In situ DESI-MSI Lipidomic Profiles of Mucosal Margin of Oral Squamous Cell Carcinoma

Purpose: The optimal safe margin distance of oral squamous cell carcinoma (OSCC) is still controversial. In this article, we apply a cost-effective molecular diagnosis method at the lipidomic level for determining the safe surgical-resection distance of OSCC by desorption electrospray ionization mass spectrometry imaging (DESI-MSI). Methods: A total of 22 participants (each with a matching tumor and continuous mucosal margin tissues) were recruited to the study. All specimens were cut into 10-μm frozen cryosections and scanned by DESI-MSI in the negative ion mode. By overlaying the mass spectrometry image with the H&E staining image, we first precisely extracted the mass spectra of those pixels across the diagnosed tumor and normal regions which served as the training and validation datasets. A Lasso regression model was developed and externally validated for distinguishing the tumor and normal region. A set of feature ions which gradually decreased from tumor to normal tissue was highlighted by assigning them higher weight coefficients during the Lasso model development. The safe surgical-resection distance of OSCC on frozen sections was measured by using this developed molecular diagnostic model for clinical reference. Meanwhile, the genetic marker (p53) of the surgical-margin was used to verify the safe surgical-resection distance determined by DESI-MSI. Results: By leave-one-out validation, the Lasso model can achieve 88.6% accuracy. The Lasso model selected 179 features from an initially collected 22,955 ions across 7,243 pixels for distinguishing tumor and normal tissues. We further screened these 179 characteristic ions, which tend to decrease gradually along with the increase in distance from tumor to normal tissue. Finally, 14 decreasing characteristic ions were tentatively assigned as fatty acids and selected as potential markers for determining the safe surgical-resection distance. These 14 lipid ions gradually decrease from tumor to normal tissue and can be used to accurately distinguish tumors from normal tissues. The spatial segmentation results based on our diagnostic model can not only clearly delineate the tumor and normal tissue but also distinguish the different status of surgical margins. Meanwhile, the safe surgical-resection distance of OSCC on frozen sections can also be accurately measured by using the developed diagnostic model. Moreover, the genetic marker of the OSCC margin (p53) validated that the safe surgical-margin boundary determined by the DESI-MSI and p53 is highly consistent. Conclusions: A diagnostic model was successfully constructed by combining DESI-MSI with the Lasso regression, which could accurately measure the safe margin distance of OSCC on frozen sections. It provides a theoretical basis for a personalized evaluation of safe margin distance of OSCC. Funding Statement: This study was supported by Jiangsu Provincial Key Medical Discipline (since 2017), Nanjing Municipal Key Medical Laboratory Constructional Project Funding (Since 2016), National Natural Science Foundation of China (No.81772880, 81702680). Declaration of Interests: The authors have no competing interests. Ethics Approval Statement: This study was approved by the medical ethics committee of the Nanjing Stomatology Hospital, Medical School of Nanjing University. All participants signed the informed consent forms.