The prognostic value of androgen to PSA ratio in predictive modeling of prostate cancer

Prostate specific antigen (PSA) is the ubiquitous biomarker used in mathematical and artificial intelligence models of prostate cancer. Yet, it is imperfect. The growth of prostate and cancer cells produces PSA, but their growth is dependent on androgen in most cases. Thus, looking at PSA by itself ignores the impact of androgen. Here, we propose the use of androgen as a reference framework for better integration of PSA in prostate cancer models. The binding of androgen to androgen receptors activates signals that promote prostate cancer growth and the production of PSA, so higher levels of androgen should correlate to higher growth of the tumor and higher production of PSA. Thus, PSA and androgen fall and rise almost synchronously during the start and stop of androgen deprivation therapy (ADT), respectively. Yet, when cancer cells become resistant to ADT, we observe a divergence to this trend, leading to a high level of PSA even during ADT. Therefore, we hypothesize that the cancer evolutionary dynamics can be represented by the ratio of androgen to PSA and its integration into prostate cancer models may help improve their predictive power.