Title: High-risk multi-region prostate cancer proteomics

Student author: Clara C. Cousins1, 2

Area of concentration: Basic, Clinical, and Translational Science Petal

Specialty: Pathology

Coauthors: Julie Livingstone2, 3, 4, Lydia Y Liu2, 3, 4, 5, Amanda Khoo2, 3, 4, 5, Vladimir Ignatchenko2, 3, 4, 5, Michelle Downes6, Thomas Kislinger*2, 3, 4, 5, Paul C. Boutros*2, 3, 4, 5, 7

1David Geffen School of Medicine, University of California, Los Angeles, USA

2Jonsson Comprehensive Cancer Center, University of California, Los Angeles, USA

3Departments of Human Genetics, University of California, Los Angeles, USA

4Institute for Precision Health, University of California, Los Angeles, USA

5Department of Medical Biophysics, University of Toronto, Toronto, Canada

6Department of Laboratory Medicine and Pathobiology, University of Toronto, Canada

7Department of Urology, University of California, Los Angeles, USA

Keywords: Proteomics; Prostate; Cancer

Background: Despite its high incidence, prostate cancer demonstrates a wide range of clinical outcomes. Predictors of aggressive prostate tumors are therefore critical for guiding treatment. Current risk stratification relies on features such as Gleason grade to characterize differentiation, tumor (T) category to stage spread, and pre-operative prostate specific antigen (PSA) levels to quantify tissue activity. Identifying molecular drivers of high-risk prostate cancer may ultimately help to clarify tumor evolution and inform targeted therapies.

Objective: To assess proteomic heterogeneity in prostate cancer using spatially aware mass spectrometry

Methods: We quantified the abundance of 3,192 proteins in 241 macro-dissected regions from 53 prostatectomy-treated patients having primary tumors of Gleason pattern 4 or 5. The association of protein abundance with clinically relevant features including Gleason grade, T category, and PSA was performed using linear mixed-effects (LME) models accounting for inter-patient and technical batch variability. Likelihood ratio tests were used to evaluate the significance of the fit for the model with versus without each clinically relevant feature. T-tests were used to evaluate whether model coefficients were significant.

Results: In LME models, protein abundance was most significantly associated with Gleason grade, T category, and PSA for: PRPF4 (FDR-adjusted p = 0.048), GGT5 (FDR-adjusted p = 0.0055), and PLTP (FDR-adjusted p = 0.0012), respectively. The corresponding coefficient estimates were 0.97 (linear term; p = 0.0051), -2.11 (quadratic term; p = 3.81e-06), and -0.96 (quadratic term; p = 5.36e-07) for PRPF4, GGT5, and PLTP, respectively.

Conclusions: Proteomic heterogeneity may be associated with high-risk clinical features and describe aggressive tumor evolution. PRPF4 (pre-mRNA processing factor 4) may increase across Gleason grades, while GGT5 (gamma glutamyl-transferase 5) and PLTP (phospholipid transfer protein) may each peak then ultimately decrease across T categories and PSA levels, respectively. Further work is needed to clarify the pathophysiology.