Nomogram Helps Predict Prostate Ca Mets

ORLANDO—A new nomogram may help physicians predict the probability of prostate cancer metastasis in patients with a biochemical recurrence after radical prostatectomy, nomogram codeveloper Zohar A. Dotan, MD, PhD, reported at the 2005 Multidisciplinary Prostate Cancer Symposium (abstract 162). "The nomogram’s prediction power was significantly higher than that of currently available prediction tools," said Dr. Dotan, a clinical fellow in the Department of Urology at Memorial Sloan-Kettering Cancer Center. Among the study coauthors were Peter Scardino, MD, and Michael Kattan, PhD.

About 50,000 to 60,000 new patients experience biochemical failure annually, Dr. Dotan said. Physicians lack a reliable way to determine probability of metastasis in these men. Some patients do well, while other patients’ cancers may spread rather quickly.

In developing the computerized tool, the researchers queried Sloan-Kettering’s multidisciplinary prostate cancer database for patients initially treated with surgery who later developed biochemical failure. They defined biochemical failure as a PSA of 0.2 ng/mL and rising, a single PSA of 0.4 ng/mL or higher, or the use of a secondary treatment for a rising PSA. Investigators excluded patients who had received preoperative radiotherapy, hormonal therapy, or chemotherapy.

From a database of 4,438 patients, they identified 675 patients with biochemical failure after radical prostatectomy. Of these 675 patients, 141 developed metastasis. About half of the men were receiving hormonal therapy at the time of metastatic progression. Median follow-up from identification of biochemical failure to the end of the study was 5.2 years. Median time from radical prostatectomy to biochemical failure was 1.4 years.

Fifty-four percent of the patients with biochemical failure had a preoperative PSA greater than 10 ng/mL; 74% had a Gleason score of 7 or higher. As secondary treatment for PSA progression, 191 received radiation therapy, and 300 were given hormonal therapy.

Constructing the Nomogram

Using a Cox proportional hazards analysis, the team looked at the following predictive factors: preoperative PSA; time from surgery to biochemical failure; pathologic findings (such as surgical margin, extracapsular extension, seminal vesicle invasion, lymph node metastases, and pathologic Gleason score); PSA doubling time; and PSA value at the time of biochemical failure. They also considered secondary treatments, such as radiation therapy and hormonal therapy, as time-dependent covariates.

Patients with extracapsular extension, seminal vesicle invasion, a Gleason score of 8 to 10, higher PSA levels at time of biochemical failure, and a short PSA doubling time were significantly more likely to develop metastasis. Those receiving radiation therapy as a second treatment were significantly less likely to have their cancer spread.

"Interestingly, radiation therapy was associated with reduction of metastatic progression," Dr. Dotan said at a media briefing. "It is, as far as we know, the first linkage to reduction of metastatic progression following another type of local therapy for patients with biochemical failure. Unfortunately, hormonal therapy, the most common systemic therapy for prostate cancer, was not associated with reduction of metastatic progression."

To construct the nomogram, Dr. Dotan and his colleagues incorporated the predictive variables that were significant on multivariate analysis, as well as preoperative PSA, surgical margins, lymph node status, and time from surgery to biochemical failure. The resulting nomogram has a concordance index of 0.88; internal validation showed its accuracy. Dr. Dotan said that the nomogram has a discrimination ability significantly superior to categorical PSA doubling time of 10 months and the Pound models of 1999 and 2003. He is now using an external database to validate the findings from this study.

"The nomogram can be used to identify high-risk patients for systemic progression, to identify patients who will need systemic and local therapy, and, most important, to identify entry point for clinical trials based on the probability of progression," Dr. Dotan said.