Autoimmunity to selenoprotein P predicts breast cancer recurrence

Abstract

Background: Low concentrations of serum selenium (Se) and its main transporter selenoprotein P (SELENOP) are associated with a poor prognosis following breast cancer diagnosis. Recently, natural autoantibodies (aAb) with antagonistic properties to SELENOP uptake have been identified in healthy subjects, and in patients with thyroid disease. Given the potential transport disrupting properties, we hypothesized that breast cancer patients with SELENOP-aAb may have a poor prognosis.

Methods: SELENOP-aAb along with serum Se, SELENOP and GPX3 activity were determined in serum samples of 1988 patients with a new diagnosis of breast cancer enrolled in the multicentre SCAN-B study. Patients were followed for ∼9 years and multivariate Cox regression models were applied to assess hazard ratios.

Results: Applying a cut-off based on outlier detection, we identified 7.65% of patients with SELENOP-aAb. Autoantibody titres correlated positively to total Se and SELENOP concentrations, but not to GPX3 activity, supporting a negative role of SELENOP-aAb on Se transport. SELENOP-aAb were associated with age, but independent of tumor characteristics. After fully adjusting for potential confounders, SELENOP-aAb were associated with higher recurrence, HR(95%CI) = 1.87(1.17-2.99), particularly in patients with low Se concentrations, HR(95%CI) = 2.16(1.20-3.88). Associations of SELENOP-aAb with recurrence and mortality were linear and dose-dependent, with fully adjusted HR(95%CI) per log increase of 1.25(1.01-1.55) and 1.31(1.13-1.51), respectively.

Conclusion: Our results indicate a prognostic and pathophysiological relevance of SELENOP-aAb in breast cancer, with potential relevance for other malignancies. Assessment of SELENOP-aAb at time of diagnosis identifies patients with a distinctly elevated risk for a poor prognosis, independent of established prognostic factors, who may respond favourably to Se supplementation.

Keywords: Cohort study; Glutathione peroxidase; Prognosis; SELENOP; Selenium.

Graphical abstract

Fig. 1

Study design and prevalence of autoimmunity to SELENOP. a 1988 patients with an incident diagnosis of primary invasive breast cancer were included in this study. Serum sampling was conducted at time of diagnosis, and follow up encompassed approximately 9 years b Samples were analysed for SELENOP-aAb in 96 well plates by immunoprecipitation of complexes formed in serum with protein A-sepharose, and detection of luminescence as light units (RLU) from precipitated SELENOP-SEAP-aAb complexes. c An outlier criterion for cut-off definition of autoimmunity was applied, and values exceeding 3-fold of binding index (BI ≥ 3, dotted line) were considered positive. d Binding indices of SELENOP-aAb are displayed on a logarithmized y-axis, and plotted as density on the right y-axis. Patients above the cut-off are marked red e SELENOP-aAb displayed a right skew, as emphasized by the marginal density plot. BI was displayed on non-logarithmized y-axis. f Applying the unbiased cut-off (BI ≥ 3), a total of 7.65% of patients were identified as SELENOP-aAb positive. g Age at diagnosis was compared to aAb-positivity, applying the Wilcoxon-Rank-sum test. h Correlation of the continuous SELENOP-aAb titre and age at diagnosis was assessed, using Spearman's rank correlation test. Blue points indicate SELENOP-aAb negative patients, and red points indicate SELENOP-aAb positive patients. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Fig. 2

Correlation of SELENOP-aAb with total serum selenium and selenoproteins. Linear regression (line) with 95% confidence intervals (shadow) was used to visualize the relationship. a Correlation of autoantibody titres to total serum selenium was assessed, with increasing cut-offs for autoantibody titres from left to right. Slope of the linear regression line has shown an increasing trend with increasing antibody titres. Above BI = 10, SELENOP-aAb were significantly correlated with total serum selenium, R = 0.336, p = 0.009. b A similar trend was seen with regard to serum SELENOP levels, which also was statistically significant above BI = 10, R = 0.273, p = 0.037. c No association was seen for serum GPX3, although it is tightly correlated to serum selenium and serum SELENOP concentrations in this study cohort. d SELENOP-aAb were significantly associated with selenium/GPX3 ratio above BI = 10, R = 0.299, p = 0.021. Spearman's rank correlation was used to assess correlation.

Fig. 3

Kaplan Meier plots for overall survival and recurrence free interval. a Overall survival according to autoantibody positivity was assessed with Kaplan Meier plots and log-rank test. Overall survival differed significantly between the two groups. b Recurrence free interval was also lower in SELENOP-aAb positive patients. c Overall survival probability stratified by Se status, cut-off was set at median of the cohort, corresponding to 70.4 μg/L Se. d Recurrence free interval stratified by Se status.