Sensitivity and specificity of magnetic resonance imaging and computed tomography for the determination of the developmental state of cranial sutures and synchondroses in the dog

Background During skull ontogenesis, growth centers in the skull base and calvarial bones allow gradual expansion of the cranial vault. Premature growth termination of cranial base synchondroses and/or calvarial sutures can result in devastating skull dysmorphologies. There is evidence to believe that a premature closure in one or more cranial growth centers contribute to the brachycephalic skull morphology in dogs. To provide a proof of concept for the non-invasive investigation of ontogenetic changes in cranial sutures and synchondroses in living dogs, we compared magnet resonance imaging (MRI) and computed tomography (CT) with histologic findings. Our aim was to determine the in vitro sensitivity and specificity for conventional clinical imaging methods in the assessment of cranial suture closure and synchondroses ossification in dogs. Results The evaluation of cranial base synchondroses in MRI had a sensitivity of up to 93.1% and a specificity of 72.7% dependent on the observer. The evaluation of cranial base synchondroses in CT had a sensitivity of 92.2% and a specificity of 86.4%. Suture assessment on MRI suture assessment had a sensitivity of 82.1% dependent on the observer and a specificity of 19.3%. CT suture assessment had a sensitivity of 85.1% and a specificity of 40.4% in dependence of the observer. Conclusion Conventional cross-sectional imaging techniques (MRI and CT) allow reliable assessment of the open or closed state of synchondroses within the cranial base. In contrast CT and MRI are not suitable for a reliable assessment of the cranial sutures in dogs.

basioccipital, basisphenoid and presphenoid bones emerge from cartilaginous precursors that ossify during fetal development. Cartilaginous segments termed synchondroses persist between the ossification centers, consisting of two mirror-image growth plates arranged in opposing directions. Analogous to endochondral growth plates in long bones, synchondroses of the skull base grow through ongoing chondrocyte proliferation and gradual osseous transformation allowing the expansion of the cranial base along its rostro-caudal axis [1,2,8].
The growth of individual bones is regulated within these centers by complex signal cascades, involving multiple receptors and transcription factors within both kinds of growth centers [9]. Any disturbance in these processes has been shown in children to lead to premature closure of both, cranial base synchondroses and/or sutures (craniosynostoses) and thereby to devastating skull dysmorphologies [10]. Regulation of growth and closure of both, sutures and synchondroses of the skull have been extensively studies in laboratory rodents [11,12], but not so in dogs. Even the knowledge of general temporary evolution of skull sutures and synchondroses in companion animals is limited. Two older investigations determined closure times for the cranium growth centers in dogs at the age of 12 months [13,14]. Two recent studies documented a higher incidence of closed facial sutures and premature closure times for the sphenooccipital synchondrosis in brachycephalic dogs compared to mesaticephalic dogs [15,16]. The authors of both studies suggest that the temporal variation of growth termination has a substantial influence for the development of a brachycephalic head morphology. In light of these new insights, the question arises as to whether these variations are part of a physiologic spectrum or, at least partially, a pathological condition. It should be considered that different forms of pathologic craniosynostoses and grades might contribute to the brachycephalic skull morphology in dogs in the same way as it does in cats [17]. Breeding of phenotypes based on pathological genetic defects clearly known to be associated with neurological or craniofacial diseases as in humans [18] would be prohibited by the German animal protection law and would also be unethical.
Cross sectional diagnostic imaging methods (MRI and CT) would allow the examination of large cohorts of dogs of different breeds in vivo. However, the value of these imaging methods for the assessment of synchondrosal and sutural status has never been determined.
The aim of the present study was, therefore, to compare imaging findings with histological preparations in order to determine the sensitivity and specificity for conventional clinical CT and MRI in the assessment of the open or closed status of cranial synchondroses and sutures in dogs.

Histology
Due to the fragility of the structures, especially in very young puppies, twenty-two sutures and five synchondroses were lost during sample preparation. 433 sutures and 112 synchondroses could be histologically evaluated.

Synchondroses morphology and status
In all immature dogs, a double-sided arrangement of chondrocytes in a hyaline homogenous cartilage matrix was observed between the bone tissue of the basicranial bones. The cartilage consisted of chondrocytes being distributed into a central resting zone, as well as bilateral proliferating, and hypertrophic zones ( Fig. 1 A, B) [19]. In the fourth zone in the peripheral portion of synchondrosis, osteoblasts and blood vessels invaded the area of cartilage. In older dogs, the resting and proliferative layer within the synchondrosis gradually decreased, leading to a relative narrowing of the growth center ( Fig. 1c and d). At the chondro-osseous junction differentiated hypertrophic chondrocytes are replaced by bony trabecula that, eventually, fill the growth plate leading to a continuous medullary cavity ( Fig. 1e and

Suture morphology and status
Suture morphology ranged from straight-edged plane sutures (lambda suture; Fig. 2 A, B) or butt-sutures (palatine fissure, Fig. 2c and d) over simple overlapping (sphenofrontal and squamosal suture, Fig. 2e and f; coronal suture, Fig. 3a and b) to serrated sutures, with the bone edges having a saw-like appearance (sagittal suture, Fig. 3c and d). The gap in between the bony edges are filled with collagen and elongated fibrocytes. Two types of open sutures were identified according to the type of connective tissue, and cellular components, which dominated the sutural space. In young dogs the sutures contained loosely arranged connective tissue showing a collagen fiber orientation preferentially parallel to the sutural alignment (Fig. 4a). These sutures also featured high amounts osteoblasts (> 50 in a mean of 3 FOV) and fibroblasts (> 20 in 3 FOV) (type A; Fig. 4a). In older dogs (> 7 months), sutures rather contained a more dense connective tissue with collagen fiber orientation being orientated oblique to perpendicular to the suture line, with a low numbers of osteoblasts (< 10 in 3 FOV) and fibroblasts (< 20 in 3 FOV) (type B, Fig. 4b  highest specificity of 100% (CI 95% of 60.7-100%) The intersphenoidal synchondrosis showed a high sensitivity of 93.7% (CI 95% of 79.2-99.2%) and specificity of 83.3% (CI 95% of 35.9-99.6%). The lowest accordance of CT and histological findings was shown with a high sensitivity of 85.2% (CI 95% 66.3-95.8%) and a high specificity of 80% (CI 95% of 44.4-97.5%) in the spheno-ethmoidal synchondrosis ( Table 2).
Sensitivity and specificity of suture assessment using MRI MRI suture assessment shows a moderate to high sensitivity of 72.2 (DF) -82.1% (MS) (CI 95% of 66.0-86.7%) but a low specificity of 14.4 (DF) -19.3% (MS) (CI 95% of 9.7-25.6%). Sensitivity and specificity for each individual suture and synchondrosis is shown in Table 1. Individual suture and synchondrosal assessment is summarized in Table 2.

Discussion
Synchondrosis and sutures are cranial growth centers that allow expansion of the brain and cranial cavity. An increasing body of evidence suggests that general head morphology in dogs may be substantially influenced by the function of these growth plates [15,16]. A premature fusion of one or more cranial sutures may result in characteristic shortening of the cranial cavity and facial bones in brachycephalic dogs [15,16,[20][21][22][23]. Ethical concerns as regards of breeding brachycephalic animals would arise if a pathological growth disorders would be the basis of their head morphology. The study of cranial sutures and synchondroses in dogs therefore merits more serious consideration. Non-invasive diagnostic imaging methods as MRI and CT are the primary imaging techniques used to evaluate the condition of the cranial  sutures and synchondroses in children [10,24]. They have also been used to investigate cranial growth centers in dogs [16]. The use of MRI and CT would permit the investigation of a large cohort of dogs, and would also allow monitoring the temporal behavior of a single suture in the same animal. However, a review of the veterinary literature revealed large gaps in knowledge about the morphology and ontogeny of cranial base synchondroses and cranial sutures and their assessment using imaging methods in dogs. This study was, therefore, designed to determine the sensitivity and specificity for both imaging methods using histological examination as a control. For the assessment of synchondrosal status, both CT and MRI showed a good sensitivity and specificity. The large size of synchondroses, as well as the consistent change of signal (or density resp.) from cartilage to bone tissue account for this result. However, not all sutures could be assessed equally well. The spheno-ethmoidal synchondrosis has a lower sensitivity compared to spheno-occipital and intersphenoidal synchondrosis in both CT and MRI. The microanatomy of this synchondrosis differs from the others and is more difficult to define within the cranial base as it is not always straight but crescent-shaped [24]. The transition to the air filled nasal cavity blurs the contrast between the synchondrosis and the medullary cavity that allows good visualization of the other synchondroses.
Assessment of suture status was more difficult in both modalities. Although CT correctly identified a cranial suture as open (high sensitivity), it was less reliable in identifying a truly closed suture (low specificity). MRI revealed a moderate sensitivity and a low specificity for the assessment of sutural state. The low specificity for both modalities can be attributed to the small size of the examined structures, which can be hardly visible even on a prepared skull, especially in small dogs. Although CT and MRI scanning had a submillimeter resolution in this study, the small size of some sutures in very young and older dogs (micrometer range) complicates clear identification. Cranial sutures can be reliably studied using micro-CT of human skull specimens [25]. However, considering the limitations of the gantry size of the micro-CT, and extreme long scanning time, micro-CT is impractical for in-vivo radiologic evaluation of cranial sutures. Flat-panel-CT, which produces optic resolution between 200 and 250 μm allows visualization of calvarial sutures in vivo [26][27][28]. It is likely that sensitivity and specificity of suture assessment in dogs would be higher using this imaging technique.
The microstructure of the suture might also have an influence on its assessment. The spheno-frontal sutures and squamosal sutures were particularly difficult to examine. These two are overlapping sutures (sutura squamosa), with extremely thin sutural gaps, which are located laterally on the skull curvature, making an assessment even more difficult. Butt sutures (sagittal fissure, palatine fissure) do not overlap and have a wider sutural gap, which makes them easier to identify. Furthermore, the localization of the sagittal, and lambdoid sutures is marked by the external sagittal crest or the nuchal crest, respectively, which simplifies their identification in both imaging and histology.
Beyond size and morphology, the variability between the observers indicates that assessment of sutures is also experience dependent. Classifying the maturation stage of cranial sutures in children requires significant training and proficiency [29]. Although diagnostic methods to evaluate cranial suture ossification/maturation in human studies usually rely on Flat-panel-CT with higher resolutions, only images read by experienced viewers achieves good to excellent agreement compared to histologic slides [30]. Interrater variability for the assessment of cranial sutures in humans are consistent with the range of this study in dogs (0.67-0.84) [31].
In humans, the end point of cranial vault growth is determined upon fusion of the sutures in the third decade of life [5], but there can be considerable variability in closure rates [32]. It is interesting to note, that even in the old dogs the majority of the cranial sutures do not ossify. The interfrontal, coronal, sagittal and lambdoid sutures as well as the palatine fissure all provide strong evidence for maintained suture patency even in advanced age (up to 11 years). In our study we identified two types of syndesmoses, type A with loose connective tissue lying parallel to the suture line and consisting of a high amount of osteoblaststs at the sutural edges. Type B was considered to have dense connective tissue which is orientated in a ninty degree to the suture line and contained low numbers of osteoblasts. Exemplary findings of coronal, parieto-sphenoidal, parieto-interparietal and lamboid type B syndesmoses in older dogs (3 to 10 years) lead to the suggestion that this might be a form of inactive (functional closed) suture. The sutural space is reduced to~100-200 μm but not bridged with bony tissue and remain as a syndesmosis. Cranial sutures in mammals do not necessarily fuse when growths stops or slows down suggesting that they have an additional role [18]. The transformation of the sutural structure allows flexibility and energy absorption in the skull bones and reduces the risk of skull fractures in mature [33], which is why they can remain in mature animals. The fact that the end-point of sutural development in the dog is not necessarily determined upon fusion demonstrates the impossibility to determine the physiological end point of bone growth in the suture on the basis of imaging techniques. A pathological craniosynostosis on the other hand might be diagnosed using CT or MRI.
Our study presents one important limitation. Histological sections without visible sutures were rated as fused. This finding might be based on the presence of a truly ossified suture, or because the suture was left out during sample cutting. Relatively simple at birth, the microarchitecture of many sutures gain complexity during growth. We observed a three-dimensional structure and their internal course must not necessarily correspond to the visible suture on the surface. The validity of our results might have increased further if the whole excised bone/suture sample would have been fully sliced, but we refrained from the examination of the whole suture course as this process is extremely time consuming.

Conclusion
Conventional imaging techniques are very useful to assess the open or closed status of synchondroses within the cranial base. Sutural closure is difficult to diagnose in MRI and CT. Assessment is dependent on observer experience. Histologic examination remains the gold standard for suture assessment.

Methods
The aim of the present study was, therefore, to compare imaging findings with histological preparations in order to determine the sensitivity and specificity for conventional clinical CT and MRI in the assessment of the open or closed status of cranial synchondroses and sutures in dogs.

Animals
The heads of 40 dogs of different breeds (24 mesaticephalic-and 16 brachycephalic dogs), were collected from the Department of Veterinary Clinical Sciences, Clinic for Small Animals and from the Clinic for Gynecology, Andrology and Obstetrics of the Justus-Liebig-University, Giessen (Table 4). The dogs were euthanized or died due to diseases unrelated to the skull and central nervous system. Age and sex was recorded. Written consent was obtained from all owners that donated their animals for the study and actual dogs remained anonymous.

Imaging techniques
Post mortem MRI of the head was obtained using a high field scanner ( Gyroscan Intera, 1.0 T, Phillips, Hamburg, Germany). Dorsal, sagittal, and transvers T2-weighted (W) images with a 2 mm slice thickness and a 0.2 mm slice interval were acquired (T2-Turbospin echo, echo time (TE) 120 ms, repetition time (TR): 2900 ms. T1-FFE weighted dorsal and transversal images with a slice thickness of 1 mm were obtained (TR 25 ms TE 6.9 ms). Field of view was 120 × 120 mm in small dogs and 210 × Post mortem CT examination was performed using a sixteen slice CT scanner ( Philips Brilliance 16, Phillips, Hamburg, Germany ) . Transvers images with a slice thickness of 0.8 mm were obtained using 120 kV, 321 mA and a field of view of 133 mm. Data were processed using a bone algorithm (window width 2500 and window level 500).

Assessment of synchondrosis and suture status in CT
In CT, an open synchondrosis or suture was defined as a hypodense zone between well-defined hyperdense borders of the calvaria or cranial base bones (Fig. 6). A closed suture or synchondrosis was defined as a continuous hyperdense bone, without interruption by a hypodense structure. Partial closure was defined as a non-continuous hypodense bone and the presence of bone-isointense bridges within this hypointense bone structure.

Assessment of synchondrosis and suture status in MRI
In MRI, an open suture was defined as a hypointense signal interruption of the hyperintense calvarial bone marrow signal in T2 weighted images. An open synchondrosis was defined based on the presence of a broad hyperintense signal zone (cartilage) with well defined, hypointense borders (endplates) (Fig. 7). A closed suture was defined as a lack of a hypointense signal within the hyperintense bony structures. A closed synchondrosis was defined based on the obliteration of the synchondrosis, replacement of cartilage and development of a continuous isointense signal from the bone marrow cavity. Partial closure was defined as bony bridges within the suture or synchondrosis, visible as a partial hypointense signal of a suture within the hyperintense bone, or as a hypointense signal within the hyperintense cartilage signal of the synchondrosis, with a narrowing of the synchondrosal cartilage.

Histological sample preparation
After postmortem scanning soft tissues were carefully removed from the skull and the whole head was immersion fixed in 10% neutral buffered formalin. A standardized procedure was devised to cut out the sutures and synchondroses from each head using a diamond coated sawblade (Proxxon Micromot System ) . Suture morphology was assessed as precisely as possible in a plane of sectioning perpendicular to the individual suture orientation [18]. Not all sutures could be macroscopically identified in all skull specimens. These individual sutures were not included in the calculation of sensitivity and specificity. Samples were washed with 1% phosphate buffered saline and decalcified over 4 weeks with Ethane-(1,2-diyldinitrilo) tetraacetic acid with a pH = 8. After decalcification, specimens were embedded in paraffin (Parablast, Sherwood ) . Three serial sections of 5 μm were stained using Masson-Goldner protocol in order to visualize bone, cartilage and fibrous connective tissue. The stain produces blue or green collagen and deep red mineralized bone.

Histomorphological analysis of the sutures
Histological slides were evaluated for structural characteristics indicating an open or closed state of the sutures.
Two types of open sutures were defined according to the type of connective tissue which dominated the sutural space. The loose connective tissue type lying parallel to the suture line was defined as open suture type A and contained a high amount of osteoblasts (> 50 in a mean of 3 field of views in a magnification of 40 (FOV)) and fibroblasts (> 20 in 3 FOV)) ( Fig. 4a.). The dense connective tissue which is orientated in a ninty degree to the suture line was characterized as open type B, it contained low numbers of osteoblasts (< 10 in 3 FOV) and fibroblasts (< 20 in 3 FOV) (Fig. 4b.). Partly closed sutures describes bony bridges within the suture. In some cases no suture could be identified in the collected sample.

Histomorphological analysis of the synchondroses
The synchondroses were defined as open when there was a continuous zone of cartilage between the bones (Fig. 1), and as closed when there was no cartilage visible and there was bony tissue showing a continuing medullary cavity. Partly synostosis was defined as non-continouous cartilage next to bony tissue between the periostal borders.
Each section was histologically analyzed in a two-step procedure. First, the existence or non-existence of a fibrous tissue or cartilage within the synchondrosis/suture was determined by use of a histological overview (40x magnification). Non-existence of cartilage or fibrous tissue was evaluated as a closed synchondrosis/suture. Furthermore, it was determined whether fibrous-or cartilage tissue was present throughout the suture/ synchondrosis, or whether bony bridges were present. These data was used to assign each synchondrosis/suture to one of the following categories: 1) open synchondrosis/ suture; 2) partly closed synchondrosis /suture; 3) closed synchondrosis/suture (Figs. 1, 2). Morphological analysis of sutural appearance, collagen structure within the sutural space and amount of osteoblasts, fibroblasts and blood vessels was also performed.

Statistical analysis
All statistical analyses were performed using the statistical software package BMDP. Sutures classified as open in MRI and CT were compared to type A syndesmoses. Sutures classified as closed in MRI and CT were compared to type B syndesmoses indicating an inactive or functionally closed condition. As the partially closed state does not allow longitudinal extension partially synostosis was also considered as a closed synchondrosis or suture and summed up for statistical analyses. An exact-fisher test was performed on statistical data to compare different suture assessment techniques of histological examination, MRI and CT. A high sensitivity means, in this regard that a synchondrosis or suture, which was assessed as open in imaging was also assessed as open in the histological examination. A high specificity means that the closed status of a synchondrosis or suture in MRI and CT was confirmed by histological evaluation. Sensitivity and specificity in MRI and CT were defined as high (> 80), moderate (65-79%) or low (< 64%). Interobserver variability was obtained using the kappa coefficient. A kappa value < 0.2 implies slight agreement, 0.21-0.4 fair agreement, 0.41-0.6 moderate agreement, 0.61-0.8 substantial agreement and > 0.81 almost perfect agreement.