Session Item

Radiogenomics is a field of research that relates genomics with medical imagining radiomics features. PTEN loss is associated with playing a role in affecting the immune response of the tumour microenvironment in prostate cancer. Fibroblast infiltration has also been associated with prostate cancer disease progression and treatment resistance. However, the relationship between PTEN loss and fibroblast infiltration has been poorly researched in prostate cancer. Additionally, radiogenomics studies have yet to investigate radiomics features related to fibroblasts infiltration. Our study sought to investigate the relationship between PTEN status and fibroblast infiltration and identify their correlated CT-based radiomics features in prostate cancer patients.

Using gene expression for 248 prostate cancer patients, a PTEN loss gene signature inferred PTEN status, and immune and stromal cell type scores were calculated. The interaction between PTEN loss and immune and stromal cell types was associated with metastatic disease using survival analysis. PTEN status and fibroblast infiltration levels were combined to generate patient subtypes, which were associated with metastatic disease and validated on an external prostate cancer patient cohort. 184 patients in our discovery cohort had matched CT-based radiomics features extracted. Correlation analysis was performed to identify radiomics features related to immune and stromal cells, PTEN loss, and our patient subtypes.

Fibroblast infiltration was significantly associated with metastatic disease in PTEN loss patients (HR = 2.43 [1.58-3.73], p-value < 0.001), but not in PTEN intact patients (HR = 0.97 [0.58-1.6], p-value = 0.89, interaction p-value = 0.006). A PTEN loss/fibroblast infiltration patient subtype (PTEN loss/fibroblast high) was significantly associated with metastatic disease (HR = 4.1 [1.75-9.61]) in comparison to the other generated subtypes (p-value < 0.001, Figure 1).

The PTEN loss/fibroblast high patient subtype was significantly associated with metastatic disease (HR = 3.23 [1.3-8.04]) when compared to the other subtypes (p-value = 0.006) in the external validation cohort. Various CT-based radiomics features were identified for fibroblast infiltration (Figure 2), PTEN loss, and the PTEN loss/fibroblast high subtype.

To our knowledge, this is the first study to demonstrate the relationship between PTEN loss and fibroblast infiltration for disease progression in prostate cancer and identify their significantly correlated radiomics features. Biological signaling pathways identified for the novel PTEN loss with fibroblast infiltration patient subtype should inspire future work to determine targeted treatment strategies for these patients. The CT-based radiomics features identified should also inspire future work to determine their ability to identify patients at risk of future disease progression during their treatment planning in clinical practice.