Adrian
The 15% figure for recurrence among men in the sample sounds too good to be true, but is probably because the median follow-up was only 5 years. To get a more realistic view of the recurrence rate when you have a study with a wide range of follow-up periods, it is better to use an actuarial calculation (sometimes described as something like "Kaplan-Meier") rather than looking at the observed incidence in the sample.
I'm not sure that the long term actuarial rate of recurrence is quoted in the paper because it is mainly concerned with the likelihood of metastasis after recurrence. However, it does reveal that the overall rate of metastasis at 15 years for the whole sample was 18% so the recurrence rate will presumably have been much higher than that.
In general, of course, the recurrence rate observed in any study will depend not just on the follow-up period but also on the characteristics of the men in the sample - e.g. a higher percentage of grade group 1 will generally mean a lower recurrence rate.
Bets wishes
Kevin