294

---

Click the image below to view the full size version of this cover.

Issue 294: In our recent work, we carried out a pilot project on personalized computational forecasting of prostate cancer growth during active surveillance using an imaging-informed biomathematical model. We describe the development of the disease as a combination of tumor cell mobility and proliferation. Longitudinal multiparametric magnetic resonance imaging data collected during standard-of-care active surveillance enable the construction of a virtual 3D representation of the patient’s prostate and provide estimates of tumor morphology and tumor cell density. Our results show that our technology can reproduce and predict spatiotemporal growth of prostate cancer during clinically relevant times in active surveillance. Additionally, we calculated a panel of metrics from the personalized tumor predictions and used them as features in a logistic classifier of clinical progression to higher risk disease, which requires moving the patient from active surveillance to radical treatment (e.g., surgery, radiotherapy). The global proliferation activity of the tumor was found to be a significant biomarker and together with the total tumor index, yielded a classifier with AUC=0.83, operating at 75% optimal sensitivity and specificity, and capable of predicting clinical progression by more than one year earlier than standard clinical practice. Thus, although further development and validation over larger datasets are required, we believe that our predictive technology is a promising approach to guide clinical decision-making and design personalized monitoring plans for each individual prostate cancer patient. See the full publication in Cancer Research Communications here.