Glen Tisman, MD: Nomograms – Predicting Survival in Patients Resistant to Castration

Quick Search


Nomograms 7 & 8
Predicting Survival in Patients Resistant to Castration

Nomograms 7 and 8 were discussed in a publication by Matt Galsky and William Kelly.7 They are two point-based nomograms (Nomogram 7 by Smaletz et al and Nomogram 8 by Halabi et al). Both are based on pretreatment variables measured on a routine basis in patients with chemical castration (a form of androgen deprivation) and refractory prostate cancer. These models provide a user-friendly format in which to make sophisticated predictions of survival in patients who demonstrate recurrent prostate cancer after hormonal ablation. These models appear to have improved the ability to predict the outcomes of patients with castrate metastatic disease.

In Nomogram 7, pretreatment clinical and biochemical determinants from 409 patients enrolled into 19 consecutive therapeutic protocols were evaluated. The variables included patient age, Karnofsky performance status, alkaline phosphatase, and serum albumin. These parameters were combined to produce a nomogram to predict median, 1-year, and 2-year survival. The nomogram was validated internally and externally. Calibration plots suggested that the nomogram was well calibrated for all predictions.

More advanced disease parameters, such as poor performance status, low albumin and hemoglobin, and high LDH and alkaline phosphatase levels, were associated with lower prognostic scores. A higher age was associated with better survival, but the overall contribution of age was modest. The predictive effect of PSA and albumin was not intuitive. Interestingly higher levels of PSA were associated with better survival. The overall contribution of PSA was small, and the changes in other predictors, which often accompany increases in PSA, likely negate the seemingly protective effect of PSA.

Nomogram 8 was recently reported by Halabi et al. In this model, data were pooled from six Cancer and Leukemia Group B (CALGB) protocols that enrolled 1101 patients with castrate metastatic prostate cancer. Variables were assessed first in a univariate model, and all of the significant parameters were then entered in a multivariable model. In the multivariable analysis, statistically significant prognostic factors of overall survival included ECOG performance status, Gleason score from the original prostatectomy or biopsy specimen, LDH, PSA, alkaline phosphatase, and hemoglobin.

Nomogram 8 was constructed based on these parameters in addition to the presence or absence of visceral disease (disease involving organs such as the liver). Unlike Nomogram 7, then, Nomogram 8 includes the presence of visceral disease and Gleason score. However, the contribution of each of these parameters is modest. Furthermore, in Nomogram 8, higher levels of baseline PSA were associated with a worse prognosis. A limitation to both of these models is that they were developed using data from highly selected patients who met eligibility criteria for clinical trials. Therefore, they may not be useful for patients with a poor performance status or severe organ dysfunction. In addition, the measures of discrimination in both models suggested that for approximately 30% of patient-pairs, the patient predicted to have a better prognosis died first (oops). This underscores the need for better models.

Clinical scenario 7: Using Nomogram 7, a castrate refractory, 75 year-old, man (15 points) with a Karnofsky performance status of 80 (normal activity with effort and some complaints caused by the disease = 40 points), with a hemoglobin of 11 (37 points), with a PSA = 44 (5 points), LDH = 300 (43 points), ALK = 99 (7 points), and albumin = 3.0 (12 points) has a total of 159 points. The chance of this patient surviving for one year is calculated to be approximately 16% and for two years of about 3%.

Clinical scenario 8: Using Nomogram 8, a man without visceral involvement (no tumor in lungs or liver etc.) (0 points), with a tumor of Gleason score 8 (17 points), an ECOG performance status = 0, (0 points), with a baseline PSA = 300 (25 points), with LDH = 20 (18 points), with AP = 40 (15 points), and with hemoglobin = 9 (32 points), would have a total of 107 points. His 12-month probability for survival would be 78% (24-month = 50%), and the median survival would be 24 months. (Median survival is the survival time of 50% of the studied patients.)

7. Galsky M, Kelly WK: Use of nomograms for predicting survival in patients with castrate prostate cancer. Urology, Vol 62 (Supplement 6B), 119-127, 2003.

Next Nomogram > >

Return to Tisman Paper

Related Posts: