NOMOGRAM TO PREDICT PATIENT-SPECIFIC PROBABILITIES OF METASTASIS-FREE SURVIVAL AT 1 AND 2 YEARS

" Overall, 78 (53%) of the 148 patients had organ-confined disease, and 9% of the patients had evidence of nodal spread. Approximately 70% had a Gleason score of 7 or lower, whereas the median PSA values at the time of diagnosis was 10.6 mg/mL (range, 1.8-200). Previous therapy was divided evenly between radiation therapy and surgery. ...Note that the median PSA values at the time of protocol enrollment was 8.2 ng/mL and median PSADT was 4.7 months. The median follow-up considered as the time to metastases, death, or to last follow-up visit, was 19.0 months, with a range of 1 to 75.3 months. Overall, 110 of the 148 (74%) patients showed evidence of metastatic progression to date, at a median of 19 months, with estimated 3- and 5-year metastasis-free survival rates of 32% (95% confidence interval, 24-40) and 16% (95% confidence interval, 9-25), respectively. At present, 15 (10%) patients have died of prostate cancer, and 133 remain alive.

In univariate analysis, patients with a higher T stage (log rank test, P = 0.07) and Gleason score of 7 or greater at the time of diagnosis were more likely to develop metastatic disease (log rank test; P = 0.006). The type of primary therapy, surgery or radiation therapy, and the presence of nodal disease were not predictive, although the number of patients with nodal involvement was small. The alkaline phosphatase (P = 0.81), hemoglobin (P = 0.16), and lactate dehydrogenase (P = 0.29) levels at the time of protocol enrollment were not predictive.

...T stage (P = 0.02), and Gleason score (P = 0.02) at the time of diagnosis, preprotocol  PSADT (P = 0.001) and PSA level at the time of protocol therapy (P < 0.001) were independent predictors for metastatic progression. Patient-specific predictions of the median time to metastatic progression and the probabilities for remaining metastasis-free at 1 and 2 years are shown by the nomogram  <above>. Each covariate scale position has a corresponding points scale (top scale). For example, a baseline PSA of 1.0 contributes 15 points to the total points scale. A point value for each predictor is determined and summed to compute the total points for a patient. This value is located on the total points scale (fourth from the bottom). The predicted 1- and 2-year progression-free survival, and the median progression-free survival is then determined by drawing a vertical line down from the total points scale to the respective predicted outcome scale.

The predictive accuracy of the nomogram was assessed using the concordance index. Similar to the area under the receiver operating characteristic curve, the concordance index provides the probability that in a randomly selected pair of patients, the patient with the longer progression-free survival had the better predicted outcome from the nomogram. The concordance index ranges between 0.5 and 1.0, with 1.0 representing a model with perfect discrimination, and 0.5 indicating that a coin flip would provide as accurate information as the model. We used bootstrapping with 200 replicates to obtain a relatively unbiased estimate. Specifically, this was done by first computing the apparent accuracy of the model using the entire data set. Second, we computed the accuracy on a sample of the same size but drawn with replacement. Third, we calculated the accuracy of this second model, using the original data set. Optimism is defined as the difference between the second and third accuracies in this process. After repeating this process 200 times, the mean optimism was subtracted from the first model accuracy to arrive at a nearly unbiased estimate. The bootstrap corrected concordance index was 0.69, indicating adequate discriminatory power for this model."

(From Slovin SF, Wilton AS, Heller G, Scher HI: Clin Cancer Res. 2005 Dec 15;11(24 Pt 1):8669-73. Time to detectable metastatic disease in patients with rising prostate-specific antigen values following surgery or radiation therapy.)