A Magnetic Bead-Based Sensor for the Quantification of Multiple Prostate Cancer Biomarkers

Abstract

Novel biomarker assays and upgraded analytical tools are urgently needed to accurately discriminate benign prostatic hypertrophy (BPH) from prostate cancer (CaP). To address this unmet clinical need, we report a piezeoelectric/magnetic bead-based assay to quantitate prostate specific antigen (PSA; free and total), prostatic acid phosphatase, carbonic anhydrase 1 (CA1), osteonectin, IL-6 soluble receptor (IL-6sr), and spondin-2. We used the sensor to measure these seven proteins in serum samples from 120 benign prostate hypertrophy patients and 100 Gleason score 6 and 7 CaP using serum samples previously collected and banked. The results were analyzed with receiver operator characteristic curve analysis. There were significant differences between BPH and CaP patients in the PSA, CA1, and spondin-2 assays. The highest AUC discrimination was achieved with a spondin-2 OR free/total PSA operation--the area under the curve was 0.84 with a p value below 10(-6). Some of these data seem to contradict previous reports and highlight the importance of sample selection and proper assay building in the development of biomarker measurement schemes. This bead-based system offers important advantages in assay building including low cost, high throughput, and rapid identification of an optimal matched antibody pair.

Fig 1. Detection Scheme.

A. The c.Ab. is covalently bound to a magnetic bead that is incubated with sample and fluorescein-tagged d.Ab. This sample is introduced into a flow cell capped with a piezo-electric membrane coated with an anti-fluorescein antibody. An electromagnet above the membrane is controlled with embedded software. B. Upon magnetic perturbation, all beads (black circles) move towards the membrane (Bi), and beads with a completed immunocomplex (black circles coated with red dots) bind to the anti-fluorescein antibody (Bii). After the field is removed and flow restored, only beads with a completed immunocomplex remain bound (Biii). These beads alter the oscillation of the membrane (represented as orange sinusoidal curve), which is interpreted as signal in arbitrary units [36]. See S1 Video. In B, the solid blue line labeled Pos refers to beads in the presence of antigen, and the dotted red line labeled Neg refers to beads in the absence of antigen.

Fig 2. Assay Construction.

A) Optimal antibody pairs were identified with a “checkerboard” assay that evaluated different antibody pairs as well as isotype controls. Metric plotted here is signal difference between the negative control and 1 ng/mL recombinant PAP. The different antibodies are defined in the Materials and Methods section. B) A full calibration curve illustrates the different sensitivities of different antibody pairs for PAP. Pair 1: Cos2 c.Ab.; Cos1 d.Ab. Pair 2: RDPo c.Ab.; RDMo. d.Ab. The red error bars represent the standard deviation of at least three replicate measurements. C) Reproducibility from day to day is <8%. D) The piezo-based approach shows 3 log orders improvement in analytical sensitivity versus direct ELISA when identical antibody pairs are used (Cos1/Cos2).

Fig 3. Serum Validation.

A) Percent recovery for three representative biomarkers. Matrix effects can either dampen signal (CA1, PAP) or inflate signal (SPARC). Ideal dilution factors were dependent on sensitivity of the assay and reference range of the biomarker (Table 1). Black dashed line indicates 100% spike recovery. B) Bland-Altman plot [38] validating specimen integrity of a subset (N = 35) of the clinical samples with high sample volumes. Red dashed line indicates 95% confidence interval.

Fig 4. Serum Data.

Raw values with means for BPH, CaP, and post-surgery (PS) samples. Biomarkers include tPSA (A), fPSA(B), CA1 (C), PAP (D), IL6-sr (E), SPARC (F), and SPON2 (G). Horizontal lines indicate mean values.

Fig 5. Clinical Data Analysis.

A) Mean values with standard deviation for the seven biomarkers with BPH, CaP, and post-surgery [39] samples. Units are ng/mL. The p value reported here is the significance between the BPH and CaP samples. AUC values are also given and report discrimination between CaP and BPH. Lower panel presents ROC curves for PSA, SPON2, and PSA OR SPON2. The OR operator increases the AUC to 0.84 from 0.80 for PSA alone.