Faculty of Health Sciences School of Oral Health Sciences 9401428a vcvibha / There is considerable confusion regarding the origin and classification of the AOT with the most recent WHO classification including the AOT as a non-inductive tumour or hamartoma showing metaplastic mineralisation. This study reviewed the clinical and epidemiological features of 51 AOTs retrieved from the archives of the Division of Oral Pathology, University of the Witwatersrand. In addition a detailed histological analysis, including histochemical and immunohistochemical investigations, was undertaken with a view to provide evidence for induction in AOTs. 4μ haematoxylin and eosin sections were examined. Selected cases were stained with PAS, alcian blue at pH 2.5, Congo red, reticulin, mucicarmine, von Gieson, Masson’s trichrome and Prussian blue. Melanin bleach was performed on certain sections. Immunohistochemistry was performed in the presence of adequate preparations and controls with MNF 116 and Vimentin antisera. Analysis of the clinical and epidemiologic data revealed that the AOT in our series had the same clinicopathological features as those reported from other parts of the world. This data will be included in a review article being prepared to commemorate the 100th anniversary of the description of this lesion. AOTs occur in both follicular (64%) and extrafollicular forms (21%) most commonly in the anterior maxilla (62.7%) in females (63.6%) in the second decade (66.6%) where they are frequently associated with unerupted canines (42%). There is some evidence ii suggesting that extrafollicular AOTs may originate in other odontogenic cysts and that this might explain why some AOTs grow to a large size and behave aggressively causing root resorption and expansion. Histologically the unique and important presence of tall columnar cells resembling ameloblasts or odontoblasts were identified in 5 cases of AOT. These cells were arranged in a circular configuration and were actively secreting PAS positive material, which we have interpreted as dental matrix material. We have called these ‘circular secretory units’. The tall columnar cells did not always surround the entire secretory unit suggesting either that there was a variable rate of differentiation or that having completed their function these cells change shape and become unrecognisable. We regard these circular secretory units as providing definite evidence of induction. Further evidence of induction is provided by the presence of clusters or strands of odontogenic epithelium intimately associated with a lace-like pattern of dental matrix material. No evidence of residual ectomesenchyme was found, but this does not rule out the possibility that induction has indeed occurred. We can also find no evidence linking the circular secretory units with the pseudo-ductular spaces, which characterise the AOT. In conclusion, based on our observations, we recommend that the AOT be classified as a benign tumour with inductive capacity.

Le cancer du sein triple négatif (CSTN) représente 10-15% des cancers du sein. Son pronostic est sombre en particulier face à la rareté des thérapies ciblées adaptées à ce sous type. Sa complexité de prise en charge est directement liée à sa grande hétérogénéité tant au niveau moléculaire que morphologique.Dans ce contexte, nous avons développés des modèles de Patient Derived Xenograft (PDX) issus de CSTN. Ce modèle, robuste, a la particularité de retenir les caractéristiques (histologiques, génotypiques mais aussi phénotypiques) des tumeurs observées chez les patients.Dans notre cohorte de 61 PDX de CSTN, nous avons confirmé l’hétérogénéité anatomopathologique et génomique de ce sous type. Les différentes anomalies moléculaires mises en évidence sont de faible fréquence (<10%) mais 88% de nos modèles ont une altération potentiellement ciblables et plus de la moitié ont au moins 2 altérations ciblables. Nous nous sommes particulièrement intéressés à 2 sous types de CSTN : (i) le sous -type LAR (Luminal Androgen Receptor) dont nous avons décrit les premiers modèles de PDX : ces modèles présentent des altérations fréquentes de la voie PI3K ainsi que des réponses majeures aux inhibiteurs de cette voie ; (ii) le sous type métaplasique, dont 4 de nos 9 modèles présentent une double altération genomique dans les voies PI3K et RTK-MAPK ainsi que des réponses complètes et durables à la combinaison d’inhibiteurs de PI3K et de MAPK.Dans les autres sous-types de CSTN, nous avons également mis en évidence des taux de réponse importants aux inhibiteurs de la voie PI3K et MAPK. Les biomarqueurs de réponse à ces différentes thérapies ciblées testées sont en cours d’étude en particulier par intégration des données génomique et protéique de nos modèles. / Triple negative breast cancer (TNBC) accounts for 10-15% of breast cancers. Its prognosis is worse, particularly due to the rarity of targeted therapies adapted to this subtype. Its complexity of management is directly related to its high heterogeneity, both at the morphological and genomical levels.In this context, we developed Patient Derived Xenograft (PDX) models from TNBC. This robust model has the specificity of retaining the characteristics (histological, genotypic but also phenotypic) of the tumors observed in patients.In our cohort of 61 PDXs of TNBC, we confirmed the anatomopathological and genomical heterogeneity of this subtype. Majority of targeted alterations are of low frequency (<10%) but 88% of our models harbour a potential targetable alteration and more than half have at least 2 targetable alterations. We were particularly interested in 2 subtypes of TNBC: (i) the LAR subtype for which we have described the first PDX models: these models present frequent alterations of the PI3K pathway as well as major responses to PI3K inhibitors; (ii) the metaplastic subtype, of which 4 of our 9 models show double alterations in the PI3K and RTK-MAPK pathways and complete and durable responses to the combination of PI3K-MAPK inhibitors.In the other CSTN subtypes, we have also demonstrated significant response rates to PI3K and MAPK inhibitors. Biomarkers of response to these various targeted therapies tested are being studied, in particular by integrating the genomic and protein data from a higher number of PDX models.