PLoS Medicine (Jul 2019)

Validation of the prognostic value of NF-κB p65 in prostate cancer: A retrospective study using a large multi-institutional cohort of the Canadian Prostate Cancer Biomarker Network.

  • Andrée-Anne Grosset,
  • Véronique Ouellet,
  • Christine Caron,
  • Gabriela Fragoso,
  • Véronique Barrès,
  • Nathalie Delvoye,
  • Mathieu Latour,
  • Armen Aprikian,
  • Alain Bergeron,
  • Simone Chevalier,
  • Ladan Fazli,
  • Neil Fleshner,
  • Martin Gleave,
  • Pierre Karakiewicz,
  • Louis Lacombe,
  • Jean-Baptiste Lattouf,
  • Theodorus van der Kwast,
  • Dominique Trudel,
  • Anne-Marie Mes-Masson,
  • Fred Saad,
  • Canadian Prostate Cancer Biomarker Network

DOI
https://doi.org/10.1371/journal.pmed.1002847
Journal volume & issue
Vol. 16, no. 7
p. e1002847

Abstract

Read online

BACKGROUND:The identification of patients with high-risk prostate cancer (PC) is a major challenge for clinicians, and the improvement of current prognostic parameters is an unmet clinical need. We and others have identified an association between the nuclear localization of NF-κB p65 and biochemical recurrence (BCR) in PC in small and/or single-centre cohorts of patients. METHODS AND FINDINGS:In this study, we accessed 2 different multi-centre tissue microarrays (TMAs) representing cohorts of patients (Test-TMA and Validation-TMA series) of the Canadian Prostate Cancer Biomarker Network (CPCBN) to validate the association between p65 nuclear frequency and PC outcomes. Immunohistochemical staining of p65 was performed on the Test-TMA and Validation-TMA series, which include PC tissues from patients treated by first-line radical prostatectomy (n = 250 and n = 1,262, respectively). Two independent observers evaluated the p65 nuclear frequency in digital images of cancer tissue and benign adjacent gland tissue. Kaplan-Meier curves coupled with a log-rank test and univariate and multivariate Cox regression models were used for statistical analyses of continuous values and dichotomized data (cutoff of 3%). Multivariate analysis of the Validation-TMA cohort showed that p65 nuclear frequency in cancer cells was an independent predictor of BCR using continuous (hazard ratio [HR] 1.02 [95% CI 1.00-1.03], p = 0.004) and dichotomized data (HR 1.33 [95% CI 1.09-1.62], p = 0.005). Using a cutoff of 3%, we found that this biomarker was also associated with the development of bone metastases (HR 1.82 [95% CI 1.05-3.16], p = 0.033) and PC-specific mortality (HR 2.63 [95% CI 1.30-5.31], p = 0.004), independent of clinical parameters. BCR-free survival, bone-metastasis-free survival, and PC-specific survival were shorter for patients with higher p65 nuclear frequency (p < 0.005). As the small cores on TMAs are a limitation of the study, a backward validation of whole PC tissue section will be necessary for the implementation of p65 nuclear frequency as a PC biomarker in the clinical workflow. CONCLUSIONS:We report the first study using the pan-Canadian multi-centre cohorts of CPCBN and validate the association between increased frequency of nuclear p65 frequency and a risk of disease progression.