“…Differences between rates may be due to case mix 2,3 , surgical expertise 3±5 , differences in de®nition of local recurrence (pelvic only versus pelvic recurrence in the presence of metastatic disease) 6±8 , incomplete reporting of histopathological ®ndings 9 , lack of clarity in the de®nition of a curative operation 10 , variability in the protocol for and length of follow-up 6,11 , variability in data collection 12 and variability in the way in which local recurrence rates are calculated 8,13 . Furthermore, there have been few convincing studies of patient and tumour characteristics that might reliably predict local recurrence 14,15 . Such a study would require a large prospective patient cohort treated by standardized surgery alone, followed over a period of time suf®ciently long for most local recurrences to occur, and would include a comprehensive range of clinical variables with standardized pathology and staging.…”