Authors: Ummugulsum Gazel, Gizem Ayan, Nicole Hryciw, Jean-Philippe Delorme, Elliot Hepworth, Marcos Sampaio, Zaid Jibri, Jacob Karsh, Sibel Zehra Aydin
Categories: Systematic Review and Meta-Analysis, new bone formation, radiography, spine, AcademicSubjects/MED00010
Source: Rheumatology Advances in Practice
Doi: 10.1093/rap/rkae061
We aimed to explore the radiographic definitions of types of New Bone formation (NBF) by focusing on the terminology, description and location of the findings.
Three systematic literature reviews were conducted in parallel to identify the radiographic spinal NBF definitions for spondyloarthritis (SpA), Diffuse Idiopathic Skeletal Hyperostosis (DISH) and Osteorathritis (OA). Study characteristics and definitions were extracted independently by two reviewers. Definitions were analysed and collated based on whether they were unique, modified or established from previous research.
We identified 33 studies that indicated a definition for the NBF in SpA, 10 for DISH and 7 for spinal OA. In SpA, the variations in syndesmophytes included the description as well as the subtypes and locations. The differentiation of syndesmophytes from osteophytes were included in 12 articles, based on the origin and the angle of the NBF and associated findings. The definitions of DISH varied in the number of vertebrae, level and laterality. For OA, five articles indicated that osteophytes arose from the anterior or lateral aspects of the vertebral bodies, and two studies required a size cut-off.
Our ultimate aim is to create formal NBF definitions for SpA, DISH and OA guided by an atlas, through a Delphi exercise with international experts. The improved ability to differentiate these conditions radiographically will not only allow the clinicians to accurately approach patients but also will help the researchers to better classify patient phenotypes and focus on accurate radiographic outcomes.
Keywords: new bone formation, spine, radiography
New bone formation (NBF) and destruction cycles provide normal bone turnover in a balanced and continuous manner. Osteoblasts play the main role in synthesis of new bone to maintain bone homeostasis. NBF occurs as a result of tissue repair mechanisms mediated by inflammatory, non-inflammatory and biomechanical forces [1].
Despite increasing research in this area, the cellular and molecular processes of NBF in humans are poorly understood. Bone morphogenic protein and wingless-type-like signalling pathways found to have a critical role in NBF in animal studies in spondyloarthritis (SpA) [2]. Furthermore, the complex bone remodelling mechanism of osteoblastic NBF with the interaction of cellular proliferation, differentiation, maturation, migration and cell death has been described in SpA. However, these osteoblastic changes are not specific to SpA, and a similar remodelling process can be seen in the degenerative changes in the spine leading to spinal damage Osteorathritis (OA) and Diffuse Idiopathic Skeletal Hyperostosis (DISH) [3, 4].
Regardless of the underlying mechanism, processes of NBF can sometimes occur in the same patient [5, 6]. Therefore, it can be challenging to differentiate NBFs in SpA, OA and DISH by conventional radiography. Moreover, these conditions can frequently coexist, which has been demonstrated to be related to a poorer outcome, and the management can be challenging [5, 7]. For example, marginal syndesmophytes can be mistaken for the early osteophytes, and bridging osteophytes mimic bridging para-marginal syndesmophytes [8, 9]. The ability to distinguish these entities through radiographic imaging holds clinical significance due to the diverse prognoses and differences in the treatment approaches associated with each. Differentiating the aetiology of NBF is also very important for the research on the pathogenesis of these diseases. This may be problematic in clinical trials that use radiographic progression on the spine as an outcome and has the potential to result in high rate of measurement error.
We believe an important element in establishing a comprehensive definition of radiographic spinal NBF is to review the radiographic definitions that are currently utilized in the literature. Our ultimate aim is to have a consensus on the definitions of NBF in order to differentiate NBF types from each other. We conducted three parallel systematic literature reviews (SLR), aiming to explore the definitions of various types of NBF in DISH, OA and SpA and combined results in this article. We will use the results to inform an international Delphi through an atlas to establish criteria to identify and differentiate NBF in each of these disease processes, which will be followed by prospective research to validate our efforts.
Three separate SLRs were performed by using a predefined PICO (population, intervention, comparator and outcome) strategy. MEDLINE, EMBASE and Cochrane Central Register databases were reviewed for publications between January 1980 and November 2023 by an experienced librarian (RS) at The University of Ottawa. Our research questions were identified as ‘What are spine NBF radiography findings in patients with 1) DISH, 2) spine OA, 3) PsA, 4) AS, 5) SpA?’
Search strategies have been developed separately for these research questions. The following terms have been used for the literature search; ‘PsA’ OR ‘AS’ OR ‘Spondyloarthritis’ OR ‘Spine osteoarthritis’ OR ‘DISH’ AND ‘Radiography’ AND ‘Spine’ as MESH terms and text terms. Review process was done after merging the results of three diseases under the SpA topic.
The protocols have been registered to the International Prospective Register of Systematic Reviews database (Registration numbers for DISH: CRD42020197545, for OA: CRD42020197584, for SpA: CRD42020197760). Details of search strategies are given in Supplementary Tables S1–S3, available at Rheumatology Advances in Practice online.
The titles and abstracts were independently screened by two reviewers (UGG&GA for SpA, NH&JD for DISH and OA). All abstracts with discrepancies were carried forward to a full-text review, to be as inclusive as possible, and the full texts were reviewed independently by the same investigators. Any disagreement at the stage of full text review was resolved by the third investigator (SZA). Articles that do not fulfil the inclusion criteria were identified, and the reason for exclusion was documented. Additionally, references of the included articles were manually searched. All screening processes for three SLRs were presented in a flow chart (Fig. 1). To be eligible for inclusion, studies had to meet the following criteria; either cross-sectional, case–control, cohort, observational (retrospective or prospective) study designs, literature reviews and case studies with more than 10 patients; studies including patients >18 years old with a diagnosis of SpA (AS, PsA, inflammatory bowel disease-associated arthritis, reactive arthritis), DISH or OA with the descriptions of axial plain radiographic NBF features and utilized this definition for diagnosis. Studies were excluded if they were in a language other than English, the wrong study type, with no displayed or inaccessible data, duplicate study population, the wrong outcome or modality (e.g. computed tomography or magnetic resonance imaging) or if only the abstract was available.
Figure 1. Flow chart for study selection
After identifying of the articles to be included, data were extracted in parallel by two independent reviewers using a standardized sheet (UGG&GA for SpA, NH&JD for DISH and OA). Any discrepancies within the data extraction phase were resolved by discussion with a third reviewer (SZA). The descriptions of axial plain radiographic NBF features in patients with DISH, OA and DISH were the primary outcomes.
The location (cervical/thoracic/lumbar) and distribution of age groups of the defining disease types were the secondary outcome measures.
The results of the SLRs are presented separately for SpA, DISH and OA:
Our literature search identified 32 studies (11 original studies and 21 review articles), which included a definition of NBFs on the spine radiography of SpA patients. The diagnostic subgroups, sample size, age distribution and spinal region in the original studies, as well as the definitions, were displayed in Table 1. For the review articles, definitions and disease subgroups were summarized in Supplementary Table S4, available at Rheumatology Advances in Practice online.
Our literature search was able to identify only one description of syndesmophytes, that was used in 13 studies with only minor modifications [10–22]. According to that, a syndesmophyte was defined as ‘bony overgrowth (protuberances/projections) along the anterior longitudinal ligament or ossification within the outer fibres/layers of the annulus fibrosus’. In the majority of these articles, syndesmophytes’ shape was specified as thin and the orientation of the growth as vertical. Six of the 12 articles mentioned that syndesmophytes may also connect the angles of adjacent vertebral bodies or connect two vertebral bodies across the disc space, leading to bridging phenomena, although this was not a mandatory feature of the definition [12–15, 19, 21].
** *Marginal * ** Seven articles used a specific terminology of ‘marginal syndesmophytes’ [14, 23–28]. Six of these articles defined the marginal syndesmophytes as ‘vertebral ossifications/calcification/bony outgrowth arose from the edge of the vertebral body vertically and extend from the corner of one vertebra to the next’. Mattar et al. [14] defined the marginal syndesmophytes as ‘horizontal projections at the level of the vertebral end-plate, with its cortex and medulla continuous with those of the parent bone’. Also, three of seven articles additionally described them as ‘being thin’ [23, 27, 28].
** *Non-marginal (para-marginal) * ** Within the included studies, eight articles had a definition for non-marginal (para-marginal) syndesmophytes [23, 26–32]. Three out of eight articles described the growing pattern of these NBFs as syndesmophytes arising from beyond/away from the edge/margin of the vertebral body [23, 26, 31]. Five articles stated that these bony growths are curvilinear. Also, the following features for the shape of para-marginal syndesmophytes were mentioned in the asymmetrical, thick, bulky, fluffy and chunky. While three articles described these ossifications as being parallel to the vertebral bodies or intervertebral discs [29, 31, 32], Eshed et al. [27] defined them as horizontally oriented syndesmophytes.
** *Paravertebral * ** Four articles included the definition of paravertebral ossification. These ossifications were defined as being close to the vertebra; however, with a gap between the margins of the ossification and the vertebra [23, 28]. Also, Klecker et al. [31] described it as *‘*coarse asymmetrical bony bridging, and relative sparing of the apophyseal joints’.
** Squaring: ** Squaring of vertebral borders was described as a result of erosive changes at the corners of the vertebrae and straightening of the anterior curve of the vertebra by NBF. This lesion is defined as a typical feature of AS and is best visualized in the lumbar spine [12, 33, 34].
** Finer ossification **: It was separately defined only by Porter et al. as ‘more closely related to the disc margins and fusing with the rim of the vertebral body’.
Among 11 original articles, five investigated the syndesmophytes on the cervical and lumbar spine, while six articles included the thoracic spine as well (Table 1). For para-syndesmophytes, three articles specifically indicated that the lower thoracic and upper lumbar spine or thoracolumbar junction were more commonly involved than the cervical and lower lumber spine [31, 32]. However, Sudol-Szopinska et al. [29] mentioned that cervical involvement may be typical for the para-marginal syndesmophytes in PsA.
For the other lesions, in two articles, squaring of vertebra was mentioned as they can be best visualized in the lumbar spine due to the concavity of the lumbar spine compared with the cervical and thoracic spine [33, 34].
A differentiation between syndesmophytes and degenerative changes was made in 12 articles [5, 12, 13, 18, 21, 26, 35–40]. According to those articles, syndesmophytes originated at the ligamentous insertion and the growth was parallel to the anterior vertebral side/anterior intervertebral ligament, whereas osteophytes originated from the cartilaginous endplate, with a horizontal growth and was associated with disc space narrowing.
In parallel to this explanation, in five articles, an angle of 45° was used to differentiate, with SpA-related changes having an angle of ≤45° to the anterior vertebral side and an angle of > 45° being representative of degenerative changes [5, 12, 37, 38, 40].
Non-marginal syndesmophytes were indicated to be more typical for PsA [27, 29]. The main differences for NBF in PsA compared with other SpA entities were larger, asymmetric distribution with skipped vertebral bodies levels, unilaterality and separation from the lateral aspect of the vertebral bodies of syndesmophytes [27, 29, 40]. Also, in Reijnierse et al.’s [12] study, ossification in AS was specified as in the outer layers of the annulus fibrosus itself, which results as intervertebral bridging, while PsA-related ossification was indicated as paraspinal and separated from the vertebral bodies and discs.
Our literature search identified several dichotomous variations in the radiographic definition of spinal DISH, that could be grouped into 10 definitions in total (Table 2).
** *Number of * **Within the 10 dichotomous definitions outlined, five definitions [41–45] required the involvement of at least three contiguous vertebrae (or two intervertebral bridges), and three definitions [4, 5, 46] required four contiguous vertebrae to classify as changes as consistent with DISH. Two studies [47, 48] did not require any specific number of contiguous vertebrae as part of their definition of DISH.
** Level of NBF: ** Five out of 10 definitions [5, 41, 43, 44, 48] mandated that lesions of NBF consistent with DISH were found on the thoracic spine, and one definition [45] required the presence of changes in the thoracic or lumbar spine. None of the definitions specified criteria for cervical spine changes. There were four definitions [4, 42, 46, 47] that did not specify a spinal level of involvement to make the diagnosis of DISH.
** Laterality: **Two out of 10 definitions required the ossifications to be present on the right side [5, 48].
Several descriptions of the bony lesions of DISH were found in the identified dichotomous definitions. The most common description specified changes as flowing or bridging ossifications/calcifications, found in seven of the definitions [4, 5, 37, 41, 43, 46]. Other descriptions included were exuberant osteophytosis [47], flame-shaped anterolateral bony bridges [45] and massive vertical osteophytes [48], all used in one definition each.
Most of the definitions differentiated DISH-related changes from OA by requiring a normal/relatively normal disc space. Preservation of disc height was mandated in five definitions [4, 41, 43, 47, 48], one of which specifically included ‘widening intervertebral disc space’ [47]. None of the studies included differentiation of NBFs of DISH and OA. One study described the NBF in DISH as osteophytes.
Three definitions [4, 41, 43] required the absence of sacroiliitis in order to make a definitive diagnosis of DISH [49, 50], with each of these definitions also including the absence of ankylosis in the facet joint and one in the apophyseal joint [43]. Outside of the 10 dichotomous definitions identified by our search, one study modified the classically accepted Resnick criteria to help differentiate the NBF of DISH and SpA based on the angle of new bone growth from the vertebrae, which was then used in another study [7, 37]. In both studies, a growth angle of >45° from a vertebral body was felt to be in keeping with DISH-related changes, whereas bony growth of ≤45° was felt to be in keeping inflammatory changes, either from PsA [7] or AS [37].
There were seven studies identified that provided a definition for the identification of osteophytes in the context of spinal OA (Table 3).
** Location of NBF: **A total of five articles indicated that osteophytes arose from the anterior or lateral aspects of the vertebral bodies, with only two of these studies specifying that osteophytosis could also occur at other locations, including the posterior, superior and inferior margins [51, 52].
** Size of NBF: **Two studies required a size cut-off in their definitions of osteophytes in the context of OA. Pfirrmann et al. [53] classified large osteophytes as those with an ‘anteroposterior diameter greater than 3 mm’. Another study required the presence of osteophytes longer than 2 mm to define spondylosis [54].
** Description of NBF: **Of the seven studies that included definitions for osteophytes, five described osteophytes as a form of outgrowth or spur arising from the bone [51, 52, 54, 55]. One study simply defined osteophyte as ‘prominent bony proliferation’ [53].
** *Osteophyte * ** Two articles defined different subtypes of osteophytes [53, 55]. Both studies outlined criteria for traction osteophytes, and one of the articles also defined claw osteophytes. These definitions were based on the shape and direction of growth of the osteophyte itself. Traction osteophytes were noted to grow horizontally in both aforementioned studies.
Eight studies were identified that outlined specific criteria for the diagnosis of spondylosis (Supplementary Table S5, available at Rheumatology Advances in Practice online). All studies defining spondylosis included the presence of osteophytes in their criteria for diagnosing spondylosis. Osteophytes were often an absolute criterion for the diagnosis of spondylosis, but not in every case. In all but one study, disc space narrowing was also included in the criteria for the definition of spondylosis, though it was not necessary to make the diagnosis in any study. Three articles included facet joint sclerosis in their descriptions of spondylosis.
Our results showed heterogeneity and variations in defining NBF in these three diseases. For SpA, our SLR revealed the inconsistencies in the literature for the definitions of syndesmophytes in terms of the shape, location and growing patterns of syndesmophyte subtypes. Also, this study identified that there are many variations of definitions of spinal DISH in the literature, which may result in different outcomes. Osteophyte formation was the only feature consistently included in the definition of spinal OA. Otherwise, joint space narrowing was frequently part of the diagnostic criteria, but the inclusion of other features was variable. These results are important to generate knowledge on how the NBFs are defined in the literature and create a standardized approach in this field.
The pathophysiological mechanism of the NBF leading to ankylosis in SpA is still unclear. The slow progression of the process requires a long-term follow-up, making it difficult to understand the natural course. In addition, the spine is not accessible for the purpose of the biopsies. Previously, it has been suggested that inflammation is the initial lesion, followed by the replacement of the subchondral bone marrow by fibrosis as a repair mechanism [56]. On the other hand, it has also been shown that syndesmophytes can grow from the areas without inflammation [36, 57]. According to this hypothesis, there might be similar underlying mechanisms in inflammatory and degenerative diseases, such as mechanical and genetic factors. For example, in DISH, where the metabolic conditions have been identified as the underlying factors, the extra spinal NBF can be seen in similar locations as SpA, such as ligaments, tendons and entheses. Therefore, it may become even more complicated to differentiate SpA from DISH [58–60]. On the other hand, DISH and SpA can also occur concomitantly, with some observation that this co-occurrence may portend worse clinical outcomes [7]. A small study comparing patients with DISH and AS demonstrated that there was a preponderance of horizontal enthesophytes in the former versus vertical enthesophytes in AS [61]. In two studies, authors attempting to differentiate spinal DISH from inflammatory arthritis used the angle of new bone growth in the spine to differentiate these two processes [7, 37]. To our knowledge, this is the only attempt in the literature on plain radiographs to differentiate the two types of NBF based on the angle, which is based on the expert opinion and has not been validated. Research on the bony changes of DISH utilizing computed tomography scans has also suggested an osteophyte angle of larger than 90° in relation to the vertebral bodies to differentiate DISH from bridging degenerative osteophytes, which has not been defined or tested in plain radiographs [62]. It will be important for any future definitions of DISH to take concurrent cases of seronegative spondyloarthropathies and DISH into consideration rather than excluding inflammatory arthritis entirely as is traditional.
From a diagnostic perspective, a clear and comprehensive definition of OA will enable clinicians to differentiate other concurrent skeletal disease processes of NBF. Overlapping features of OA and inflammatory arthropathies have also been reported [63]. For instance, enthesophytes, more classically connected to inflammatory processes, have been associated with OA in patients where SpA has been excluded [63, 64]. Erosive OA, an uncommon presentation of OA, can be especially difficult to discern from other inflammatory arthropathies, particularly PsA of the hand [65]. Especially with the ageing population, it may be complicated during the disease course to differentiate OA changes from SpA progression.
It is important to emphasize that the radiographic features of spinal NBF are not meant to be diagnostic for any disease as a stand-alone modality but rather be complimentary to the clinical features as well as other radiographic features- such as the sacroiliac joint findings. One key element to be able to differentiate the various NBF types lies under the recognition of which anatomical structures are getting Syndesmophytes are the ossification process of the annulus fibrosus, whereas para-syndesmophytes involve the soft tissues around the vertebral corners. The ‘flowing ossification’ in DISH mainly includes the ossification of the anterior longitudinal ligament, leading to a more widespread process exceeding the vertebral corners. An illustration of a variety of NBF types is provided in Fig. 2 with corresponding examples of the radiographs.
Figure 2. Illustration of some of the new bone formation (NBF) types and representative radiographs. (A) Osteophyte: Horizontal bony outgrowth, with an angle of >45° to the hypothetical line that crosses the vertebral corner/ (B) Marginal vertebral ossifications/calcification/bony outgrowth arising from the edge of the vertebral body vertically, having an angle of ≤45° to the hypothetical line that crosses the vertebral corner. (C) Non-marginal (Para-marginal) asymmetrical, thick and bulky ossifications/calcification/bony outgrowth arising from away from the edge of the vertebral body. (D) DISH: Flowing ossifications and/or calcifications of the anterior longitudinal ligament.
The major limitation of this study is having only focused on plain radiographic definitions, excluding definitions identified by computed tomography and magnetic resonance imaging. While more advanced imaging techniques can certainly be important in differentiating different types of NBF, plain films balance both cost-effectiveness and radiation exposure for patients followed over time.
We propose that the description of NBF in DISH, SpA and OA needs to include a detailed description of the anatomical location, highlight the differences in different levels of the spine and include how to differentiate DISH from syndesmophytes seen in the context of PsA and other axial SpA well as the ‘tractions spurs’ or osteophytes in OA. The detailed findings from the literature will allow us to propose the definitions and conduct a Delphi exercise with a group of international experts to be able to create formal NBF definitions for disease groups.
The improved ability to differentiate these conditions radiographically will not only allow the clinicians to accurately approach their patients but also will help the researchers to better classify patient phenotypes and focus on accurate radiographic outcomes.
Dr. Ozun Bayindir Tsechelidis and Dr. Ricardo Sabido-Sauri contributed to this manuscript for review and data extraction at the stage of revision.
Ummugulsum Gazel, Division of Rheumatology, Department of Medicine, The Ottawa Hospital, Ottawa, ON, Canada.
Gizem Ayan, Division of Rheumatology, Department of Medicine, The Ottawa Hospital, Ottawa, ON, Canada.
Nicole Hryciw, Department of Medicine, The Ottawa Hospital, University of Ottawa, Ottawa, ON, Canada.
Jean-Philippe Delorme, Department of Radiology, The Ottawa Hospital, University of Ottawa, Ottawa, ON, Canada.
Elliot Hepworth, Division of Rheumatology, Department of Medicine, The Ottawa Hospital, Ottawa, ON, Canada.
Marcos Sampaio, Department of Radiology, The Ottawa Hospital, University of Ottawa, Ottawa, ON, Canada; Division of Radiology, Department of Medical Imaging, The Ottawa Hospital, Ottawa, ON, Canada.
Zaid Jibri, Department of Radiology, The Ottawa Hospital, University of Ottawa, Ottawa, ON, Canada; Division of Radiology, Department of Medical Imaging, The Ottawa Hospital, Ottawa, ON, Canada.
Jacob Karsh, Division of Rheumatology, Department of Medicine, The Ottawa Hospital, Ottawa, ON, Canada; Ottawa Hospital Research Institute, Ottawa, ON, Canada.
Sibel Zehra Aydin, Division of Rheumatology, Department of Medicine, The Ottawa Hospital, Ottawa, ON, Canada; Ottawa Hospital Research Institute, Ottawa, ON, Canada.
Supplementary material is available at Rheumatology Advances in Practice online.
The data underlying this article will be shared on reasonable request to the corresponding author.
This research did not receive any specific grant from funding agencies in the public, commercial or not-for-profit sectors.
*Disclosure * S.Z.A. received honoraria from Abbvie, Celgene, UCB, Novartis, Jannsen, Pfizer and Sanofi. The other authors have disclosed no conflicts of interest.
The data underlying this article will be shared on reasonable request to the corresponding author.