The source of pretreatment serum prostate-specific antigen in clinically localized prostate cancer--T, N, or M?

Prostate-specific antigen (PSA) is an important marker for prostate cancer and has been shown to be secreted from the primary tumor and from metastases. However, the relative contribution of the primary and micrometastatic disease to the serum level of PSA in patients with clinically localized disease has not been delineated. This study addresses the source of pretreatment serum PSA in patients with clinically localized disease.

The fall in serum PSA level following radical prostatectomy (280 patients; 105 T1, 165 T2, 10 T3) or definitive radiotherapy (427 patients; 122 T1, 147 T2, 158 T3/T4) was analyzed with the assumption that any fall in PSA following local treatment reflects the fraction of PSA produced in the prostate and its primary tumor.

Serum PSA level became undetectable in 277 of the 280 (99%) patients within 6 months of radical prostatectomy. The three patients who did not achieve undetectable levels had postsurgical values < or = 0.9 ng/ml. Following definitive radiotherapy, nadir serum PSA values were between < or = 0.3 and 20.3 ng/ml, with mean and median values of 1.9 and 1.2 ng/ml, respectively. Nadir PSA was undetectable in 52 patients (12%). Four patients' PSA did not fall, but rose from the start, and each developed metastatic disease within 9 months, and in each metastases appeared to contribute to pretreatment serum PSA. In the remaining patients, the maximal factor by which PSA fell to its nadir was higher the higher the pretreatment PSA level. We present arguments that this is most consistent with the hypothesis that virtually all detectable pretreatment serum PSA derives from the primary tumor. Confirmatory evidence that little of the pretreatment serum PSA came from metastases was obtained by extrapolating the rising PSA profile in 97 patients back to pretreatment time. Back-extrapolated PSA contributed a mean of 7% and a median of 5% to the pretreatment serum value. Because such back-extrapolated values estimate the potentially maximal micrometastatic contribution, metastatic disease at diagnosis contributes little to pretreatment serum PSA.

In patients with clinically localized prostate cancer, putative micrometastatic disease contributes negligibly to the pretreatment serum PSA level even when the latter is high. Most likely such patients, when they have metastases, have a very low metastatic burden.