Article Text


Evidence for a distinctive pattern of bone formation in enthesophytes
  1. M Benjamin1,
  2. H Toumi1,
  3. D Suzuki1,
  4. K Hayashi1,
  5. D McGonagle2,3
  1. 1
    School of Biosciences, Cardiff University, Cardiff, UK
  2. 2
    National University of Ireland Galway, Ireland
  3. 3
    Academic Unit of Musculoskeletal Disease, University of Leeds, Chapel Allerton Hospital, Leeds, UK
  1. M Benjamin, School of Biosciences, Cardiff University, Museum Avenue, Cardiff CF10 3AT, UK; Benjamin{at}


Objectives: The mechanism of new bone formation at entheses in spondyloarthritis (SpA) is poorly understood, but it is a key factor contributing to disability in disease. As bony spur development is also an age-related phenomenon, spurs in elderly dissecting room cadavers were studied in order to establish general principles relating to their development.

Methods: Spurs of different sizes were studied by routine histology at 26 different entheses (a total of 76 specimens) from the upper limb, lower limb and spine. The percentage of bone:marrow was compared in the posterior part of the calcaneus in cadavers with and without Achilles spurs to ascertain the relationship between spurs and immediately adjacent trabecular bone.

Results: Bony spur formation was a common age related phenomena and typically occurred in the most fibrous part of an enthesis. Paradoxically, it was often heralded by the initial appearance of a thick zone of calcified fibrocartilage that subsequently developed bony nodules within it. Uncalcified fibrocartilage was more prominent around large spurs. Endochondral, intramembranous and chondroidal ossification all contributed to spur formation and growth, but cell hypertrophy and florid vascular invasion of a cellular calcified cartilage, typical of endochondral ossification, were not conspicuous features.

Conclusion: Entheseal new bone formation occurs by a combination of three methods of ossification. However, endochondral ossification was atypical and differed from that seen in the normal development of cartilage bones or during fracture healing. How the inflammatory process modulates these processes could lead to a better understanding of entheseal new bone formation in SpA.

Statistics from

Enthesopathy is a key factor involved in musculoskeletal disability and can manifest as a wide array of degenerative or inflammatory conditions. It is particularly significant in spondyloarthritis (SpA), where new bone often develops at the attachment sites of tendons and ligaments and may eventually culminate in fusion of adjacent spinal segments. Indeed, this enthesis-related bone formation is a major cause of disability in ankylosing spondylitis (AS). The mechanism of new bone formation in SpA has assumed an even greater importance with the advent of tumour necrosis factor (TNF)α blockers, since the link between inflammation and spinal fusion is poorly defined.1

As spur development is also common at entheses as an age related phenomenon, studies on attachment sites taken from elderly dissecting room cadavers could help us to understand fundamental principles of enthesophyte formation. This is even more pertinent since young normal patients also have evidence for small spurs on high resolution ultrasound.2 Surprisingly however, even the basic structure of spurs in man has largely been ignored at most attachment sites, let alone their method(s) of ossification. In animal studies, it has been suggested that enthesophytes can develop following vascular invasion along the longitudinal rows of fibrocartilage cells that characterise the enthesis, as an extension of normal development.3 Whether this is also true of entheses in man is not known, for it was not seen in the limited number of Achilles tendon specimens examined by Benjamin et al3 in the same study. Thus, the purpose of the present paper is to establish fundamental principles of spur development by comparing spurs of different sizes from the limbs and spine of dissecting room cadavers.


The study was based on the examination of a large collection of histologically-sectioned entheses used in previous investigations with a different focus.47 From an initial analysis of the large number of available histology slides (representing over 50 different attachment sites), 26 different entheses with spurs were selected for further study (table 1), representing a total of 32 cadavers of both sexes. The entheses included ones from the upper limb, lower limb and spine. The cadavers had been donated to Cardiff University for anatomical investigation under the provision of the 1984 Anatomy Act and the 1961 Human Tissue Act. The initial embalming fluid (employed because the cadavers were also used for student dissection) contained 4% formaldehyde and 25% alcohol, but the material subsequently removed for histology was further fixed in 10% neutral buffered formal saline. Further details of histological techniques are given elsewhere.5 6 Briefly, paraffin wax embedded entheses were serially sectioned and sections mounted on glass slides at 1 mm intervals throughout the thickness of each specimen. The slides were stained with Masson trichrome, toluidine blue and Hall and Brunt quadruple stain. As in previous studies, the medical history of the cadavers was not known.

Table 1 Histologically-sectioned entheses in which spurs were present.

An assessment of the known types of new bone formation was made including endochondral, intramembranous and chondroidal ossification. Endochondral ossification was recognised by the presence of calcified cartilage showing evidence of matrix destruction as it is replaced by bone.8 Intramembranous ossification was suggested by the deposition of new bone matrix (osteoid) on pre-existing bony surfaces and chondroidal ossification by the direct conversion of cartilage to bone without cartilage erosion. Hence, the key evidence for chondroidal ossification was the presence of chondroid bone (ie, bone characterised by cartilage-like cells) and the lack of any distinct boundary between cartilage and bone.9

Quantitative analysis of the calcaneus in cadavers with and without Achilles tendon spurs

In order to explore the link between enthesophytes and adjacent bone, serially sectioned entheses from 10 cadavers were used to analyse the structure of the posterior part of the calcaneus. Because of the large size of the Achilles tendon, its enthesis was collected as three equal sized pieces from the medial, central and lateral thirds of the attachment site. In six of the specimens, a spur was identified (in at least one part of the enthesis), but in the remainder, there were no spurs at all. For each cadaver, one section at each 1 mm sample point, spanning from medial to lateral was digitised using a ×10 objective lens and all measurements were performed on the digital images. The percentage of bone:marrow was calculated using algorithms developed in our laboratory with Matlab software (V., release 14; Mathworks, Natick, Massachusetts, USA) as described previously.10 Briefly, a segmentation process permitted separation of cancellous bone spicules from marrow. The segmentation was made with an edge detection method, using a LaPlacian-Gaussian filter. The percentage of bone:marrow was estimated at three sites in the posterior part of the calcaneus: the superior tuberosity, and the proximal and distal halves of the enthesis. For each cadaver, a single mean value was obtained by averaging the measurements taken at every individual sample point. The significance of the difference at each of the three sites between the percentage of bone:marrow in cadavers with and without spurs was assessed by Student t test.


A total of 76 spurs were analysed at 26 different entheses (table 1; 32 cadavers, both sexes, mean age 80 years). Among the most common were those at the insertions of the quadriceps, Achilles, biceps brachii, fibularis (peroneus) brevis and supraspinatus tendons; at least five specimens were available for study in each case. The largest spurs were always those in the largest tendons, with some spurs in the Achilles and quadriceps tendons reaching 1 cm or more in length or 4 mm or more in thickness (fig 1A,B). Although some large spurs had a recognisable enthesis with the overlying tendon (fig 1C), there were many other spurs (large and small) that were not conspicuously anchored to adjacent collagen fibres (fig 1A,D). Some of these spurs had obvious gaps in their cortical bony shell, so that marrow was in direct contact with soft tissue at local sites (fig 1D). Although large spurs were more striking, small ones were more common and particularly characteristic of the insertions of the biceps brachii, fibularis brevis and gluteus medius tendons (fig 2A).

Figure 1

(A) A sagittal section of the quadriceps tendon enthesis in which there is a large, blunt bony spur (S) at the most superficial part of the attachment site. In this region, the parent bone is characterised by densely-packed, highly anisotropic trabeculae (T), but the bony architecture of the spur is clearly different. Note the invasion of blood vessels (BV) into loose connective tissue (arrow) that caps the tip of the spur and which reduces the extent of spur–tendon attachment. This is probably associated with the less regular arrangement of its trabeculae. Note also the local defect (D) in the bone cortex at the tip of the spur. Masson trichrome. Scale bar: 3 mm. (B) A macroscopic view of a long, thin bony spur (S) in the most superficial (most fibrous) part of the quadriceps tendon enthesis. Note again the anisotropic trabeculae (T) in the parent bone. Scale bar: 4 mm. (C) A large bony spur (S) that has a distinctive enthesis with the overlying quadriceps tendon. Thus, note the thick zone of calcified fibrocartilage (CF) knitting deeply into the underlying bone (B) and the prominent tidemark (TM). Scale bar: 400 μm. (D) A large bony spur (S) in the quadriceps tendon with an “open tip” (ie, a large part of the spur surface where bone is absent (arrows) and where the tendon (TEN) does not attach to it). Fat (F) overlies the bone defect instead. Scale bar: 400 μm.

Figure 2

(A) A small bony spur (S) at the attachment of the tendon (T) of gluteus medius. Note the greater prominence of uncalcified (UF) rather than calcified (CF) fibrocartilage around this particular spur. Scale bar: 400 μm. (B) A macroscopic view of several bony spurs that have developed within the Achilles tendon. Unusually, spur S1 lies in the middle of the tendon, but spurs S2 and S3 are at the periphery of the attachment site. Between the two sets of spurs is an abundance of uncalcified fibrocartilage (UF) that is clearly highlighted by its intense staining with toluidine blue. Note the presence of streaks of loose connective tissue in the tendon near the spurs (arrows). Scale bar: 2 mm. (C). A thickened zone of calcified fibrocartilage (CF) that has developed in the superficial (ie, most distal) part of the Achilles tendon enthesis. This is a region where such tissue is normally inconspicuous, but also where bony spurs most typically develop. B Bone, T Tidemark. Masson trichrome. Scale bar: 300 μm. (D) In this example (taken from the tibial attachment of the anterior cruciate ligament) of what is interpreted as an early stage of spur development, there are numerous small bony nodules (BN) within a greatly thickened zone of calcified fibrocartilage (CF). B Bone, T Tidemark. Masson trichrome. Scale bar: 200 μm.

General features of new bone formation at entheses

A widespread feature of spur formation was their development at the periphery of attachment sites in what was often the most distal (or superficial) and fibrous part of a fibrocartilaginous enthesis (fig 1A,B). The “parent” bone from which spurs protruded at the periphery of most large entheses, had anisotropic trabeculae, orientated parallel to the long axis of the spur (fig 1A,B). Paradoxically, the development of spurs at fibrous sites was often accompanied by fibrocartilage formation as described further below (fig 2A,B).

Histology of spurs

There was commonly a marked increase in the quantity of calcified enthesis fibrocartilage and the appearance of multiple, bony nodules within this tissue (fig 2C,D) at sites where spurs were known to develop (though had yet to appear) and in the immediate vicinity of small spurs. Occasionally however, small spurs were surrounded by uncalcified rather than calcified fibrocartilage. The bony nodules could be small, isolated islands of bone within the calcified fibrocartilage or microscopic protrusions from the subchondral plate. Although some uncalcified fibrocartilage was indeed present at the time the calcified fibrocartilage thickened or small spurs formed, the non-mineralised tissue became increasingly evident as the spurs grew, capping their tips and/or sides (fig 2B).

Ossification mechanisms

Bony spurs developed by endochondral, intramembranous and chondroidal ossification. Evidence of endochondral ossification was provided by the ample signs of remnants of calcified fibrocartilage within bony spurs of all sizes (fig 3A). Indeed, such is the prominence of calcified fibrocartilage in some spurs that they cannot always be regarded purely as bony structures, but as bone-cartilage complexes (eg, that illustrated in fig 3A). Curiously (and unlike classical endochondral ossification) it was difficult to find examples of a highly cellular, calcified cartilage that was richly endowed with hypertrophic cells. However, there was abundant evidence of cartilage destruction, vascular invasion and osteoclastic activity (fig 3B,C).

Figure 3

(A) A macroscopic view of a spur (S) at the peripheral part of the quadriceps tendon attachment site. There are prominent remnants of calcified fibrocartilage (CF) within the spur, indicating that its development involved endochondral ossification and that not all the calcified fibrocartilage had been remodelled. Note the area of loose connective tissue (L) overlying the spur and the associated blood vessels (BV). Toluidine blue stain. Scale bar: 1 mm. (B) A low power view of a region of calcified fibrocartilage in a large Achilles spur showing evidence of vascular invasion and osteoclast erosion (OC). The cells in the calcified fibrocartilage of the spur are neither hypertrophied nor closely packed. BV Blood vessels. Inset: enlarged view of a multinucleate osteoclast. Both sections were stained with Masson trichrome. Scale bars: 100 μm (main fig) and 20 μm (inset). (C) Close-up of blood vessels (BV) in a region of vascular invasion within an Achilles spur (the location of the vascular region is shown by the rectangle in the inset). Scale bar: 100 μm. (D) Intramembranous ossification in a large Achilles spur with evidence of osteoid (O) formation. The photograph is taken at the tip of the spur in the region indicated by the rectangle in the small inset. The osteoblasts (OB) that have secreted the osteoid have become artefactually detached from the surface of the newly-deposited bone matrix. LB, Lamellar bone. Masson trichrome. Scale bar: 100 μm.

There were also signs of osteoid deposition by osteoblasts at numerous locations within the spurs (fig 3D). Where the osteoid was laid down on pre-exisiting bone, this was interpreted as evidence of intramembranous ossification (figs 3D, 4A). The frequent contribution of chondroidal ossification to spur enlargement was suggested by the presence of chondroid bone (fig 4B). Where such bone was identified, it was generally difficult to determine the boundary between hard and soft tissues, for there was no well defined tidemark. Both of these processes were evident in the distal part of spur that is predominantly subject to tensile loading.

Figure 4

(A) Large multinucleated cells (putative chondroclasts shown by arrows) at an erosion front within the soft tissue capping the tip of a spur in the quadriceps tendon. The presence of these cells, together with a layer of osteoid (O) on the bone surface suggests that this spur was actively growing by intramembranous ossification. Scale bar: 100 μm. (B) Evidence of chondroidal ossification in a growing spur at the tibial attachment of the anterior cruciate ligament. Note the presence of chondroid bone (CB) and the difficulty in distinguishing any sharp border between the highly cellular cartilage (C) and the chondroid bone. BV Blood vessel, LB Lamellar bone. Scale bar: 50 μm. (C) It is common to find spurs (S), as for example here in the Achilles tendon, in which the bone surface is punctured by small holes that allow blood vessels (BV) in the soft tissue and marrow to be continuous with each other. Scale bar: 300 μm. (D) Mucoid degeneration (MD) at the tip of a spur that has developed within the Achilles tendon. Scale bar: 300 μm. All sections are stained with Masson trichrome.

In addition to osteoclasts that eroded or remodelled spurs from the “inside out”, there were other multinucleated cells of similar appearance on the soft tissue side of the spurs (fig 4A). They were thus regarded as putative chondroclasts. Occasionally, thin-walled and dilated blood vessels were present in the soft tissue overlying the spurs (fig 4C). Such vessels were usually surrounded by loose connective tissue and may be associated with small nerves.

There were usually signs of microdamage on the soft tissue side of entheses near spurs. These included matrix degeneration, fibrocartilage cell clustering and fissure formation (fig 4D). Again, this differs from the features that are typical of normal bone development, as was the common presence of gaps in the cortical bone at the periphery of spurs that allowed thin-walled blood vessels (conspicuous in the overlying soft tissues) to be continuous with vessels in the marrow spaces of the spurs themselves (figs 1D, 4C). The marrow itself was similar to that in the parent bone in all spurs.

Quantitative analysis of the calcaneus in cadavers with and without Achilles tendon spurs

The relative quantity of cancellous bone compared with marrow in the posterior part of the calcaneus was significantly greater (p<0.01) in cadavers with than without spurs (table 2). This was the case not only in the distal part of the entheses (ie, where the spurs formed), but also in the proximal part of the enthesis and in the adjacent superior tuberosity.

Table 2 The percentage of bone:marrow in the posterior part of the calcaneus in cadavers with and without bony spurs. Mean values (standard error of the mean (SEM)). In all three regions, the values are significantly different (p<0.01).


The study provides a novel analysis of a large number of spurs from a wide variety of attachment sites, in order to improve our understanding of the mechanism of new bone formation at entheses. The results suggest that a common trigger for spur development is the appearance of a thickened region of calcified fibrocartilage—paradoxically, in a part of an enthesis that is normally relatively fibrous rather than fibrocartilaginous. Consequently, spurs typically appear in the most superficial part of attachment sites. It is likely that bony nodules develop once the calcified fibrocartilage has thickened pathologically and that these nodules then fuse with each other to form a macroscopically-recognisable spur. At the cellular level, it therefore appears that entheseal new bone formation differs from that which occurs during normal development or fracture healing (fig 5). Spurs evidently grow by a unique combination of endochondral, intramembranous and chondroidal ossification, yet with no suggestion in man that vascular invasion can proceed along rows of enthesis fibrocartilage cells as reported previously in the rat.3

Figure 5

A diagrammatic comparison of ossification methods as they typically occur in a developing long bone and as they are hypothesised to occur in an enthesophyte. (A) A developing long bone in the region of a growth plate. The calcified cartilage (CC) that is eroded during endochondral ossification contains a highly cellular array of greatly hypertrophied chondrocytes (HC), many of which are arranged in longitudinal rows. There is considerable vascular invasion of the spaces left by the degenerating calcified cartilage and ample evidence of osteoclastic (Oc) activity and matrix destruction (arrows). Osteoid (O) is deposited on fragments of calcified cartilage by osteoblasts (Ob) to form “mixed spicules” (MS) and at the edge of the long bone, the perichondrium (P) differentiates into a periosteum. This produces a periosteal bone collar (PBC) by intramembranous ossification. BV, blood vessels. (B) Part of the tip of an enthesophyte. The spur consists of a mixture of lamellar bone (LB) and calcified fibrocartilage (CFC), but small mixed spicules of cancellous bone similar to those at a growth plate, are not obvious. Some tendon collagen fibres (T) are anchored to the spur, but parts of the spur are devoid of such attachment. There is ample evidence of cartilage matrix destruction, together with osteoblastic (Ob) and osteoclastic (Oc) activity. Note however that the fragments of calcified fibrocartilage (CFC) within the spur are not densely cellular and the cells (C) are rarely aligned in columns or hypertrophied. Vascular invasion is easy to demonstrate, but the florid angiogenesis typical of endochondral ossification is less evident. Osteoblasts contribute to forming bone by intramembranous ossification, but it is less easy to find bone deposited on the surface of cartilage remnants. In contrast to growth plate development, chondroidal ossification leading to the formation of chondroid bone (CB), may be a feature of spur development and/or growth. In this process, fibrocartilage cells (FC) are directly incorporated into the bone without cartilage erosion. Note the presence of chondroclasts (Cc) in the soft tissue overlying the spur, together with signs of matrix destruction (MD). O, osteoid.

The development of spurs in the more fibrous regions of entheses is in line with the common view that many enthesophytes are traction spurs that appear in response to abnormal patterns of tensile loading. Indeed, the antecedent thickening of calcified fibrocartilage may in itself be a response to recurrent trauma. Thus in the Achilles tendon, spurs typically develop in the distal part of the enthesis, which is subject to greater impact on heel strike than is the proximal part. Furthermore, if one regards superficial olecranon traumatic/idiopathic bursitis as an indicator of trauma that stems from recurrent pressure, it is informative to note that spurs are more common in the elbow with bursitis rather in the contralateral limb.11 However, it does not necessarily follow that all spurs seen in elderly dissecting room cadavers are traction spurs. Kumai and Benjamin12 argue that as spurs are deep to the plantar fascia rather than within it, they are unlikely to develop in response to excessive traction forces.

The link between spur development and mechanical loading is further supported by the current finding that spurs in the Achilles tendon are characteristic of cadavers where there is a particularly densely packed network of spicules in the parent cancellous bone. It seems that spur formation is in some way linked to a more generalised bone-forming or bone-preserving capacity with age—in line with the suggestion that some individuals are “bone-formers”.13 As Wolff’s law dictates that the architecture of spongy bone reflects changes in loading patterns, the possibility needs to be considered that some enthesophytes at least, represent adaptive changes in the “parent bone” that culminate in an “overflow” of bone into soft tissue. This seems to occur in regions where spicule thickness and/or number are already high.

The conspicuous absence of compact bone at fibrocartilaginous entheses suggests that there must be some mechanism preventing cancellous bone being infilled to such an extent that it is converted into compact bone. Curiously, it is not necessarily the case that spurs are firmly anchored to the tendon/ligament at their tips. Furthermore, as some small bony nodules are initially islands surrounded by calcified fibrocartilage, it cannot be assumed that all spurs are connected to their “parent bone” from the moment they appear (ie, some may initially be “microscopic sesamoids” within the tendon or ligament). It further follows that the multiple bony projections seen clinically in radiographs at the tips of spurs, may indicate the prior existence of separate bony nodules. The presence of loose, vascular connective tissue near many spurs and the dramatic difference in the bony architecture between the spur and the adjacent region of its parent bone (fig 1A) challenges the idea that once formed, spurs necessarily continue to be subject to high tensile load.

We have previously emphasised how non-immunological factors or tissue specific factors could determine the phenotypic expression of disease, in this case, at the enthesis.5 14 In the present study, we have noted that the fibrous part of an enthesis has abundant evidence for osteoblastic and osteoclastic bone responses, in addition to tissue microdamage and altered vascularity. Although spur development in elderly cadavers is unlikely to hinge around inflammatory changes, one can speculate that in the inflammatory milieu associated with SpA, physiological tissue remodelling responses are altered and that by poorly defined mechanisms new bone formation occurs at sites of predominant tensile loading of entheses. At the cellular level, this process appears to be quite distinctive and needs further study. The relationship between bone formation responses at the distal insertion and diffuse adjacent osteitis that is typically evident on MRI in SpA is by no means clear.

There are a few features of spur development emerging from our study that are strikingly different from classic endochondral ossification as it occurs in normal long bone development8 and these are summarised schematically in fig 5. There is less obvious cell hypertrophy, the cells in the calcified fibrocartilage of the developing spur are far less densely packed, and the vascular invasion of that fibrocartilage (though present) is less striking. All of these features are likely to be interconnected and to reflect a difference in the rate of bone formation, with spurs probably formed slowly. This may also explain why vascular invasion along rows of fibrocartilage cells aligned with the long axis of the spur is not a feature of human spurs as described in the Achilles tendon of the rat.3 The short life span of the rat means a faster growth rate.

Terminal differentiation of chondrocytes without cell enlargement has been suggested previously,15 though not in relation to bony spurs. Furthermore, even where chondrocytic hypertrophy does occur during ossification, cell enlargement is not necessarily followed by cell death.16 Evidence of erosion itself is clear enough in bony spurs, for there were numerous examples of enthesophytes in which matrix destruction was ongoing via osteoclasts and/or chondroclasts. However, the actual process of endochondral ossification must differ from that typical of cartilage bones during development. Here, a striking hypertrophy of closely-packed chondrocytes precedes matrix calcification and is followed by cell death and florid vascular invasion. During spur enlargement, the vascular response was often less striking. It therefore appears that for a variety of reasons, new bone formation at entheses differs from that which occurs during normal development or fracture healing. Spurs evidently grow by a unique combination of endochondral, intramembranous and chondroidal ossification—yet with no suggestion in man that vascular invasion can proceed along rows of enthesis fibrocartilage cells, as reported previously in the rat.3

As far as we are aware, this is the first account of the structure of enthesophytes of different sizes, from a wide variety of tendons or ligaments within a single manuscript. We have tried to piece together from the histological sections, a plausible account of the formation, growth and remodelling of bony spurs. However, it is important to recognise that our study was based on elderly dissecting room cadavers and not on patients with SpA. Thus, most of the spurs we observed are likely to be degenerative rather than inflammatory in nature. The reader should also be aware that the method of spur formation could differ with age. Thus, our common findings of degenerative changes at the tips of spurs may reflect an aging change rather than one intrinsic to spur formation. Finally, as with any histological approach to studying a developmental process, a definitive account of the stages of spur development is not possible. Any histological image always captures a single moment in time and there is inevitably an element of speculation as to what the tissue might have looked like before or after the time at which the sample was taken.

To summarise, an appreciation of the present findings may be important for studies that explore the basis for new bone formation in SpA where it is unclear how current treatment protocols may influence ossification. It is possible that the inflammatory response at the enthesis associated with SpA could significantly modulate the vascular pattern and in combination with tensile loading could in some way enhance new bone formation over time.


We thank S Redman and D Scarborough for their technical help with the histology.


View Abstract


  • Competing interests: None declared.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.