Associations between protein biomarkers and adverse pathologic features in MRI-mapped prostate cancers: An interim analysis.

Abstract

405 Background: While standard mpMRI has shown benefit in increasing detection of clinically significant prostate cancer, it is limited in its ability to detect small lesions, low Gleason scores, and certain tumor architectures. Proteomics may supplement imaging and enhance prediction of clinical outcomes. We investigated associations between MRI-mapped tumors and protein levels obtained from radical prostatectomy (RP) specimens and correlated them to adverse pathology features. Methods: This was an interim analysis of a planned cohort of 124 subjects. Patients underwent routine preoperative mpMRI and whole-mount histopathology after RP, with proteomic analysis performed on tissue from MRI-mapped tumors. A panel previously shown to improve prediction of distant metastasis in RP patients included five proteins: TGFβ-1, SPARC, FOLH1 (PSMA), CAMKK2, and tumor PSA. Protein concentrations were log transformed. Logistic regression assessed protein associations with adverse pathologic features, including extracapsular extension (ECE), seminal vesicle invasion (SVI), lymph node invasion (LNI), Gleason score ≥8, and a composite of these findings. Multivariate analyses were adjusted by patient age, PSA, Gleason Grade Group, and T stage at diagnosis. Results: 61 patients had proteomic data; mean age at diagnosis was 64 years and mean PSA at diagnosis 10 ng/dL. ECE was present in 67%, SVI 20%, LNI 21%, and Gleason score ≥8 26%; 72% had at least one adverse pathologic feature. On initial multivariate analysis of proteins alone, TGFβ-1 significantly correlated with LNI (odds ratio [OR] 24.4 [95% confidence interval (CI) 3.7-160.2], p = 0.0009) and composite adverse pathology (OR 3.1 [95% CI 1.0-9.3], p = 0.05). Even after including clinical characteristics in the model, TGFβ-1 still significantly correlated with LNI (Table). None of the other proteins significantly correlated to these markers of adverse pathology. Conclusions: In this study assessing protein associations with MRI-identified prostate tumors, TGFβ-1 was strongly predictive for LNI. TGFβ signaling has been associated with both prostate cancer suppression and promotion depending on context. Further investigation will assess if the combination of proteins in the panel is more predictive of adverse pathology. Multivariate logistic regression analyzing association between proteins and lymph node invasion. Protein Odds ratio (95% CI) p-value TGFβ-1 235.0 (4.5-N/A) 0.007* SPARC 1.4 (0.3-7.3) 0.7 FOLH1 0.8 (0.3-2.4) 0.7 CAMKK2 2.0 (0.7-5.8) 0.2 PSA 2.0 (0.7-5.6) 0.2 Protein concentrations were log transformed and models included patient age, PSA, Gleason Grade Group, and clinical T-stage at diagnosis.