Another study in coffee intake and its association with lower PCa risk. It does NOT address post-diagnosis correlations, however. There's a good discussion on the limitations of the study, and reasonable explanation on what statistical methods they used to investigate their influence. Looks to me they looked into 0-10 "cups" per day range of consumption, and the linear relationship worked.
"Results Sixteen prospective cohort studies were finally included, with 57 732 cases of prostate cancer and 1 081 586 total cohort members. Higher coffee consumption was significantly associated with a lower risk of prostate cancer. Compared with the lowest category of coffee consumption, the pooled relative risk (RR) was 0.91 (95% CI 0.84 to 0.98), I2= 53.2%) for the highest category of coffee consumption. There was a significant linear trend for the association (p=0.006 for linear trend), with a pooled RR of 0.988 (95% CI 0.981 to 0.995) for each increment of one cup of coffee per day. For localised, advanced and fatal prostate cancer, the pooled RRs were 0.93 (95% CI 0.87 to 0.99), 0.88 (95% CI 0.71 to 1.09) and 0.84 (95% CI 0.66 to 1.08), respectively. No evidence of publication bias was indicated in this meta-analysis."