Image analysis of androgen receptor immunostaining in prostate cancer accurately predicts response to hormonal therapy.

PURPOSE: Immunostaining for androgen receptor in prostate tumor specimens has revealed that the majority of primary and advanced stage cancers are positive for this regulatory transcription factor. Consequently, its use as a marker for tumor behavior and therapeutic response has been discounted. However, past reports have noted significant heterogeneity of androgen receptor immunostaining between prostate tumor cells in contrast to staining homogeneity in normal epithelium, which indicates that variability in androgen receptor content may exist within certain tumor specimens. To analyze this phenomenon more thoroughly and to determine whether this variability possesses clinical correlates, androgen receptor immunostaining profiles within androgen receptor positive prostate tumor specimens were categorized using an image analysis based system.
MATERIALS AND METHODS: Tumor specimens were obtained before hormone therapy from 44 patients with advanced stage prostate cancer and 4 with early stage disease who later had progression. Response to antiandrogen therapy and survival was monitored. Paraffin embedded tumor sections were processed for immunocytochemistry and stained for androgen receptor. A Quantimet image analysis system was used to analyze nuclear immunostaining for androgen receptor and Receptogram patterns were established for each specimen based on univariate distributions of nuclear receptor content and concentration.
RESULTS: Data revealed that 17 of 18 responders to hormone therapy possessed type 1 (15) or type 3 (2) Receptograms, which are characterized by a unimodal peak or multimodal peaks within a narrow concentration range. Of the 17 cases that stabilized following therapy 16 had type 3 Receptograms and 1 was characterized as type 1. In contrast, all 13 patients in whom endocrine treatment failed had either type 2 or 4 Receptograms, which are characterized by a highly skewed or bimodal androgen receptor distribution. Positive and negative predictive values for this assay were 100 and 93%, respectively. In addition, the type 1/3 Receptogram patterns were correlated with longer mean survival.
CONCLUSIONS: Image analysis of prostate cancer androgen receptor immunostaining with a pattern oriented approach for response is capable of accurately predicting response to hormone therapy in patients with advanced stage disease. Application of this analytic scheme may assist the clinician with therapeutic management of advanced prostate cancer.