P Romundstad et al. American Journal of Epidemiology. May (2001) Vol. 153 #10 p 978.
If there had been no other exposure to carcinogens this research would be a strong indicator of causation but with several details to be worked out.
The Radar report is available to subscribers: