Pointed in the right direction.
Top Line: Treating prostate cancer well begins with appropriate risk assessment.
The Study: In 2016, with a rise in proposed risk models for all cancers, the AJCC decreed that moving forward it would adopt only models meeting prespecified criteria. Since then, no newly-proposed risk models for localized prostate cancer have surpassed their bar. Until now. This massive cohort study used data from nearly 20,000 men (11% Black) with localized prostate cancer from 55 international medical centers to create its model. Factors important to predicting prostate cancer mortality (PCM) in the training cohort (50% of subjects) included age, T and N stage, Gleason score, percent of positive cores, and PSA. Each of these factors was then assigned 0-8 points per its regression coefficient, with involved nodes and/or high Gleason scores buying the most points. Point totals then put patients into one of nine newly-classified group stages IA-IIIC. In the validation cohort (the other 50% of subjects), the new group staging effectively distinguished PCM. Nearly 40% of all men were categorized as stages IA-IB with a 10-year PCM of <1%--men who might safely opt for active surveillance. Stages IIC-IIIC, on the other hand, carried a 10-year PCM 10x that at 10%. The authors do recognize gene expression classifiers could very well add even more power to this predictive model, but such information wasn’t available on this cohort. Without leaning on gene expression, though, this model remains widely applicable in most international settings.
TBL: A new points-based group staging for localized prostate cancer outperforms current AJCC staging and NCCN risk categorization at predicting prostate cancer mortality. | Dess, JAMA Oncol 2020