Validating clinical assumptions for characterizing prostate cancer aggressiveness utilizing historical cancer registry data

The prognosis of prostate cancer is highly dependent on disease characteristics at diagnosis, and dilution effects can compromise etiological studies that are not stratified by disease aggressiveness. However, such classifications are often hindered by missing staging data. This study aimed to develop and validate clinical assumptions to consolidate information about clinical TNM stage and Gleason score nnGS) for improved classification of aggressive and non-aggressive disease. 17,551 males diagnosed with prostate cancer (1974-2022) in the Norwegian Janus Serum Bank cohort were included. The developed clinical assumptions were validated using positive predictive values (PPVs) among males with non-missing data, and long-term cancer-specific mortality. Cases were classified as aggressive (cT4, N1, M1, or GS ≥8) or non-aggressive (complete information without aggressive features) with and without applying the clinical assumptions. Eight of the 14 developed clinical assumptions had a PPV >95%. Applying these assumptions nearly halved the proportion of cases with unclassifiable aggressivity (from 59% to 32%). Fifteen-year cumulative mortality [95%CI] was higher for previously unclassifiable cases newly as classified aggressive (53.4% [47.6-59.7]) than for cases originally classified as aggressive (42.6% [40.5-44.8]). Clinical assumptions to consolidate TNM and GS information can substantially improve the classification of aggressive and non-aggressive prostate cancer.