The comparison with the regular histology
The regular intercentrum is built of two domains, endochondral and periosteal. The trabeculae in the endochondral part are not organized, in contrast to periosteal bone in which the trabeculae are regularly arranged. Endochondral ossification developed first, followed by periosteal ossification, initially fast and subsequently slowing down during the ontogeny [17, 18]. The border between these two domains is usually well defined in small individuals thanks to a different orientation of trabeculae, but in older it is less visible due to remodelling, resulting in a smooth transitional zone built of secondary trabeculae. The older the bone ontogenetically, the thicker the secondary layer. Numerous calcified cartilage residua are always preserved in the primary endochondral domain, but this tissue is not related to the ontogenetic age of Metoposaurus [17]. The preservation of calcified cartilage extending long into the ontogeny seems to be plesiomorphic for all Stereospondyli [18].
In a transversally sectioned normal intercentrum, primary trabecular bone of the endochondral domain is almost entirely surrounded (except for the dorsal surface) by highly vascularized primary cortex, representing the periosteal domain [17, 18]. Periosteal bone visible in the normal intercentrum represents the lamellar-zonal bone tissue type with parallel-fibred matrix and numerous vascular canals organized in regular rows in zones, whereas annuli are avascular. Deeper parts of the periosteal bone are usually strongly remodelled and primary bone is visible only in a form of small patches between the cavities. A few deep canals, perforating the surface and reaching to the endosteal domain, are visible, representing probably nutrient canals [17] (Additional file 6).
In the pathological bone analysed herein, the entire inner part of the star-shaped structure in the pathological intercentrum, with the exception of the tips of its arms, is filled with secondary trabecular bone and it is not possible to determine the border between the primarily periosteal and endochondral domains. Surprisingly, the analysed specimen does not show any calcified cartilage in endosteal domain in both sections. The lack of calcified cartilage in the sub-horizontal section of the intercentrum of the analysed specimen is due to the sectioning plane, limited to the periosteal part of the affected bone. However, the lack of any remains in the transverse section is unusual. The pathological processes causing an uncontrolled deposition of the bone in the periosteal domain may have also accelerated the physiologically slower ossification rate of the endochondral domain. Thus, even if on the microstructural level the endochondral domain does not differ very much from the regular bone, the character of the cancer affected its histological picture. Remains of the non-affected cortex are limited only to a thin, avascular layer of parallel-fibred matrix, separating the neoplastic bone from the regular secondary trabecular tissue. Small fragments of residual periosteal bone are also visible in the tips of the remaining star. These residues of tissue indicate the relatively long growth of the periosteal cortex, despite the developed pathological tissue, and represent the ontogenetically youngest part of the regular structures.
Differential diagnosis
The massive, irregular outgrowths on the metoposaur intercentrum spread in multiple directions. The absence of draining sinus tracts, defects surrounded by new bone formation (abscess) or filigree reaction rules out an infection-related bone involvement [21, 22]. No bone deformations indicative of healed fractures were present. Absence of subperiosteal bone resorption or large resorption cavities rules out hyperparathyroidism, which would also not be associated with massive new bone deposition [22, 23]. The documentation collected from gross anatomy, histology, and X-ray computed micro-tomography suggests a fast-growing bone mass (e.g., huge lacunae and higher vascularization) and abnormal new bone deposition as a result of increased cell proliferation, characteristic for neoplasia [21, 22].
The abnormal bone deposition in this case indicates that the individual was clearly afflicted with a primary neoplasm of bone. Benign bone tumours (e.g., osteoid osteoma, osteoblastoma) form spherical protuberances composed of very dense bone [21, 22] which are absent in the studied specimen. The aggressive bone destruction and massive new bone formation indicates a sarcoma of the osteosarcoma or chondrosarcoma variety. Absence of internal curly-cue or popcorn-like calcifications rules out a cartilage-derived tumour (chondrosarcoma) [22].
Osteosarcoma is the of the most common primary neoplasms affecting musculoskeletal system [24], characterized by proliferation of malignant cells of mesenchymal origin. Spinal osteosarcomas are rare and aggressive neoplasms [25]. There are different variations of osteosarcoma: central periosteal, parosteal, or telangiectatic. The latter is refuted because of absence of large osteolucent areas related to telangiectatic blood vessels [22, 26]. The location of the bone mass surrounding the affected vertebra and the clear boundary between normal and altered bone suggest that the neoplasm originated from the periosteum, rather than being centrally-derived. Further, the osteosarcoma in this individual developed from the surface of the bone (rather than within the periosteum that occurs with periosteal osteosarcoma [22]) and penetrated into the vertebral intercentrum though natural canals [17] forming the star-shape of the remaining intercentrum (Fig. 4), thus identifying it as a parosteal osteosarcoma [7, 24].
Osteosarcoma in paleontological record
The evolutionally history of osteosarcoma is poorly understood [7]. The fossil record provides several cases of osteosarcomas in amniotes—an early Pleistocene hominid [27], a Late Jurassic dinosaur [28], and a Middle Triassic stem-turtle [2]. The identification of osteosarcoma in a Late Cretaceous ceratopsid [7] is questionable since the mentioned in its description small unconnected islands with circular morphology are not typical of osteosarcoma and more suggestive of a cartilaginous component of sarcoma, a chondrosarcoma [21, 22]. The oldest report suggesting osteosarcomatous involvement is an abnormality of integumentary cranial bones in an Early Triassic capitosaurid [13] from Russia. However, the data presented suggest extra bone pieces, typically found in Wormian bone overgrowth [29], without identifying the destructive changes characteristic of osteosarcoma [21, 22]. Since the case studied here, there is no known obvious record of osteosarcoma in extinct amphibian.
Tumour growth dynamics
Three-dimensional analysis and distribution of the pathological tissues allow reconstruction of the dynamic of cancer growth in during its ontogeny. The morphology and the XMT visualisation reveal that the maximal development of the neoplasm is asymmetric, so the maximal deformations are present below the left parapophysis and around the right one, probably including even the region of the neural arch connection. Posterior surface of the intercentrum is the least affected by the overgrowth, whereas the anterior surface is strongly deformed. This suggests the entrance of the neoplasm from the anterior side. The latter shows extensive alterations in the deep parts of neoplasm, confirming the ontogenetically older stage of the tissue [30]. The malignant process resulted in loss of star shape visibility in the transverse section next to the anterior surface. The size of numerous areas of trabecular destruction-derived cavities in the pathological tissue exceed these known from normal tissue [17] and are typical for osteosarcoma-related bone lesions.
The most interesting structure visible in both digital and thin sections is the star-shaped structure. Histological thin sections show clearly that the characteristic for tumour, fast growing tissue is separated from the inner trabeculae by remains of the cortex. Crucial for the spread of the pathological tissue seem to be five nutrient foramina penetrating the periosteal domain of the affected intercentrum (compare Fig. 4). The comparison of the arms and the position of the canals in non-pathological bone reveals that the arms fit well between the canals, whereas the concavities are located in the regions where the canals occurred. Moreover, the almost continuous structure of the periosteal remains suggests that the neoplasm minimally permeated the endochondral domain, mostly deforming the periosteal bone. With time, the pathological tissue started to develop next to the anterior surface, overgrowing ventrally and laterally. The nutrient canals constitute a natural way for invasion of the neoplasm into the cortex. Pathological tissue was deposited rapidly as indicated by the structure of the tissue. Normal deposition was still possible in the not yet affected fragments of the periosteal bone, located between the canals, and the periosteal cortex increased its thickness. During ontogeny, both processes occurred in parallel, but the neoplasm grew faster and limited the space where periosteal growth was still possible to constantly narrowing wedges. The huge amount of osteocyte lacunae and the radial orientation of vascular canals confirm the much faster growth of the pathological tissue than the normal tissue, where only moderate number of osteocyte lacunae and considerable lower porosity are observed [17]. In the final stage, the neoplasm overgrowth surrounded almost the entire intercentrum and the physiological periosteal growth was no longer possible.
Neoplastic bone deposited close to the original vertebral centrum is well ordered. The orderliness decreases, and sinuses and cavities appear as one proceeds further from the intercentrum. The presence of other structures such as ribs of soft tissues may have limited the growth of the bone tissue. Small structures visible in the ground section in the left lateral overgrowth, clearly separated from the pathological tissue by sediment (Fig. 3B), seem to be remains of anatomically nearby-located bones such as ribs.
The notch visible in the dorsal part of the section is intriguing. Normal Metoposaurus spp. intercentra from the post-cervical or anterodorsal regions are not permanently fused with the neural arch [17, 31], and even in the histological sections no sign of articulation or co-ossification of both elements is visible [18, 20]. Here, the clear sutures show that in the dorsal bone mass, other elements have been included and pathologically co-ossified. The location of the notch and its shape suggest that in that case two rami of the base of the neural arch were permanently fused with the intercentrum, similar as it is in the atlas. In the case described here of the pathological condition, the notch may represent the neural canal [17, 31].
Physiological aspects of bone tumour formation
In mammals, bone neoplasms are formed as woven bone [32], which is a fast growing matrix type resulting from the static osteogenesis [33]. Characteristic for that tissue is an unorganised structure of collagen fibres and unordered osteocytes lacunae. The studied neoplastic tissue is highly organized with numerous osteocyte lacunae. The disordered arrangement of osteocytes observed in the neoplastic bone points to static osteogenesis [33], which is connected with the formation of fast-growing woven bone. Surprisingly, in the studied metoposaur, an orderly arrangement of osteocytes lacunae is noted [33]. As a result of static osteogenesis, a bone with a high degree of packing is produced, which seems to be a structural novelty. A characteristic feature of the studied bone tumour in the metoposaur is that it is macroscopically similar to that known in amniotes, but it is differently at the level of histology. The unordered arrangement of the osteocyte lacunae can result from the very fast maturation process of osteoblasts, which mature before they are able to transit to the bone surface, as it occurs in the normal dynamic osteogenesis [33]. The dominant tissue types known from healthy bones of Metoposaurus are parallel-fibred or lamellar bone, both resulting from dynamic osteogenesis [18], typically found in bones that exhibit incipient fibro-lamellar bone [34]. However, even if the incipient woven bone was produced, the overall tissue was not as highly vascularised and rich in osteocyte lacunae, and the overall growth rate was not as high as observed in the mammalian neoplastic bone [32]. It is possible that genetic limitations of the organism made it impossible to produce fast growing true woven bone and it was metabolically more effective and only possible to accelerate the growth rate via increasing the number of osteocyte lacunae, their maturation rate, and extremely high vascularisation. However, it is not clear how the scattered bone cells were able to produce a highly organised tissue. The huge amount of osteocyte lacunae and the radial orientation of vascular canals confirm the much faster growth of the pathological tissue than the normal tissue, where only a moderate number of osteocyte lacunae and considerably lower porosity are observed [17]. Large and subspherical osteocyte lacunae (Additional file 7) were noted in pathological states in fossils [35,36,37]. However, there are no explanations of the role of these osteocytes in cancer biology known from paleohistological studies. Recent medical studies have shown that direct attack by cancer cells on osteoblasts induces the less-organized osteoblast arrangement [38] and proved that osteocytes as important components of the cancer microenvironment in the bone where cancer cells alter osteocyte viability and their gene expression profile [39, 40]. Thus, osteoblasts which in a physiological state deposit highly organized collagen fibres and established on the bone surface regular rows of osteocytes [33], mature earlier in the bone matrix before shifting to the bone surface and deposit bone in an abnormal fashion, still lamellar on the collagen fibres level but disorganized structurally, enormous porous with a high number of primary trabeculae. The last character seems to be crucial for the estimation of bone growth rate. The fact that the tumour finally overgrowth the entire healthy tissue and limited its growth indicates that pathological tissue was growing faster than the healthy one, however only in a relative way. It is not possible to state how long it takes for the neoplasm to develop to the size observed here. Despite the diagenetic alteration and destruction, the pathologic process is clearly visible in the examined specimen. Wherever the tumour had a space to grow, it grew in an orderly manner (blood vessels arranged radially); where there was an obstacle, the tumour bone lost its organization.
It is, therefore, noteworthy that while no woven bone is observed in ZPAL Ab III/2467 and the pathological region is composed of well-organized, lamellar matrix accompanied by high number of radial vascular canals, the tissue is rich in subspherical (regardless of the sectioning plane) osteocyte lacunae and relatively loosely arranged spatially trabeculae, with the axes locally oriented predominantly outwardly, in a sense mimicking rapidly deposited radial fibrolamellar bone tissue of amniotes [41, 42]. This suggests that while on one hand the growth rate was limited by the deposition rate of the tissue determined by its organized weave, on the other hand, that was compensated by its spatial layout. While radial (or spicule-like) deposition of periosteal bone is known to occur in pathological scenarios [37, 43], no comparable tissue type was thus far described in Mesozoic amphibians.
Theories on aetiology of cancer
Neoplasms are common across the animal kingdom and affect most vertebrates [8, 44] constituting a part of class heritage. Tumorigenesis seems to be plesiomorphic for animals—spontaneous tumours appear even in basal metazoans [45]. As metazoans exhibit a natural propensity to proliferate [46], the clonal cell divisions are controlled by genetic mechanisms. In the conventional, currently accepted paradigm, called Somatic Mutation Theory (SMT) [47], development of neoplasm is related with the loss of genetically-conditioned coordination of cell proliferation and differentiation [48]. It is a consequence of evolution, development of clonal multicellularity, and increase of somatic complexity. According to this paradigm, the transformation of a normal cell into a neoplastic cell occurs directly through DNA damage and any cell in the organism can become cancerous. Hypothetically, it means that the greater the somatic complexity of the organism, the more susceptible it should be to cancer. However, this is not true, since in large animals such as pachyderms [49], the prevalence of cancer is lower than in small mammals. This phenomenon is known as Peto’s paradox [50] and is one of big challenges in modern comparative oncology.
The somatic mutation paradigm is prevalent and sets the directions for current cancer research. In 1999, Sonnenschein & Soto [51] proposed an alternative theory, in which tumorigenesis takes place when the normal interaction between the functional calls (parenchyma) and structural cells (stroma) of organism is disturbed. According to that view, neoplasm should be examined from a hierarchical perspective of the organism and defined as a problem of tissue organization. This theory was named Tissue Organization Field Theory (TOFT) [47], because it recognizes that the primary source of cancer is the loss of organization at the tissue level, not genetic mutations [47]. Cells in an organism’s tissues undergo numerous processes that regulate their metabolism. A number of intercellular communication mechanisms which guarantee a correct organization and coordination of cells at every stage of ontogenesis, including histogenesis, are responsible for the regulation of tissue physiology. Disruption of communication processes, related, for example, with the state of cell membrane polarization and disruption of ion transport through ion channels and ion gaps, may consequently lead to changes in the organization and functioning of tissues [52]. Such changes may include: (1) disorders of tissue metabolism, (2) increase in cell proliferation and mobility, (3) transformation of mature cells into stem cells, which in turn may lead to the development of neoplasms. Both theories describing carcinogenesis have been widely studied and discussed. Although SMT and TOFT describe two different mechanisms of cancer formation, there are attempts to combine the two theories. However, as suggested by Montévil and Pocheville [53] because of the differences in the logical assumptions of these theories, they should not be combined.
The discussed case of cancer in a metoposaur (Fig. 5) is consistent with the Tissue Organization Field Theory, which locates the causes of neoplastic transformations in disorders of tissue architecture. This is expressed in several ways: (1) the fast growth characteristics of the newly formed bone which mixes a slowly deposited matrix type with spatial distribution typical for rapidly growing bone; (2) both the affected intercentrum and the overgrowth being subject to physiological remodelling processes, as evidenced by the numerous areas of bone tissue destruction within the tumour and the vertebra itself; it appears that the physiological processes occur in the neoplasm and the original bone alike; (3) it is difficult to explain why the remains of the cortex exhibited as the star-shaped structure are so well marked; in case of a typical neoplastic invasion, lesions and a chaotic organisation could be expected but in the described specimen the border between the physiological bone and the overgrowth is ordered and clear. In the analysed bone a change of tissue state by an increase of its cell proliferation and its subsequent hypertrophy can be observed, which led to the modification of the epigenetic state of its cells into a stem-like state.
Amphibians, thanks to their developmental and ecological plasticity and high regeneration potential, are a vertebrate group in which neoplasms are rarely described. As the amphibians go through the process of metamorphosis, during which the organism undergoes a reconstruction, it is assumed that they have mechanisms protecting them from carcinogenesis and point mutations within oncogenes. Because of that, they are an object of research in the area of both comparative and clinical oncology, as well as an aid in search of new strategies of neoplasm therapies and are utilized in research focused on understanding of potential mechanisms of anti-neoplasm defence. Ruben et al. [54] indicate that the mechanisms involved in the amphibian metamorphosis may also protect the organism from neoplasms. On the other hand, the rarity of identification and description of neoplasms in amphibians may also result from the issue of poor general recognition of pathology in that group [55]. A literature review of 50 cases of neoplasia in exotic amphibians (between 1954 and 2018) revealed that the most common neoplasms, mostly concern skin [12]. The resistance for cancer in some bigger mammals like pachyderms [56] and whales [57] is now recognized that an increase in repeated representation of a specific gene provides protection against cancer development [58] and was developed during a natural selection in the course of evolution [57, 58]. Molecular sequencing indicate the remote antiquity of the tumour suppressor family genes [59, 60]. A genome duplication and production of tetraploids happen in the course of amphibians evolution [61], so the duplications of tumour suppressor genes also can occur.