ESTRO 2024 - Abstract Book

S5227

Radiobiology - Microenvironment

ESTRO 2024

Material/Methods:

The study included fresh-frozen biopsies from 95 prostate cancer patients. Hypoxia levels were quantified digitally from pimonidazole-stained sections in 83 patients, who received this hypoxia marker before prostatectomy. Gene and miRNA expression profiles were determined by Illumina bead arrays and RNA sequencing in 91 patients, respectively. Target genes were predicted from gene expression profiles and subjected to gene set enrichment analysis (GSEA) of cancer hallmarks. Expression levels of the oncogene MYC and tumor suppressor gene PTEN were validated by qRT-PCR. Cell proliferation by Ki67 and PTEN protein expression were quantified in histological sections by immunohistochemistry. Additionally, selected sections underwent miR-210 staining by in situ hybridization. A method for spatial analysis of multiple sections was developed, including co-registration of sections and representation of parameter fractions in microregional tiles. Spatial co-localization indices were computed for each biopsy by correlation analysis of parameter fractions within the tile representation. Aggressiveness was evaluated by biochemical recurrence (BCR). Results were validated in the external TCGA-PRAD cohort, consisting of 497 patients, where the Ragnum 32-gene hypoxia signature served as measure of hypoxia. The patient population displayed large heterogeneity in the distribution of hypoxia levels, ranging from normoxia to severe hypoxia. Correlation analysis showed that different sets of miRNAs were linked to different hypoxia levels. Of these, 7 miRNAs exhibited their strongest downregulation at moderate hypoxia, whereas 27 were downregulated and 8 upregulated at severe hypoxia. Scores based on miRNAs and their target genes were calculated and shown to be associated with the Ragnum hypoxia score in the TCGA-PRAD cohort (P<0.0001). MiRNA and target gene scores predicted BCR in both cohorts (P<0.05). GSEA on target genes identified enrichment of proliferation-related gene sets, including MYC targets at all hypoxia levels and PTEN inactivation at severe hypoxia (FDR<0.1). The association with these gene sets was validated by qRT-PCR of MYC and PTEN (P<0.05). Spatial analysis revealed significant co-localization with hypoxia for proliferating cells in 38 biopsies and cells with loss of PTEN in 11 biopsies. The co-localization index for hypoxia and proliferation was predictive of BCR (P<0.002) and correlated with miRNA scores (P<0.005). The co-localization index for hypoxia and PTEN was associated with a higher fraction of severe hypoxia (P=0.01). In a proof-of-concept investigation, miR-210-3p was detected in hypoxic regions where Ki67 and hypoxia overlapped. Results:

Conclusion:

We here demonstrated, in a cohort of patients, that the regulatory mechanisms governing gene expression in prostate tumors depend on hypoxia level. Different miRNAs seemed to be involved at moderate and severe hypoxia levels. At both levels, expression of these miRNAs and their target genes were associated with increased cell proliferation. Spatial histopathology confirmed cell proliferation, PTEN depletion, and miRNA expression within hypoxic regions of different severity levels. Furthermore, ISH demonstrated the presence of a regulatory miRNA within the regions displaying this phenotype. Additionally, we established a link between PTEN inactivation and miRNA regulation under severe hypoxia, with spatial analysis verifying the association between PTEN loss and severely hypoxic regions.

Keywords: Prostate cancer, miRNA regulation, hypoxia levels