, and occur at the noncortical bone, also known calcified zones (i.e., borders of the subchondral and enthesial calcified bone with the adjacent underlying trabecular structures). The content of pseudoerosions is visible with US and MRI and may be normal or degenerated ligament tissue, or blood vessels [44,56] and the development of edematous changes [58]. With contrast media, a slight enhancement can be observed, however, in one publication rare cases of strong enhancement was documented [56].

Anatomic Pseudoerosions
Anatomic pseudoerosions, i.e., normal concavities of a bone with a potential for misinterpreting them as arthritis-related erosions, were described in twelve original papers and reviews and may be classified into four types according to their anatomic form and configuration (Table 1): (1) a groove or notch or its incomplete form, i.e., a jutty, (2) a sulcus as part of an osteofibrous channel, (3) a subcapital neck on long bones, or (4) a nutritional channel or a zonal roughness [3,11,41,[51][52][53][54][55][56][57]. According to their shape, they may be grouped into (1) shallow or broad concavities and (2) subchondral cysts, if en-face displayed on an image and occasionally with a small opening to the joint space, or (3) channel-like structures (Figure 2a,c) [3,54,55]. The anatomic location of pseudoerosions is predominantly at the carpal bones, the MCP-and the MTP-joints. Almost always they are linked to a ligament insertion (Figure 2b), a mucosal fold fixation or the hood of a tendon sheath, and occur at the noncortical bone, also known calcified zones (i.e., borders of the subchondral and enthesial calcified bone with the adjacent underlying trabecular structures). The content of pseudoerosions is visible with US and MRI and may be normal or degenerated ligament tissue, or blood vessels [44,56] and the development of edematous changes [58]. With contrast media, a slight enhancement can be observed, however, in one publication rare cases of strong enhancement was documented [56].

Discussion
From the viewpoint of imaging anatomy, a misinterpretation of erosions in RA may occur due to (1) anatomic pseudoerosions, or (2) artifact-related pseudoerosions as a result of an inadequate investigation technique. Pseudoerosions and erosions are commonly located at certain areas of the surface outline of the calcified bone, also known as calcified zones. These may therefore, besides cortical bone and trabecular bone, be regarded as a third type of organization of the bone matrix.
The term "calcified zones" (Figure 3) in this context is therefore proposed to describe the borders of the subchondral and enthesial calcified bone with the adjacent underlying trabecular structures. It may be extended for describing all parts of intraarticular bone apart from the cortex. With its overlying tissue of hyaline cartilage, synovium or capsule-ligamentous structures it forms anatomic units. The relationship between these zones and the adjacent tissues is so tight that the fibrous layers of tendon sheaths, bursae, periosteum or the cartilaginous zones of entheses or hyaline cartilage are in direct continuation with the subjacent bone, thus providing direct contact with synovial tissue. The concept of the subchondral zone was used by Dihlmann [71] to describe the mineralized zone of hyaline cartilage as part of the subchondral bone. It may be extended to describe a subligamentous, subtendinous or subbursal zone of the bone. Utilizing sub-millimeter spatial resolution CT, these calcified zones can be displayed. Differentiating the normal calcified zone from erosional changes, i.e., irregular margins and sclerotic reaction, is the main challenge in differentiating true erosions from pseudoerosions [72].

Anatomic Pseudoerosions
An anatomic pseudoerosion can be defined as a normal concavity of a bone outlined by a smooth and thin calcified zone with the potential for a false-positive misinterpretation of an erosion. In this form, the term pseudoerosion is more precise than "notch" or "bony depression" and may be preferred as it contains a prognostic impact for the imaging assessment of arthritis. Such clinically oriented annotations, examples are the scaphoid waist and the metacarpal neck as typical sites for fractures, have been in use in traumatology and may be of help in the assessment of arthritis-related erosions, too (list of described pseudoerosions in Table 3, an overview of anatomical pseudoerosions in the hand may also be found in Figure 4). Pseudoerosions have to be differentiated from other pathologies as ganglion cysts, crystal-induced arthropathies, tuberculosis or other infections, and from degenerative lesions in the form of erosions, subchondral (pseudo)cysts or beak-shaped osteophytes as there are so many similarities in location [38,60]. Intraosseous ganglion cysts are common and almost always have a continuity with a ligament which underwent mucous degeneration [73,74]. Especially in the elderly population, the more prevalent degenerative changes of the bone may be difficult to be differentiated from RA-related erosions [38,75]. However, in children interpretational problems may arise. There, normal concavities simulating erosions have been referred to as "bony depressions" at certain locations in the wrist [76][77][78]. Such pseudoerosions in children may be big, indicating that size is not a reliable feature for differentiating normal variants from true erosions.

Anatomic Pseudoerosions
An anatomic pseudoerosion can be defined as a normal concavity of a bone outlined by a smooth and thin calcified zone with the potential for a false-positive misinterpretation of an erosion. In this form, the term pseudoerosion is more precise than "notch" or "bony depression" and may be preferred as it contains a prognostic impact for the imaging assessment of arthritis. Such clinically oriented annotations, examples are the scaphoid waist and the metacarpal neck as typical sites for fractures, have been in use in traumatology and may be of help in the assessment of arthritis-related erosions, too (list of described pseudoerosions in Table 3, an overview of anatomical pseudoerosions in the hand may also be found in Figure 4).    Table 3. Right skeletal hand, on the left view from palmar, on the right view from dorsal.
Grooves due to ligament or tendon insertions have a varying appearance as described in the enthesis concept by Benjamin and McGonagle [79]. Such prominent grooves can cause the appearance of a pseudoerosion (Figure 2a). A groove may occur in three forms: (1) at a nonapophyseal direct tendon or ligament attachment where the uncalcified components of the enthesis enters the bone, (2) at an apophysis with overhanging edges, or (3) at an incomplete apophysis, a jutty, at the indirect attachments of a tendon or ligament with a tangential transition into the  Table 3. Right skeletal hand, on the left view from palmar, on the right view from dorsal. Grooves due to ligament or tendon insertions have a varying appearance as described in the enthesis concept by Benjamin and McGonagle [79]. Such prominent grooves can cause the appearance of a pseudoerosion (Figure 2a). A groove may occur in three forms: (1) at a non-apophyseal direct tendon or ligament attachment where the uncalcified components of the enthesis enters the bone, (2) at an apophysis with overhanging edges, or (3) at an incomplete apophysis, a jutty, at the indirect attachments of a tendon or ligament with a tangential transition into the periosteum. For example, pseudoerosions resulting from the first form are the metacarpal ligament insertions at the bases of the metacarpal bones [80]. At the dorsal aspect of the triquetral bone, such a pseudoerosion may be formed by the distal insertion of the radiotriquetral ligament along with other components of the dorsal radiocarpal ligament. On the capitate, on which several strong carpal ligaments have their insertion, and many other carpal bones, intercarpal ligaments may cause pseudoerosions [51]. Examples for the second form may be the non-spherical form of metacarpal and metatarsal heads, which can be explained by the collateral ligament complexes running laterally and medially with smoothly outlined shallow metacarpal grooves containing these structures. At the metacarpals, these grooves are bordered by little tubercles for the proximal attachment of the collateral ligaments ( Figure 4) [81]. Moraes do Carmo et al. [54] identified three concavities in the first metacarpal head (intersesamoid, ulnar, and radial) and two in those of the fingers (ulnar and radial). They described dorsal depressions of the metacarpal heads due to the extensor digitorum tendons in one third of their anatomic specimens which correlated with observations with ultrasound made by Boutry et al. [82,83]. A similar study was done for defining pseudoerosions of the metatarsal heads by Torshizy et al. [55] who described anatomic variations in the normal osseous concavities of the lateral and medial aspects of each metatarsal head. Typical jutties, i.e., examples for the third form of grooves, are the small round or oval subcapsular notches at the proximal phalangeal bases [80,84]. At the Achilles tendon insertion, proximal to its jutty shallow irregularities beneath the calcaneal bursa may represent true erosions [85].
Osseous sulci are commonly roofed with a ligament, fascia or other fibrous tissue, thus forming an osteofibrous channel for a tendon within a synovial tendon sheath. A subcapital neck of the distal metacarpal and the metatarsal bones is a small metaphyseal narrowing that may cause a pseudoerosion on projection radiographs, ultrasound or MRI [86]. At the distal fifth metacarpal bone, due to its slight varus angulation this neck may be more prominent.
Nutritional channels may appear as pseudoerosion on MR if their orifice is displayed as a little T2-weighted hyperintense spot [51]. Their superficial orifice is often located at a roughness of the calcified zones which as a whole may simulate an erosion [11,34,87]. Some of these iuxtaarticular surface roughnesses may be specified as crests or ridges that correspond to attachment sites for redundant joint capsule [55]. Others, especially on carpal bones, may be due to indentations of innominate ligamentous attachments or synovial folds [51]. Such typical structures visible between the radial aspect of the scaphoid and the radial carpal collateral ligament may be called scapho-capsular ligaments ( Figure 2C). Roughness of the calcified zones may be visible at various sites and should be differentiated from shallow extensive true erosions and from advanced cartilage degeneration [88].

Artefact-Related Pseudoerosions
Artefact-related pseudoerosions are defined as an interruption of the sharp outline of the calcified zones. Important causes are a low signal-to-noise ratio, a partial volume artefact, or in case of ultrasound irregular backscattering with artefacts on an incongruent or rough surface. A low signal-to-noise ratio could be caused by over-penetration of the X-ray beam through the bone or due to insufficient spatial or contrast resolution. This effect is more severe in cases with low calcium content in the calcified zones or the subjacent trabecular bone, previously referred to as subchondral osteoporosis or as pre-erosions, and may be enhanced by swelling of the overlying soft tissue. With ultrasound, diffraction or a complex backscattering of the waves on a curved or irregular surface may cause various pseudo-effects on the retrieved image [51,69,89].
Although X-ray is most commonly used in the diagnosis of RA it is CT which can be regarded as the best imaging modality for differentiating pseudoerosions from true erosions [53,62,63,[90][91][92][93][94]. Several studies [34,[95][96][97][98] describe a significant decrease of trabecular volume and number and an increased trabecular heterogeneity in patients with rheumatoid arthritis by using HRpqCT. This trabecular bone loss as the intramedullary component of bone erosions may contribute the largest part and may therefore be a reason for misinterpretation of erosions or pseudoerosions in radiographs as this imaging method is relatively insensitive to trabecular bone loss [60,99].
In addition, MRI and US are reported to be more sensitive than plain radiography [53,62,[90][91][92][93][94], but this especially seems to be dependent on the location investigated [88,100]. In some cases, radiography may even be superior to MRI in detecting bony erosions despite its lack of three dimensionality [1,3,5,58,99,101,102]. Through its high spatial resolution it can differentiate smaller erosions which otherwise would present themselves as continuous on MRI [61,65].
Thus, it is important to recognize several parameters to achieve a decrease of cognitive diagnostic errors especially in early arthritis. These include slight variations in the respective projection technique and individual ligament laxity or postinflammatory scarring of ligaments. In addition, the roughness of a calcified zone, and the transitional changes between normal bone and true inflammatory erosions are until now not or only scarcely addressed. Even the projection of the joints, even if the relevant anatomic landmarks are displayed according to the standards, is highly variable. One has also to keep in mind that discrete forms of malalignment due to ulnar deviations or other forms of arthritic subluxation, ligament laxity with a slight rotation of bones, and variations in their arrangement may cause a more prominent appearance of a pseudoerosion [3,51,57,102].

Erosions-in-Pseudoerosions
Both anatomic and artefact-related pseudoerosions are located at sites with direct or indirect contact to inflammatory tissue in arthritis, and therefore, are at higher risk for destruction. Areas of the articular bone without any cartilage covering are more prone to erosive destructions by synovial tissue and effusion [3,32]. Hence, in an anatomically preformed concavity a true inflammatory erosion may develop. McQueen et al. [51] described these erosions-in-pseudoerosions ( Figure 5) for the attachment sites of the intercarpal ligaments. It may be observed at the site of ligamentous attachments covered by synovial folds at the metacarpal or metatarsal heads or at the wrist. On the other hand, true erosions may be classified as normal variants. It remains unclear whether these are incidental findings or subclinical erosions [3,57,102].  [61,65]. Thus, it is important to recognize several parameters to achieve a decrease of cognitive diagnostic errors especially in early arthritis. These include slight variations in the respective projection technique and individual ligament laxity or postinflammatory scarring of ligaments. In addition, the roughness of a calcified zone, and the transitional changes between normal bone and true inflammatory erosions are until now not or only scarcely addressed. Even the projection of the joints, even if the relevant anatomic landmarks are displayed according to the standards, is highly variable. One has also to keep in mind that discrete forms of malalignment due to ulnar deviations or other forms of arthritic subluxation, ligament laxity with a slight rotation of bones, and variations in their arrangement may cause a more prominent appearance of a pseudoerosion [3,51,57,102].

Erosions-in-Pseudoerosions
Both anatomic and artefact-related pseudoerosions are located at sites with direct or indirect contact to inflammatory tissue in arthritis, and therefore, are at higher risk for destruction. Areas of the articular bone without any cartilage covering are more prone to erosive destructions by synovial tissue and effusion [3,32]. Hence, in an anatomically preformed concavity a true inflammatory erosion may develop. McQueen et al. [51] described these erosions-in-pseudoerosions ( Figure 5) for the attachment sites of the intercarpal ligaments. It may be observed at the site of ligamentous attachments covered by synovial folds at the metacarpal or metatarsal heads or at the wrist. On the other hand, true erosions may be classified as normal variants. It remains unclear whether these are incidental findings or subclinical erosions [3,57,102].

Limitations
A limitation of this study was that the defined search terms resulted in a large quantity of papers, which had to be screened. However, generally accepted terms for mimickers of true erosions do not exist, are described in various forms and additionally with more equivocal definitions than expected at the beginning of this project. Nonetheless, this wide search net allowed for the inclusion of all relevant sources describing the phenomenon of pseudoerosions and minimized the possibility of

Limitations
A limitation of this study was that the defined search terms resulted in a large quantity of papers, which had to be screened. However, generally accepted terms for mimickers of true erosions do not exist, are described in various forms and additionally with more equivocal definitions than expected at the beginning of this project. Nonetheless, this wide search net allowed for the inclusion of all relevant sources describing the phenomenon of pseudoerosions and minimized the possibility of excluding the respective publications.

Conclusions
In conclusion, a pseudoerosion is more than just a negative definition of an erosion. It can be defined as a normal osseous concavity (anatomic pseudoerosion) and/or an artefactual interruption of the calcified zones (artefact-related pseudoerosion). It can be classified according to their configuration, shape, content, and can be directly anatomically named. "Calcified zone" is a term to describe the deep components of the subchondral, subligamentous and subtendinous bone and may be applied for all non-cancellous borders of a bone, thus representing a third type of the bone matrix beside the cortical and the trabecular bone. Anatomic pseudoerosions are almost always related to a ligament insertion or the osteo-fibrous channel of a tendon sheath, therefore, being of high risk for microdamage and the development of a "true" arthritic erosion. Understanding these peculiar aspects of the bony surface with relation to ligament insertions and osteofibrous channels may be of help in improving the assessment of erosions and for reducing over-and underdiagnosis of true erosions.

2.
The term "calcified zone" describes the deep components of the subchondral, subligamentous and subtendinous bone and may be applied for all non-cancellous borders of a bone.

3.
Pseudoerosions can be regarded as anatomic sites at risk for the development of "true" arthritic erosions.

Conflicts of Interest:
The authors declare no conflicts of interest.