1Royal (Dick) School of Veterinary Studies, University of Edinburgh, Roslin, United Kingdom

1 Morphological variation of the caudal fossa of domestic cat skulls assessed with computed

2 tomography and geometric morphometrics analysis

4 Caroline R Gordon1, Thomas W Marchant2, Joanna Lodzinska1, Jeffery J Schoenebeck2, Tobias

5 Schwarz1

7 1Royal (Dick) School of Veterinary Studies, University of Edinburgh, Roslin, United Kingdom

8 2Roslin Institute, University of Edinburgh, Roslin, United Kingdom

10 Corresponding author: Caroline R Gordon BVSc MRCVS

11 [email protected]

12 Hospital for Small Animals, Royal (Dick) School of Veterinary Studies, University of Edinburgh

13 Midlothian EH25 9RG, United Kingdom

14 Tel: +44 (0) 7729 246706

16 Keywords: caudal fossa, foramen magnum, computed tomography, geometric morphometrics

17 analysis, feline, occipital

19 Author Note

20 Preliminary results of this study were presented at the 2016 European Veterinary Diagnostic Imaging

21 Conference in Wroclaw, Poland.

24 25 Abstract

26 Objectives This study aimed to investigate differences and demonstrate a normal range of

27 morphological variation of the caudal fossa of the cranium of domestic cats.

28 Methods Computed tomographic scans of thirty-two domestic cat heads of eleven breeds were

29 included. Isosurfaces from skulls were characterised through three-dimensional geometric

30 morphometrics using geographic landmarks placed on the internal surface of the caudal fossa and

31 foramen magnum. Raw data was transformed with a Procrustes fit and coordinate covariance was

32 analysed by principal components to establish breed- and sex-level differences. Skulls were also

33 classified according to the number of concavities along the mid sagittal vermiform impression.

34 Differences were investigated between breed groups and sex, and correlation was sought with age.

35 Results Analyses revealed size-independent differences in occipital bone morphology across breeds

36 and sex, however no clustering was evident. Most variability was observed at the exoccipital bones,

37 ventral portion of the supraoccipital bone, dorsum sellae of the basisphenoid, and the osseous

38 tentorium cerebelli. No statistically significant differences were identified via 2-sample t-tests

39 between breed groups or sexes. No statistically significant correlation using Spearman Rho

40 correlation coefficient was identified with age.

41 Conclusions and relevance The feline caudal fossa displays a wide range of intra- and inter-breed

42 variation, not linked to age or sex. Concavities along the vermiform impression have not previously

43 been described. As advanced imaging modalities are becoming more frequently used for domestic

44 felids, an established range of normality is important for discriminating pathological changes from

45 anatomical variances.

49 50 Introduction

51 The caudal fossa has been the focus of extensive research in humans and canids, in part due to its role

52 in the pathogenesis of various disorders such as Chiari and Chiari-like malformations as well as

53 occipital dysplasia.1-8 There is, however, a paucity of information regarding variability in skull

54 morphology amongst existing breeds of the domestic cat (Felis catus). With the increasing use of

55 advanced diagnostic imaging in this species, recognition of the breadth of normal anatomical

56 variation is crucial. In the authors’ experience, cats display a wide variability in caudal fossa shape in

57 the absence of associated clinical signs.

59 The caudal fossa of the cranial cavity encompasses the midbrain, pons, and medulla oblongata

60 ventrally and the cerebellum dorsally. Rostrally it extends to the dorsum sellae of the sphenoid bone

61 and the osseous tentorium cerebelli of the occipital bone. Caudally it is enclosed by the supraoccipital

62 bone, which forms the dorsal border of the foramen magnum.9 The internal surface of the

63 supraoccipital bone displays shallow depressions which conform to the surface of the cerebellum; the

64 vermiform impression.9 The basioccipital bone forms the ventral border of the foramen magnum and

65 joins the basisphenoid via a cartilaginous suture. The exoccipital bones form the lateral portions of the

66 foramen magnum and bear the convex occipital condyles.9

68 The objectives of this study were to investigate and describe differences in caudal fossa shape and

69 determine whether these differences are stereotyped by breed or sex. To detect shape trends of the

70 caudal fossa, the authors used computed tomography, an imaging modality that is amenable to

71 geometric morphometrics analysis.

73 Materials and methods

74 Sample population 75 Imaging records of the authors’ academic veterinary referral hospital were searched retrospectively

76 between January 2011 and December 2015 to identify adult cats (>12 months of age) that had

77 undergone a CT examination of the entire head to at least the level of the second cervical vertebra.

78 Studies were excluded if there were any pathological cranial or neural abnormalities identified on the

79 imaging study or clinical records. All CT examinations were acquired with the same 4-slice CT unit

80 (Siemens Somatom) using a 1mm slice width, 1.5 pitch, 130mAs, 120kV, 0.75s tube rotation time, and

81 high resolution image kernels (proprietary term U90).

83 Three-dimensional landmarking

84 Skull isosurfaces were generated from DICOM images using Stratovan Checkpoint

85 (v2016.11.21.0711). Twelve geographical landmarks were placed on the internal surface of the caudal

86 fossa and foramen magnum of each skull (Figure 1a,b) to represent the overall shape and dimensions.

87 Specifically, landmarks were placed at the osseous tentorium cerebelli, along the internal sagittal

88 curvature of the caudal fossa, opisthion, basion, the basioccipital bone, dorsum sellae, and the lateral-

89 most points of the internal surface of the foramen magnum.

91 Geometric morphometric analysis

92 Landmark coordinates and classifier data were imported into MorphoJ (v1.06d)10 for geometric

93 morphometric analysis. Superimposition, translation, rotation, and scaling of landmark

94 configurations were performed via a generalised Procrustes fit prior to generating a covariance

95 matrix. A one-way analysis of variance (ANOVA) was used to evaluate differences in centroid size

96 (the square root of the sum of squared distances of landmarks from the central point) amongst breed

97 groups. A multivariate regression of the symmetrical component of shape versus log centroid size

98 was performed to remove the influence of allometry, the shape changes correlating to changes in size.

99 Principal components analysis (PCA) is a mathematical algorithm which reduces and expresses a 100 dataset in terms of vectors along which variation is maximal. A PCA of the resulting regression

101 residuals11 was used to explore patterns of covariance amongst landmarks.

103 Classification

104 A midline sagittal CT image was created for each cat using multi-planar reconstruction on freely

105 available DICOM viewing software (Horos v2.0). The skulls were classified into groups based on the

106 number of concavities along the vermiform impression of the caudal fossa. One reader, a second year

107 radiology resident, categorised each image, blinded to the signalment of each case. A two-sample t-

108 test was used to evaluate differences in concavity number between sexes, and between breed groups.

109 A Spearman rank correlation coefficient (rs) sought correlations between the number of concavities

110 and age, PC1, PC2, and centroid size of the caudal fossa. A p-value of < 0.05 was considered

111 significant.

113 Results

114 The study population comprised of thirty-two CT scans of domestic cat heads. There were two entire

115 females, one entire male, eleven neutered females, and eighteen neutered males. Entire and neutered

116 cats were grouped together for statistical purposes; thus the overall study numbers were thirteen

117 (41%) female and nineteen (59%) male. Eleven breeds were represented including Abyssinian (n=1),

118 British shorthair (n=1), Bengal (n=1), Burmese (n=2), Chinchilla (1) domestic long hair (DLH) (n=3),

119 domestic short hair (DSH) (n=15), Maine Coon (n=4), Ragdoll (n=1), Siamese (n=2), and Tonkinese

120 (n=1). For statistical purposes, cats were grouped according to phylogenetic origin into Asian cats

121 (Burmese, Siamese, Ragdoll, Tonkinese; n=6), Western European cats (Abyssinian, Maine Coon,

122 British shorthair, Chinchilla; n=7), and mixed breeds (DLH, DSH, Bengal; n=19)12. CT images of all

123 cats were sufficiently detailed to allow consistent placement of all landmarks on stereotyped locations

124 of the caudal fossa and foramen magnum.

125 126 Geometric morphometrics

127 For the symmetric component of shape with allometry, principal component (PC) 1 accounted for

128 21.6% of the variation amongst all cats. PC2 accounted for 15.4% of variation. PC1 and PC2 thus

129 cumulatively accounted for 37.0% of the observed variation. Most variability explained by PC1 was

130 observed at the exoccipital bones and dorsum sellae. As the exoccipital markers shifted caudally, the

131 dorsum sellae shifted dorsally, whilst the roof of the cranial fossa shifted, less markedly, in an overall

132 rostral direction. Neighbouring landmarks along the roof of the caudal fossa varied between

133 rostroventral and rostrodorsal trends. Variability explained by PC2 was observed at the osseous

134 tentorium cerebelli and foramen magnum. As the tentorium shifted ventrally, the basion and ventral

135 portion of the supraoccipital bone shifted dorsally and caudally, respectively. None of the PCs

136 discriminated subpopulations based on sex (Figure 2a) or breed groups (Figure 2b). Western

137 European cat breeds demonstrated the largest mean caudal fossa size and Asian cats the smallest,

138 however centroid sizes were not significantly different between breed groups (p = 0.432) (Figure 3).

140 After controlling allometry by regressing out isometric size, PC1 accounted for 20.3%, and PC2 16.0%

141 of the shape variation, cumulatively 36.3%. Overall shape trends matched the raw data in terms of

142 direction, however varied slightly in increments. Size-corrected PCA revealed no segregation of PC

143 scores between males and females (Figure 4a), nor was there any clustering amongst breeds (Figure

144 4b).

146 Occipital bone classification

147 Three cats displayed zero internal concavities along the vermiform impression, fourteen displayed

148 one concavity, eight cats displayed two, and seven cats displayed three (Figure 5a,b,c,d and Tables 1

149 and 2). No statistically significant differences were identified between the number of concavities

150 along the mid sagittal vermiform impression of males compared to females (p = 0.917), nor was there

151 significant difference between the mixed breeds and purebreds (p = 0.634). No significant correlation 152 was identified between the number of concavities and age (p = 0.527), PC1 (p = 0.562), PC2 (p = 0.873),

153 and centroid size (p = 0.086).

155 Table 1 Relationship between vermiform impression morphology and sex

Concavities Male Female 156

0 2 (11) 1 (8) 1 8 (42) 6 (46) 2 5 (26) 3 (23) 3 4 (21) 3 (23) Total 19 (100) 13 (100) 157 Data are n (%) unless otherwise specified

159 Table 2 Relationship between vermiform impression morphology and breed group

Concavities Purebred Mixed breed 160

0 2 (15) 1 (5) 1 5 (38) 9 (47) 2 3 (23) 5 (26) 3 3 (23) 4 (21) Total 13 (99) 19 (99) 161 Data are n (%) unless otherwise specified

163 Discussion

164 A wide range of morphometric variation of the caudal fossa and foramen magnum was observed in

165 this study population of neurologically normal cats, with no identified links to breed, sex, or age.

166 Because PC shape trends were similar between the raw data and the regression residuals, and there

167 were no significant differences in centroid sizes between breed groups, the authors conclude that

168 allometry had negligible influence on shape differences of the caudal fossa.

170 Shape variability demonstrated by PC1 may be interpreted anatomically as a lengthening of the

171 exoccipital bones occurring alongside a relative dorsoventral flattening of the entrance to the caudal

172 fossa caused by dorsal shifting of the dorsum sellae. Rostral shifting of the roof of the caudal fossa 173 reflects an alteration in space distribution. A previous morphometric study categorised feline skulls

174 into three phenotypically different skull formations; rounded, triangular, and cuneiform.13 This study

175 identified a frequent variant of the round-shaped skulls where the foramen magnum extended more

176 distantly between the occipital condyles thereby acquiring a bilaterally narrowed oval shape, similar

177 to the findings of the current study. Variability of rostroventral versus rostrodorsal directions of this

178 shift amongst neighbouring landmarks of the caudal roof of the fossa demonstrates an irregular

179 contour of this internal bony surface. The vermiform impression is described in dogs as an irregular

180 excavation of the median portion on the cerebellar surface of the squamous part of the occipital

181 bone.14 This is, to the authors’ knowledge, the first description of this anatomical feature specifically

182 in the cat.

184 Shape variability demonstrated by PC2 may be interpreted anatomically as an overall dorsoventral

185 flattening of the entrance to the caudal fossa owing to ventral shifting of the osseous tentorium

186 cerebelli. Caudal extension of the foramen magnum coinciding with relative dorsoventral narrowing

187 of the foramen reflects an overall redistribution of space. The ventral portion of the supraoccipital

188 bone shifted caudally, independently of the more dorsal landmarks, demonstrating a variable caudal

189 angulation of this site relative to the internal contour of the caudal fossa. Similar findings have been

190 reported in a recent MRI study of cats which described variants of the ventral aspect of the

191 supraoccipital bone of cats being straight throughout, caudally angulated, blunt ended, and/or

192 ventrally thickened.15 The same study also found that cats could be divided into those with a thin and

193 those with a thick occipital bone. Compared with MRI, CT has a higher sensitivity for accurately

194 displaying osseous contours, and adds further validation to these findings.

196 Morphology of the external squamous part of the occipital bone and foramen magnum of cats has

197 been described in a study involving gross morphology of 50 European cat cadavers.16 Two

198 morphological patterns of the squamous part of the occipital bone were identified; a triangular shape 199 and a semi-oval shape, denoting normal variation of this external surface amongst European cats.

200 None of the specimens in this previously reported study displayed a dorsal notch, nor any other

201 defect at the foramen magnum. The authors concluded that, in the European cat, the foramen

202 magnum is free from pathology and its shape is conservative. In contrast, a morphometric study of 69

203 purebred cat skulls identified two individuals with a keyhole shaped foramen.13 Dogs display a wide

204 diversity of shape of the foramen magnum, many of which are considered as physiological variations

205 of morphology.8,17 It is apparent that a degree of physiological variation is also present in the cat and

206 that the spectrum of morphologies observed in purebreds is also seen in mixed-breed cats. From a

207 clinical perspective, this variation is important to recognise so that pathological processes are not

208 falsely assigned.

210 The majority (69%) of the study population displayed one or two concavities along the vermiform

211 impression on median plane CT images. No sexual dimorphism was identified, nor did these

212 concavities differ between purebreds and mixed breeds, nor correlate with age. Concavities along the

213 vermiform impression have not previously been described. A certain degree of care must be taken

214 comparing the appearance of the caudal fossa on CT with the anatomy. Post mortem examinations to

215 confirm the shape of the caudal fossa were not performed because none of these clinical cases were

216 available for necropsy examination. The high spatial frequency allowed by CT, however, ensures a

217 detailed and accurate representation of the osseous surface.18 Although the current study did not

218 identify significant differences between breed groups or sexes, or correlate with age, it is evident that

219 the shape of this vermiform impression is not uniform amongst cats and these variants should be

220 clinically recognised.

222 A limitation of this study is the relatively small sample size. The lack of segregation in the PCA and,

223 likewise, the lack of differences amongst vermiform impression groups may reflect the small numbers

224 of individuals representing each breed, age, and sex, and thus the inability to achieve statistical 225 significance. Categorisation of vermiform impression concavities was somewhat subjective and prone

226 to bias due to the use of only one evaluator. Additional studies inclusive of larger groups of feline

227 breeds would be recommended to further investigate interbreed morphology differences. Breeds of

228 particular interest are those that are brachycephalic, as brachycephaly has been shown in dogs to be

229 associated with Chiari-like malformation19 and occipital dysplasia.17 Further studies could also

230 include lateralised landmarks along the caudal fossa, not included in the current study, which may

231 uncover further variants.

233 Conclusions

234 A wide range of inter-breed morphometric variation of the feline caudal fossa was observed in this

235 population, which was not linked to age or sex, demonstrating that caudal fossa anatomy is not

236 homogeneous amongst cats. The findings support and expand on those previously reported13,15,16. An

237 established range of normality is important for accurate discrimination between pathological changes

238 and normal anatomical variants, and for the recognition of potential trends associated with disease in

239 this popular companion animal.

241 Conflict of Interest

242 The authors declared no potential conflicts of interest with respect to the research, authorship, and/or

243 publication of this article.

245 Funding

246 The authors received no financial support for the research, authorship, and/or publication of this

247 article.

248 249 Figure 1 Landmarks along the internal surface of the caudal fossa (a) and foramen magnum (b).

250 Images are not to scale. In (b), a portion of the occipital bone has been cropped out.

251 Figure 2 PC1 vs PC2 of the symmetric component displaying PC scores classified by sex (a) and

252 breed (b)

253 Figure 3 Box and whisker diagram of centroid size amongst breed groups. The horizontal line within

254 each box represents the median.

255 Figure 4 PC1 vs PC2 after controlling for allometry displaying PC scores classified by sex (a) and

256 breed (b)

257 Figure 5 Concavities along the vermiform impression; (a) zero (b) one (c) two (d) three

259 References

260 1. Noudel R, Jovenin N, Eap C, et al. Incidence of basioccipital hypoplasia in Chiari

261 malformation Type I: comparative morphometric study of the posterior cranial fossa. J

262 Neurosurg 2009; 111: 1046-1052.

264 2. Urbizu A, Poca MA, Vidal X, et al. MRI-based morphometric analysis of posterior cranial

265 fossa in the diagnosis of Chiari malformation type I J Neuroimaging 2013; 24: 250-256.

267 3. Yan H, Han X, Jin M, et al. Morphometric features of posterior cranial fossa are different

268 between Chiari I malformation with and without syringomyelia Eur Spine J 2016; 25: 2202-

269 2209.

271 4. Shaw TA, McGonnell IM, Driver CJ, et al. Caudal cranial fossa partitioning in cavalier King

272 Charles spaniels Vet Rec 2013; 172: 341.

274 5. Knowler SP, McFayden AK, Freeman C, et al. Quantitative analysis of Chiari-like

275 malformation and syringomyelia in the Griffon Bruxellois dog PLoS ONE 2014; 9: e88120. 276

277 6. Knowler SP, Kiviranta AM, MacFayden AK et al. Craniometric analysis of the hindbrain

278 and craniocervical junction of Chihuahua, Affenpinscher, and cavalier King Charles

279 spaniel dogs with and without syringomyelia secondary to Chiari-like malformation PLoS

280 ONE 2017; 12: e0169898.

282 7. Rusbridge C, Knowler SP. Coexistence of occipital dysplasia and occipital

283 hypoplasia/syringomyelia in the cavalier King Charles spaniel J Small Anim Pract 2006; 47:

284 603-606.

286 8. Watson AG, de Lahunta A, Evan HE. Dorsal notch of the foramen magnum due to

287 incomplete ossification of the supraoccipital bone in dogs J Small Anim Pract 1989; 30: 666-

288 673.

290 9. König HE and Liebich HG (eds). Veterinary Anatomy of Domestic Mammals. Stuttgart,

291 Germany: Schattauer GmbH, 2004.

293 10. C. P. Klingenberg. 2011. MorphoJ: an integrated software package for geometric

294 morphometrics. Molecular Ecology Resources 11: 353-357.

296 11. Klingenberg CP, Barluenga M, Meyer A. Shape analysis of symmetric structures:

297 quantifying variation among individuals and asymmetry Evolution 2002; 56: 1909-1920.

299 12. Lipinski MJ, Froenicke L, Baysac KC et al. The ascent of cat breeds: Genetic evaluations of

300 breeds and worldwide random-bred population Genomics 2008; 91: 12-21.

301 302 13. Kunzel W, Breit S, Oppel M. Morphometric investigations of breed-specific features in

303 feline skulls and considerations on their functional implications Anat Histol Embryol 2003;

304 32: 218-223.

306 14. Evans HE and de Lahunta A (eds). Miller's anatomy of the dog. 4th Edition. St Louis, MO:

307 Elsevier, 2013.

309 15. Huizing X, Sparkes A, Dennis R. Shape of the feline cerebellum and occipital bone related

310 to breed on MRI of 200 cats J Feline Med Surg Prepublished November 10 2016, DOI:

311 10.1177/1098612X16676022.

313 16. Wasowicz M, Kupczyńska M, Wielądek A et al. Morphometric analysis of occipital bone in

314 the domestic cat in comparison with selected skull size parameter and with special regard

315 to skull morphotype Pol J Vet Sci 2009; 12: 251-258.

317 17. Kupczyńska M, Czubaj N, Barszcz K et al. Prevalence of dorsal notch and variations in the

318 foramen magnum in dogs of different breeds and morphotypes Biologia 2017; 72: 230-237.

320 18. Ruff CB and Leo FP. Use of computed tomography in skeletal structure research Yearb Phys

321 Anthropol 1986; 29: 181-196.

323 19. Schmidt MJ, Neumann AC, Amort KH et al. Cephalometric measurements and

324 determination of general skull type of cavalier King Charles spaniels Vet Radiol Ultrasound

325 2011; 52: 436-440.