Construction of noninvasive prognostic model of bladder cancer patients based on urine proteomics and screening of natural compounds.

BACKGROUND

Bladder cancer (BCa) has a high incidence and recurrence rate worldwide. So far, there is no noninvasive detection of BCa therapy and prognosis based on urine multi-omics. Therefore, it is necessary to explore noninvasive predictive models and novel treatment modalities for BCa.

METHODS

First, we performed protein analysis of urine from five BCa patients and five healthy individuals using liquid chromatography-tandem mass spectrometry (LC-MS/MS). Combining multi-omics data to mine particular and sensitive molecules to predict BCa prognosis. Second, urine proteomics data were combined with TCGA transcriptome data to select differential genes that were specifically highly expressed in urine and tissues. Further, the Lasso equation was used to screen specific molecules to construct a noninvasive prediction model of BCa. Finally, natural compounds of specific molecules were selected by combined network pharmacology and molecular docking to complete molecular structure docking.

RESULTS

A noninvasive predictive model was constructed using PSMB5, P4HB, S100A16, GET3, CNP, TFRC, DCXR, and MPZL1, specific molecules screened by multi-omics, and clinical features, which had good predictive value at 1, 3, and 5 years of prediction. High expression of these target genes suggests a poor prognosis in patients with BCa, and they were mainly involved in cell adhesion molecules and the IGF pathway. In addition, the corresponding drugs and natural compounds were selected by network pharmacology, and the molecular structure 7NHT of PSMB5 was found to be well docked to Ellagic acid, a natural compound in Hetaoren that we found. The 3D structure 6I7S of P4HB was able to bind to Stigmasterol in Shanzha stably, and the structure 6WRV of TFRC as an iron transport carrier was also able to bind to Stigmasterol in Shanzha stably. The structures 1WOJ, 3D3W, and 6IGW of CNP, DCXR, and MPZL1 can also play an important role in combination with the natural compounds (S)-Stylopine, Kryptoxanthin, and Sitosterol in Maqianzi, Yumixu, and Laoguancao.

CONCLUSION

The noninvasive prediction model based on urinomics had excellent potential in predicting the prognosis of patients with BCa. The multi-omics screening of specific molecules combined with pharmacology and compound molecular docking can promote the research and development of novel drugs.