Heng Yang,1,2,* Jiayue Yang,1,3,* Huilan Zheng,1,2 Yao Dai,4 Xiqian Chen,5 Jingping Wu,1,6 Xiao Ma,7 Hongbin Cheng1,2 

1School of Clinical Medicine, Chengdu University of Traditional Chinese Medicine, Chengdu, 610075, People’s Republic of China; 2Department of Dermatology, Hospital of Chengdu University of Traditional Chinese Medicine, Chengdu, 610072, People’s Republic of China; 3Department of Rheumatology, Hospital of Chengdu University of Traditional Chinese Medicine, Chengdu, 610072, People’s Republic of China; 4The Fourth Affiliated Hospital of Xinjiang Medical University, Xinjiang Medical University, Urumqi, 830054, People’s Republic of China; 5Department of Dermatology, the Affiliated Hospital of Southwest Medical University, Luzhou, 646000, People’s Republic of China; 6Department of Medical Cosmetology, Hospital of Chengdu University of Traditional Chinese Medicine, Chengdu, 610072, People’s Republic of China; 7School of Pharmacy, Chengdu University of Traditional Chinese Medicine, Chengdu, 611137, People’s Republic of China

*These authors contributed equally to this work

Correspondence: Xiao Ma, Email tobymaxiao@cdutcm.edu.cn Hongbin Cheng, Email chenghongbin@cdutcm.edu.cn

Objective: The aim of this study was to systematically elucidate the crosstalk mechanisms between metastatic melanoma and vitiligo and to establish vitiligo and metastasis-based biomarkers as well as to find drug candidates.
Methods: The genes associated with vitiligo and metastatic melanoma were obtained through differential expression analysis and WGCNA using publicly available data from GEO and TCGA. A prognostic model and nomogram for melanoma were subsequently constructed using hub genes based on machine learning algorithms. A comprehensive assessment was conducted of the correlation between hub genes and overall survival, functional annotations, immune cells and immune checkpoint genes. At the single-cell level, we conducted scoring using the AUCell algorithm and CellChat analysis to facilitate more profound biological exploration. The Cmap database and molecular docking methods were used to screen drug candidates.
Results: Following the screening process, a total of six hub genes (DUOX1, GJB3, NOTCH3, PKP1, PTK6 and PTPRF) were employed in the construction of prognostic model by machine learning. Patients were stratified into high-risk and low-risk groups based on the model. The expression of hub genes and the predictive ability of the model were validated in independent cohorts. The high-risk group exhibited worse prognosis, greater immunosuppression and tumor-associated macrophage infiltration. A nomogram based on the risk score had great performance in predicting survival of melanoma patients at 1-, 3-, and 5-year time points. The scRNA-seq results indicated that hub genes may exert an influence on tumor progression and metastasis by affecting fibroblasts and thus promoting epithelial-mesenchymal transition. Methyl-angolensate, byssochlamic-acid, homoharringtonine, piperacillin and cephaeline were potentially targeted therapeutic compounds for hub genes based on molecular docking.
Conclusion: Our study firstly provides new insight into the genetic crosstalk between metastatic melanoma and vitiligo that may facilitate the development of personalized treatments.

Keywords: metastatic melanoma, vitiligo, machine learning, scRNAseq, molecular docking