Digital pathology-based artificial intelligence models for differential diagnosis and prognosis of sporadic odontogenic keratocysts

Abstract

Odontogenic keratocyst (OKC) is a common jaw cyst with a high recurrence rate. OKC combined with basal cell carcinoma as well as skeletal and other developmental abnormalities is thought to be associated with Gorlin syndrome. Moreover, OKC needs to be differentiated from orthokeratinized odontogenic cyst and other jaw cysts. Because of the different prognosis, differential diagnosis of several cysts can contribute to clinical management. We collected 519 cases, comprising a total of 2 157 hematoxylin and eosin-stained images, to develop digital pathology-based artificial intelligence (AI) models for the diagnosis and prognosis of OKC. The Inception_v3 neural network was utilized to train and test models developed from patch-level images. Finally, whole slide image-level AI models were developed by integrating deep learning-generated pathology features with several machine learning algorithms. The AI models showed great performance in the diagnosis (AUC = 0.935, 95% CI: 0.898-0.973) and prognosis (AUC = 0.840, 95%CI: 0.751-0.930) of OKC. The advantages of multiple slides model for integrating of histopathological information are demonstrated through a comparison with the single slide model. Furthermore, the study investigates the correlation between AI features generated by deep learning and pathological findings, highlighting the interpretative potential of AI models in the pathology. Here, we have developed the robust diagnostic and prognostic models for OKC. The AI model that is based on digital pathology shows promise potential for applications in odontogenic diseases of the jaw.

References

1. Boffano, P.et al.The epidemiology and management of odontogenic keratocysts (OKCs): a European multicenter study.J. Craniomaxillofac. Surg. 50, 1-6 (2022).
2. Li T. J.The odontogenic keratocyst: a cyst, or a cystic neoplasm?J. Dent. Res. 90, 133-142 (2011).
3. Mendes, R. A., Carvalho, J. F.& van der Waal, I. Characterization and management of the keratocystic odontogenic tumor in relation to its histopathological and biological features.Oral. Oncol. 46, 219-225 (2010).
4. Dhanuthai, K.et al.Cysts of the jaws: a multicentre study. Oral Dis., https://doi.org/10.1111/odi.14722(2023).
5 5.Wang, Y. J.et al. [Clinicopathological analysis of 844 cases of odontogenic keratocysts].Beijing Da Xue Xue Bao Yi Xue Ban. 52, 35-42 (2020).
6. Noy D., Rachmiel A., Zar K., Emodi O.& Nagler, R. M. Sporadic versus syndromic keratocysts-Can we predict treatment outcome? A review of 102 cysts.Oral. Dis. 23, 1058-1065 (2017).
7. Bresler, S. C., Padwa, B. L.& Granter, S. R. Nevoid basal cell carcinoma syndrome (Gorlin Syndrome).Head. Neck Pathol. 10, 119-124 (2016).
8. Pan S., Xu L. L., Sun L. S.& Li, T. J. Identification of known and novel PTCH mutations in both syndromic and non-syndromic keratocystic odontogenic tumors.Int J. Oral. Sci. 1, 34-38 (2009).
9. Vered M.& Wright, J. M. Update from the 5th edition of the World Health Organization classification of head and neck tumors: odontogenic and maxillofacial bone tumours.Head. Neck Pathol. 16, 63-75 (2022).
10. Wang Y. J., Zhang J. Y., Dong Q.& Li, T. J. Orthokeratinized odontogenic cysts: a clinicopathologic study of 159 cases and molecular evidence for the absence of PTCH1 mutations.J. Oral. Pathol. Med. 51, 659-665 (2022).
11. Dong Q., Pan S., Sun L. S.& Li, T. J. Orthokeratinized odontogenic cyst: a clinicopathologic study of 61 cases.Arch. Pathol. Lab Med. 134, 271-275 (2010).
12. Kaczmarzyk, T.et al.Investigation of clinicopathological parameters and expression of COX-2, bcl-2, PCNA, and p53 in primary and recurrent sporadic odontogenic keratocysts.Clin. Oral. Investig. 22, 3097-3106 (2018).
13. Cunha, J. F.et al.Clinicopathologic features associated with recurrence of the odontogenic keratocyst: a cohort retrospective analysis.Oral. Surg. Oral. Med. Oral. Pathol. Oral. Radio. 121, 629-635 (2016).
14. Cottom H. E., Bshena F. I., Speight P. M., Craig G. T.& Jones, A. V. Histopathological features that predict the recurrence of odontogenic keratocysts.J. Oral. Pathol. Med. 41, 408-414 (2012).
15. Pan S.& Li, T. J. PTCH1 mutations in odontogenic keratocysts: are they related to epithelial cell proliferation?Oral. Oncol. 45, 861-865 (2009).
16. Kaplan I.& Hirshberg, A. The correlation between epithelial cell proliferation and inflammation in odontogenic keratocyst.Oral. Oncol. 40, 985-991 (2004).
17. Shmatko A.,Ghaffari Laleh, N., Gerstung, M. & Kather, J. N. Artificial intelligence in histopathology: enhancing cancer research and clinical oncology.Nat. Cancer 3, 1026-1038 (2022).
18. Foersch, S.et al. Multistain deep learning for prediction of prognosis and therapy response in colorectal cancer. Nat. Med., https://doi.org/10.1038/s41591-022-02134-1 (2023).
19. Bashir R. M.S. et al. A digital score of peri-epithelial lymphocytic activity predicts malignant transformation in oral epithelial dysplasia. J. Pathol., https://doi.org/10.1002/path.6094(2023).
20. Zheng, X.et al.A deep learning model and human-machine fusion for prediction of EBV-associated gastric cancer from histopathology.Nat. Commun. 13, 2790(2022).
21. Yang, S. Y.et al.Histopathology-based diagnosis of oral squamous cell carcinoma using deep learning.J. Dent. Res. 101, 1321-1327 (2022).
22. Wang, R.et al.Development of a novel combined nomogram model integrating deep learning-pathomics, radiomics and immunoscore to predict postoperative outcome of colorectal cancer lung metastasis patients.J. Hematol. Oncol. 15, 11(2022).
23. Vanguri, R. S.et al.Multimodal integration of radiology, pathology and genomics for prediction of response to PD-(L)1 blockade in patients with non-small cell lung cancer.Nat. Cancer 3, 1151-1164 (2022).
24. Saldanha, O. L.et al.Swarm learning for decentralized artificial intelligence in cancer histopathology.Nat. Med. 28, 1232-1239 (2022).
25. Fujii, S.et al.Rapid screening using pathomorphological interpretation to detect BRAFV600E mutation and microsatellite instability in colorectal cancer.Clin. Cancer Res. 28, 2623-2632 (2022).
26. Feng, L.et al.Development and validation of a radiopathomics model to predict pathological complete response to neoadjuvant chemoradiotherapy in locally advanced rectal cancer: a multicentre observational study.Lancet Digit. Health 4, e8-e17 (2022).
27. Chen, D.et al.Prognostic and predictive value of a pathomics signature in gastric cancer.Nat. Commun. 13, 6903(2022).
28. Klein, S.et al.Deep learning predicts HPV association in oropharyngeal squamous cell carcinomas and identifies patients with a favorable prognosis using regular H&E stains.Clin. Cancer Res. 27, 1131-1138 (2021).
29. Gehrung, M.et al.Triage-driven diagnosis of Barrett’s esophagus for early detection of esophageal adenocarcinoma using deep learning.Nat. Med. 27, 833-841 (2021).
30. Cai, X.et al.Development of a pathomics-based model for the prediction of malignant transformation in oral leukoplakia.Lab. Investig. 103, 100173(2023).
31. Lehman C. D.& Wu, S. Stargazing through the lens of AI in clinical oncology.Nat. Cancer 2, 1265-1267 (2021).
32. Bera K., Schalper K. A., Rimm D. L., Velcheti V.& Madabhushi, A. Artificial intelligence in digital pathology—new tools for diagnosis and precision oncology.Nat. Rev. Clin. Oncol. 16, 703-715 (2019).
33. Figueroa, A.et al.Keratocystic odontogenic tumor associated with nevoid basal cell carcinoma syndrome: similar behavior to sporadic type?Otolaryngol. Head. Neck Surg. 142, 179-183 (2010).
34. Qu, J.et al.Underestimated PTCH1 mutation rate in sporadic keratocystic odontogenic tumors.Oral. Oncol. 51, 40-45 (2015).
35. Qu, J.et al.PTCH1 alterations are frequent but other genetic alterations are rare in sporadic odontogenic keratocysts.Oral. Dis. 25, 1600-1607 (2019).
36. Kalogirou, E. M.et al.Decoding a gene expression program that accompanies the phenotype of sporadic and basal cell nevus syndrome-associated odontogenic keratocyst.J. Oral. Pathol. Med. 51, 649-658 (2022).
37. Verghese, G.et al.Computational pathology in cancer diagnosis, prognosis, and prediction—present day and prospects. J. Pathol., https://doi.org/10.1002/path.6163(2023).
38. Zhang, Q.et al.Deep learning to diagnose Hashimoto’s thyroiditis from sonographic images.Nat. Commun. 13, 3759(2022).
39. Boehm, K. M.et al.Multimodal data integration using machine learning improves risk stratification of high-grade serous ovarian cancer.Nat. Cancer 3, 723-733 (2022).
40. Topol E. J.High-performance medicine: the convergence of human and artificial intelligence.Nat. Med. 25, 44-56 (2019).
41. Yilmaz, E., Kayikcioglu, T.& Kayipmaz, S. Computer-aided diagnosis of periapical cyst and keratocystic odontogenic tumor on cone beam computed tomography.Comput. Methods Prog. Biomed. 146, 91-100 (2017).
42. Lee, J. H., Kim, D. H.& Jeong, S. N. Diagnosis of cystic lesions using panoramic and cone beam computed tomographic images based on deep learning neural network.Oral. Dis. 26, 152-158 (2020).
43. Florindo, J. B., Bruno, O. M.& Landini, G. Morphological classification of odontogenic keratocysts using Bouligand-Minkowski fractal descriptors.Comput. Biol. Med. 81, 1-10 (2017).
44. Chai, Z. K.et al.Improved diagnostic accuracy of ameloblastoma and odontogenic keratocyst on cone-beam CT by artificial intelligence.Front. Oncol. 11, 793417(2021).
45. Bispo, M. S.et al.Computer tomographic differential diagnosis of ameloblastoma and odontogenic keratocyst: classification using a convolutional neural network.Dento Maxillo Facial Radiol. 50, 20210002(2021).
46. Frydenlund, A., Eramian, M.& Daley, T. Automated classification of four types of developmental odontogenic cysts.Comput. Med. Imaging Graph. 38, 151-162 (2014).
47. Pan S., Dong Q., Sun L. S.& Li, T. J. Mechanisms of inactivation of PTCH1 gene in nevoid basal cell carcinoma syndrome: modification of the two-hit hypothesis.Clin. Cancer Res. 16, 442-450 (2010).