Prostate cancer molecular subtypes in systematic versus MRI-targeted biopsy cores at active surveillance: association of PTEN and ERG status with extreme grade reclassification.

[AIMS] Biomarkers are needed to identify patients with favourable-risk prostate cancers who could benefit from treatment versus active surveillance (AS). MRI-targeted biopsies may allow for better sampling of the 'dominant' tumour and therefore more accurate risk stratification. We evaluated ERG fusions and PTEN loss for predicting subsequent grade reclassification (GR) in a multiparametric MRI-screened cohort, comparing results in systematic versus targeted biopsies.

[METHODS AND RESULTS] Validated immunohistochemistry assays for ERG and PTEN were performed on all cancer-containing cores of early AS biopsies, in a cohort of 127 Grade Group (GG) 1 AS patients with targeted and systematic biopsies. ERG fusions were identified in 43% of the cohort- 24% of these cases exhibited heterogeneous (subclonal or multi-clonal) ERG expression in sampled tumour. PTEN loss was identified in 16% of the cohort and all cases were heterogeneous. Neither ERG fusions nor PTEN loss were more likely to be found in targeted versus systematic cores. On multivariable Cox proportional hazard regression, PTEN loss was associated with extreme GR on subsequent biopsy (to GG ≥3) (hazard ratio [HR] 9.27, 95% confidence interval [CI] 2.10-40.9, P = 0.003) but ERG fusion was not (HR 2.82, CI 0.73-10.9, P = 0.13). PTEN loss was most strongly associated with this outcome when measured in systematic biopsy cores.

[CONCLUSIONS] Frequent ERG heterogeneity suggests prevalent multiclonal disease in MRI-visible low-risk prostate cancers. Higher risk AS patients might benefit from PTEN loss evaluation to predict extreme GR, in both targeted and systematic biopsies, since targeted biopsies were not more likely to identify PTEN loss.