Prostate Cancer Case Study
Identification of a Multivariate Model for Prediction of Response to Androgen Deprivation Therapy
Challenge: To determine if Androgen Receptor (AR) levels from the primary prostate tumor tissue may be predictive of therapeutic response to androgen deprivation therapy (ADT) in terms of elapsed time from initial treatment with ADT after prostatectomy, to castrate rise in PSA levels. By utilizing quantitative immunofluorescence (IF) and spectral imaging we evaluated AR content as an independent predictor of therapeutic outcome.
Methods: 63 of 881 patients treated by radical prostatectomy had received ADT for a biochemical rise in PSA and or clinical progression of their disease. 32 of 63 patients had progressed post-ADT with a castrate rise in PSA. Tissue microarrays with triplicate patient cores were stained with a multiplex immunofluorescent assay (IF) which contained the nuclear marker DAPI, along with the following antibodies: Androgen Receptor (AR), Racemase (AMACR), CK18, HMWK and p63. IF images were acquired with spectral un-mixing employed to develop antibody-cell-specific gray scale images. Utilizing image analysis algorithms, quantitative features including intensity and area for selected antigens were generated.
Results: Eleven IF antigen features were evaluated after feature selection and filtering using the concordance index with respect to outcome. A total of 5 features were statistically significant for predicting time to progression post therapy of which 2 features (AR intensity within AMACR (+) and (-) epithelial cells; p=0.0003 and p=0.0021, respectively) demonstrated that elevated levels of AR were associated with a shortened time to castrate rise in PSA post ADT. Clinical features alone were not found to be statistically significant with respect to predicting outcome.
Conclusions: AR levels in the prostatectomy sample appears to be a useful indicator for determining
therapeutic ADT response and potentially guiding future treatment decision and patient monitoring.
Next steps: A larger cohort is needed to conduct a properly powered study, which then needs further validation.
|
Patient Demographics |
63 Patients |
| Clinical Stage |
T1c(17), T2a-c(38) |
| Avaerage PSA |
14.34ng/ml |
| Dom Biopsy Grade |
GG 3(37), GG 4(25) |
| Biopsy Gleason Score |
GS 6(23), 7(21) 8>(16) |
| Dom Prostatectomy Grade |
3(32), 4(29) |
| Prostatectomy Gleason Score |
6(19), 7(24), 8>(16) |
| SVI |
0(41), 1(22) |
| LN |
0(52), 1(11) |
| Margins |
0(27), 1(34) |
| ECE |
0(33), 2(29) |
Table 1. Patient demographics for study
|
Clinical Features |
CI |
Chi Square |
p-value |
|
Clinical Stage |
0.50 |
0.00 |
0.9689 |
|
Pre-Op PSA |
0.48 |
0.26 |
0.6107 |
|
Dominant Biopsy Gleason Grade |
0.49 |
0.05 |
0.8223 |
|
Biopsy Gleason Sum |
0.46 |
0.54 |
0.4608 |
|
Dominant Prostatectomy Gleason Grade |
0.43 |
0.87 |
0.3499 |
|
Prostatectomy Gleason Sum |
0.49 |
0.04 |
0.8462 |
|
Seminal Vesicle Invasion Status |
0.44 |
1.67 |
0.1956 |
|
Lymph Node Status |
0.53 |
0.32 |
0.5731 |
|
Surgical Margin Status |
0.51 |
0.21 |
0.6476 |
|
Extra Capsular Extension Status |
0.51 |
0.23 |
0.6335 |
|
Aureon Multivariate Model for ADT Response |
0.68 |
11.00 |
0.0009 |
Table 2. Comparison of univariate analysis of various clinical-pathological attributes used to risk stratify patients for duration of ADT response and a multivariate model developed by Aureon Laboratories utilizing morphometric image analysis of Androgen Receptor. The CI and p-values generated using Aureon Systems pathology show promise in being able to more accurately risk stratify these patients in terms of ADT reponse than clinical features alone.
 |
 |
Figure 1. Aureon image capture of prostate gland: Original multiplex IF assay is on left Aureon segmented and classified image is on right. Patients who have a predominance of Androgen Receptor (AR) expressed in epithelial nuclei are at higher risk for biochemical recurrence and/or clinical failure. Aureon’s technology utilizes object-based quantitation of AR (or any other biomarker of interest) in gland units, vessels, cytoplasm and/or the nuclei of a variety of cell types (e.g., epithelial, stroma, proliferative). Epithelial nuclei is in green, AR+ epithelial nuclei is in light pink, AR+ stromal nuclei is dark pink.