Salivary metabolomics for skin cancer screening

Abstract

Background

If skin cancer (SC) can be screened using saliva, it would be extremely important for improving the prognosis of skin cancer. However, there is currently no available data on salivary screening for SC.

Objective

This study aimed to identify salivary metabolomic biomarkers for SC screening by comparing tissue and saliva samples.

Methods

This single-center study involved SC patients from Yamagata University Hospital, enrolled between May 2017 and April 2020. For the healthy controls (HCs), our database comprising salivary metabolomics data of HCs was used. Whole unstimulated saliva samples were collected from patients with SC (n = 75) and HCs (n = 77). Paired tumor and control tissues after SC resection were obtained from the same patients who donated saliva. Hydrophilic metabolites in the tissue and saliva samples were comprehensively analyzed using capillary electrophoresis–mass spectrometry. Using these candidate metabolites, a multiple logistic regression (MLR) model was developed to differentiate SC from HCs

Results

Sixty-six and 12 metabolites showed significant differences in tissues and saliva, respectively. Of these, six metabolites were commonly different between SC and HCs. Spermidine, 2-aminobutyric acid, and isoleucine were selected to develop the MLR model. The model exhibited a large area under the receiver operating characteristic curve (0.802, 95 % confidence interval: 0.731–0.874, P < 0.0001). Additionally, no significant differences were shown in candidate salivary metabolites between stages in both malignant melanoma and squamous cell carcinoma.

Conclusions

The salivary metabolomic profiles between SC and HCs differed clearly. This combination of salivary metabolites could serve as non-invasive biomarkers for screening SC.