AI-empowered cellular morphometric risk score improves prognostic stratification of cutaneous squamous cell carcinoma

Manuel J Pérez-Baena; Jian-Hua Mao; Jesús Pérez-Losada; Ángel Santos-Briz; Hang Chang; Javier Cañueto

doi:10.1093/ced/llad264

AI-empowered cellular morphometric risk score improves prognostic stratification of cutaneous squamous cell carcinoma

Clin Exp Dermatol. 2023 Aug 12:llad264. doi: 10.1093/ced/llad264. Online ahead of print.

Authors

Manuel J Pérez-Baena^{1

2}, Jian-Hua Mao^{3

4}, Jesús Pérez-Losada^{1

2}, Ángel Santos-Briz^{2

5}, Hang Chang^{3

4}, Javier Cañueto^{1

2

6}

Affiliations

¹ Instituto de Biología Molecular y Celular del Cáncer (IBMCC-CIC), Laboratory 20, Universidad de Salamanca/CSIC, Campus Miguel de Unamuno s/n, 37007 Salamanca, Spain.
² Instituto de Investigación Biomédica de Salamanca (IBSAL), Hospital Universitario de Salamanca, Paseo de San Vicente 58-182, 37007 Salamanca, Spain.
³ Berkeley Biomedical Data Science Center, Lawrence Berkeley National Laboratory, Berkeley, CA, USA.
⁴ Biological Systems and Engineering Division, Lawrence Berkeley National Laboratory, Berkeley, CA, USA.
⁵ Servicio de Anatomía Patológica, Hospital Universitario de Salamanca, Paseo de San Vicente 58-182, 37007 Salamanca, Spain.
⁶ Departamento de Dermatología, Hospital Universitario de Salamanca, Paseo de San Vicente 58-182, 37007 Salamanca, Spain.

PMID: 37572359
DOI: 10.1093/ced/llad264

Abstract

Introduction: Risk stratification of cutaneous squamous cell carcinoma (CSCC) is essential for managing patients. Artificial intelligence and machine learning might help stratify patients with CSCC by risk using more than solely clinical and histopathological factors.

Methods: A retrospective cohort of 104 CSCCs excised with clear margins was retrieved. Clinical and histopathological risk factors were evaluated. Hematoxylin and eosin-stained slides were scanned and analyzed by an algorithm based on the stacked predictive sparse decomposition technique. Cellular morphometric biomarkers (CMBs) were identified via machine learning and used to derive a cellular morphometric risk score (CMRS) that classified CSCC into clusters of differential prognosis. Concordance analysis, sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy were calculated and compared with results obtained with the Brigham and Women's Hospital (BWH) staging system. The performance of the combination of the BWH staging system and the CMBs was also analyzed.

Results: There were no differences among CMRS groups in terms of clinical and histopathological risk factors and T-stage assignment, but there were significant differences in prognosis. Combining the CMRS with BWH staging systems increased distinctiveness and improved prognostic performance. C-indices were 0.92 for local recurrence and 0.91 for nodal metastasis when combining the two approaches. The NPV was 94.41% and 96.00%, the PPV was 36.36% and 41.67%, and accuracy reached 86.75% and 89.16% with the combined approach.

Conclusion: CMRS is helpful for CSCC risk stratification beyond classic clinical and histopathological risk features. Combining the information from the CMRS and the BWH staging system offers outstanding prognostic performance for high-risk CSCC patients.