PSA-detected prostate cancer and the potential for dedifferentiation--estimating the proportion capable of...

The aims were to determine whether prostate-specific antigen (PSA)-detected prostate cancers progress to higher Gleason score during the preclinical screen-detectable phase, and, if so, to estimate the proportion of tumours with progressive potential. We developed two multi-state Markov chain models to represent the natural history of two tumour populations, one with (Model I) and the other without (Model II) the potential for progression. For each, we derived the transition rates between the states and used these estimates to calculate the expected prevalence of preclinical low and intermediate-to-high Gleason score prostate cancers, using data from the Prostate Testing for Cancer and Treatment (ProtecT) study on 2,310 prostate cancers detected by PSA testing in 71,511 men 50-69 years. We compared the expected prevalence for each tumour population to that of the observed based on ProtecT and the European Randomised Study on Screening for Prostate Cancer (ERSPC)-Rotterdam Centre's first round screening data, the latter allowing independent assessment of the two models. The overall expected proportion of low Gleason score tumours was 0.56 under Model I and 0.81 under Model II, whereas the observed proportion based on either ProtecT or ERSPC-Rotterdam was 0.69. Using the observed prevalence from ERSPC-Rotterdam, we estimated that 22, 33 and 66% of the tumours in men aged 55-59, 60-64 and 65-69 years, respectively, had the potential for progression in the preclinical phase. PSA-detected prostate cancers are a mixture of progressive and non-progressive tumours with respect to Gleason score. The former may potentially benefit from screening. Identifying cancers with the potential for progression is important to target screening.