Identification and validation of novel biomarkers affecting bladder cancer immunotherapy via machine learning and its association with M2 macrophages

Abstract

Background: Immunotherapy has shown promising results in bladder cancer therapy options.

Methods: Analysis of open-access data was conducted using the R software. Open-access data were obtained from The Cancer Genome Atlas (TCGA), Gene Expression Omnibus (GEO), and IMvigor210 databases. Immunofluorescence and co-culture systems were utilized to validate the effect of PTHLH on M2 macrophage polarization.

Results: Here, through the combined (TCGA, GSE128959, GSE13507, and GSE83586) and IMvigor210 cohorts, we comprehensively investigated the biological and immune microenvironment differences in patients with diverse immunotherapy responses. Meanwhile, we found that M2 macrophage could affect bladder cancer immunotherapy sensibility. Moreover, based on the machine learning algorithm (LASSO logistics regression), PTHLH, BHMT2, and NGFR were identified, which all have good prediction abilities for patient immunotherapy. Then, a logistics regression model was established based on PTHLH, BHMT2, and NGFR, and each patient was assigned a logistics score. Subsequently, we investigated the difference in patients with high low logistics scores, including biological enrichment, immune microenvironment, and genomic characteristics. Meanwhile, data from the Human Protein Atlas database indicated a higher protein level of PTHLH in bladder cancer tissue. Immunofluorescence indicated that the knockdown of PTHLH in bladder cancer cells can significantly inhibit the M2 polarization of co-culture M0 macrophages.

Conclusions: Our study investigated the difference between bladder cancer immunotherapy responders and non-responders. Meanwhile, the PTHLH was identified as a novel biomarker for bladder cancer immunotherapy.

Keywords: M2 macrophages; PTHLH; bladder cancer; immunotherapy; machine learning.