Race and prostate weight as independent predictors for biochemical recurrence after radical prostatectomy

Abstract

We hypothesized that factors beyond pathological stage, grade, PSA and margin status would be important predictors of biochemical recurrence (BCR) after radical prostatectomy (RP). A cohort of 3194 patients who underwent RP between 1988 and 2007 and who had neither neoadjuvant therapy nor postoperative adjuvant hormonal therapy was retrieved from the Duke Prostate Center database. Age, prostate-specific antigen (PSA), pathological Gleason score (pG), lymph node status, seminal vesicle invasion (SVI), extracapsular extension (ECE), positive surgical margin (PSM) status, year of surgery, race, adjuvant radiation therapy (XRT), percent tumor involvement in the RP specimen and prostate weight were evaluated as possible predictors of BCR in multivariate Cox regression analysis. BCR was defined as a PSA of 0.2 ng ml−1 or higher at least 30 days after surgery. A nomogram was developed from the Cox model. Predictive accuracy was obtained by calculating bias-corrected Harrell's c and by bootstrap calibration. In multivariate analysis, PSA (hazard ratio 1.39 (95% confidence interval 1.29–1.51)), ECE (1.22 (1.04–1.44)), pG score (1.38 (1.14–1.68), 2.23 (1.76–2.84), 2.69 (2.12–3.40) for pG 3+4, 4+3, >7, respectively), SVI (1.72 (1.40–2.12)), PSM (2.05 (1.73–2.42)), year of surgery (0.65 (0.54–0.77)), African-American race (1.37 (1.13–1.66)), adjuvant XRT (0.19 (0.11–0.34)) and prostate weight (0.83 (0.76–0.92)) were identified as independent predictors of BCR (P0.018 for all factors). Predictive accuracy of the nomogram was 0.75. Race and prostate weight were independent predictors for BCR after RP. By incorporating these variables, we developed a nomogram, which provides a highly accurate means for estimating risk of BCR after RP.

Introduction

Biochemical recurrence (BCR) after radical prostatectomy (RP) is a well-accepted endpoint for the assessment of cancer control in patients with localized prostate cancer. As patients with BCR after RP are often subjected to salvage radiation or hormonal therapy,1 prediction tools have become a part of patient assessment and counseling after RP. Several nomograms have been developed to predict BCR after RP, using either preoperative2 or postoperative3, 4 variables. These nomograms mainly use preoperative prostate-specific antigen (PSA) and pathological variables as predictors. We hypothesized that additional variables might be clinically important independent predictors of BCR after RP. We sought to identify these variables and incorporate them into a nomogram for prediction of BCR after RP.

Methods

Patients

A cohort of 3242 patients who underwent RP between 1988 and 2007 and who had not undergone neoadjuvant therapy were retrieved from the institutional review board-approved Duke Prostate Center database. Forty-eight patients had received postoperative adjuvant hormonal therapy before BCR or last follow-up and were, therefore, excluded, leaving 3194 patients for further analysis. Patients were followed postoperatively as per the attending urologist's discretion with serial PSA determinations, and all postoperative PSA values were entered into the database.

Data analysis

Biochemical recurrence was defined as a PSA of 0.2 ng ml−1 or higher at least 30 days after surgery. Patients who had radiation therapy (XRT) before BCR and within 1 year of RP were defined as having adjuvant XRT, whereas patients who had XRT before BCR but more than 1 year after RP were treated as having failed primary therapy at the time point of XRT. Therefore, primary treatment failure was defined by either BCR or treatment with XRT more than 1 year after RP.

Age, PSA, pathological Gleason score, lymph node status, seminal vesicle invasion, extracapsular extension, positive surgical margin status, year of surgery, race, adjuvant XRT, percent tumor involvement in the RP specimen (measured as described previously5) and prostate weight were evaluated as possible predictors of BCR in multivariate Cox regression analysis. As PSA and prostate weight had a non-normal distribution and a nonlinear effect of these variables was expected, these variables were transformed to the natural logarithm before analysis. Age, PSA, year of surgery, percent tumor involvement and prostate weight were used as continuous variables, whereas Gleason score (<7, 3+4, 4+3, >7), lymph node status, seminal vesicle invasion, extracapsular extension, positive surgical margin status, race (African-American vs non-African-American) and XRT were used as categorical variables. Clustering of these variables was examined based on the Spearman's ρ2 value. Variables that did not reach statistical significance at a level of 0.05 were removed from the model in a stepwise process.

We did not consider body mass index (BMI) as a predictor while constructing our final model, as 31.7% of the patients were missing this variable. However, a subgroup analysis in the patients who had BMI data available did not identify BMI as an independent predictor for BCR and race remained significant even after adjusting for BMI (data not shown).

A nomogram was developed based upon the significant variables from the final Cox model. Predictive accuracy was obtained by calculating bias-corrected Harrell's c and by 200 samples bootstrap calibration. Bootstrap calibration plots were obtained, which compare the nomogram- or model-predicted chance for recurrence-free survival with the actual likelihood of recurrence-free survival in the modeling cohort.

All statistical analysis was performed with R version 2.5.1,6 using the packages ‘Design’7 and ‘Hmisc’.8

Results

The patients’ baseline clinicopathological variables are summarized in Table 1. Sixty-five patients (2.0%) received adjuvant radiation therapy by our definition. Median age was 62.6 years. Median and mean follow-up was 2.6 years and 3.9 years, respectively. Mean follow-up of the patients without primary treatment failure was 4.7 years. A total of 867 (27.1%) patients failed primary treatment. Figure 1 shows a 10-year estimated overall recurrence-free survival of 62.9% (95% confidence interval 60.5–65.2%).

Table 1 Clinicopathological characteristics of the patient cohort
Figure 1
figure1

Overall biochemical recurrence-free survival of the patient cohort. The 95% confidence interval is marked by bars and the number of patients at risk is given for 0, 5, 10 and 15 years.

Clustering analysis revealed that PSA clustered with year of surgery, age and prostate weight, whereas extracapsular extension (ECE) clustered with percent tumor involvement, positive surgical margins, seminal vesicle invasion, Gleason score of 8–10 and adjuvant radiation therapy. Race did not cluster with any of the variables. The highest Spearman's ρ2 value among the variables used for model development was observed between ECE and percent tumor involvement (ρ2 0.19).

After stepwise backward selection, the final Cox proportional hazards multivariate model identified preoperative PSA, positive surgical margins, seminal vesicle invasion, extracapsular extension, Gleason score, prostate weight, African-American race, year of surgery and adjuvant radiotherapy as independent predictors for biochemical recurrence (Table 2). This model was based on a total of 2746 patients, as 292 patients (9.1%) with missing preoperative PSA and 191 patients (6.0%) with missing prostate weight data were excluded. Overall, 706 of these 2746 patients (25.7%) experienced a recurrence.

Table 2 Multivariate Cox model predicting biochemical recurrence

From the Cox model, a nomogram was developed that allows calculating an individual patient's risk for BCR within 1, 2, 5, and 10 years after RP (Figure 2). As can be seen in the nomogram, preoperative PSA, prostate weight and Gleason score were the most important predictors for recurrence-free survival. In addition, race was a more relevant predictor for recurrence than ECE. Predictive accuracy of the nomogram was 75% (Harrell's c), which implies that correct discrimination of risk between a pair of patients with higher and lower risk was accurately achieved by using the nomogram in 75% of cases.

Figure 2
figure2

Nomogram for the prediction of biochemical recurrence (BCR) after radical prostatectomy. Instructions: draw a perpendicular line from the patient's value to the ‘Points’ scale on the top of the nomogram. Thus, you will obtain points for each variable. Prostate-specific antigen should be entered in nanograms per milliliter; prostate weight should be entered in grams. Add all points obtained and then draw a perpendicular line from the corresponding point on the ‘Total Points’ axis downwards. This will give you the estimated 1-, 2-, 5- and 10-year probabilities for this patient to remain without BCR.

Similarly, calibration plots showed a high accuracy of the nomogram (Figure 3). Every predicted probability fell within the 95% confidence interval for the actual probability.

Figure 3
figure3

Calibration plots illustrating the predictive accuracy of the nomogram for the prediction of 1-, 2-, 5- and 10-year recurrence-free survival after radical prostatectomy. For each probability of recurrence-free survival predicted by the nomogram, the actual probability with 95% confidence interval (bars) from the modeling cohort is given.

Discussion

In addition to well-known predictors for BCR after RP, we identified race and prostate weight as additional independent predictors and developed a highly accurate nomogram, which incorporates these factors.

Several studies have addressed the importance of race for BCR of prostate cancer after RP with conflicting results. Some did not find race to be an independent risk factor associated with BCR,9, 10, 11, 12, 13, 14 whereas others did identify race as an independent risk factor for BCR.15, 16, 17, 18 Interestingly, Iselin et al.,10 did not find race to be an independent predictor of BCR in a cohort of patients from our institution 10 years ago. In their study, 77 (9.8%) out of 784 PSA-era patients were African-American. In multivariate analysis, they found African-American race to be associated with a hazard ratio of 1.56 (95% confidence interval 0.93–2.63, P=0.09). This risk is very similar to the risk found in the present study. Therefore, one could argue that Iselin's study was not powered enough to detect the modestly increased risk for African-American men at that time.

Powell et al.19 found that the effect of race on BCR was conditional by showing that race was an independent risk factor for BCR in patients with ECE, but not in patients without ECE. This effect was not seen in our data, as race remained a significant independent predictor in the subgroup of patients without ECE (data not shown). Despite these differences, one could argue that race was still a clinically significant factor for BCR after RP in patients without ECE in the study of Powell et al.,19 as the hazard ratio for African-American race in this population was 2.7 with a 95% confidence interval of 0.9–7.9.

It has been previously suggested that the significance of race for BCR might, at least in part, be driven by the fact that African-American men tend to be more obese.14, 20 In our study, BMI was not identified as an independent predictor for BCR in a subanalysis and race remained significant even after adjusting for BMI. This is in line with data reported by Amling et al.,20 who found race, but not obesity to be an independent predictor of BCR in multivariate analysis. In summary, data on the importance of race in predicting BCR after RP are conflicting, and different results in different studies might be due to different patient populations examined in these studies.

Previous studies have found that a larger prostate decreases the likelihood of positive surgical margins in robot-assisted laparoscopic21 and open22 radical prostatectomy. Similarly, higher prostate weight has been found to be protective against BCR.22, 23, 24, 25 Our finding that increased prostate weight is an independent predictor for improved recurrence-free survival is in concordance with these reports. Of note, the effect of prostate weight was not driven by outliers, as prostate weight remained a significant independent factor when the analysis was limited to men with a prostate weight between 20 and 100 grams (data not shown). Several previous studies have suggested that lead-time bias could explain the predictive significance of prostate weight. Increased prostate weight would lead to higher PSA values and, thus, these patients might be subjected to biopsy earlier than patients with small prostates.23, 24, 25 However, Freedland et al.22 argued that this phenomenon could not account for the effect of prostate weight in their study, as an analysis of only the patients who had a palpable nodule on prostate biopsy still showed prostate weight to be a significant independent predictor for higher recurrence-free survival.

Others have suggested a more physical effect of prostate weight. A similar size tumor in a larger prostate would have a lower chance to penetrate the capsule or reach the surgical margin than in a smaller prostate, simply due to geometrical considerations.26

Regardless of the reasons leading to a larger prostate being protective against BCR after RP, we believe that incorporating this variable into a nomogram is a valuable addition. Prostate weight can significantly vary between individual patients and between patients in different care settings.27, 28 In fact, it was suggested previously to incorporate prostate weight into predictive nomograms to increase their generalizability across practice settings.22

Our study has several limitations. Our patient population underwent RP at a tertiary care center and all surgeons had extensive experience in performing RP. As surgeon experience has been shown to be an important factor influencing BCR-free survival,29 our nomogram may predict an overly optimistic chance of recurrence-free survival for patients who are operated on by less experienced surgeons.

Two randomized controlled trials have shown a benefit of adjuvant radiation therapy for reducing the risk of BCR in patients with high risk for BCR.30, 31 Although adjuvant radiation therapy was found to be a clinically important and statistically significant independent predictor for better recurrence-free survival in the current retrospective analysis, these results are based on a relatively low number of patients who underwent adjuvant radiation therapy as per the treating physician's discretion. Moreover, some of the patients might have been misclassified as not having received adjuvant XRT, if they had XRT outside of the Duke medical system and their medical record did not reflect this treatment. The purpose of our nomogram is, therefore, not to identify patients who would benefit from adjuvant radiation therapy, but rather to provide a more general means to estimate a specific patient's ‘cure rate’ during postoperative counseling. In fact, other studies have shown that patients with a positive surgical margin might benefit more from radiation therapy than patients with a negative margin but who had other risk factors, such as ECE or seminal vesicle invasion.32 This suggests that predictors for success of adjuvant XRT might differ significantly from predictors of BCR after RP.

Conclusion

Race and prostate weight were readily available and statistically significant independent predictors for BCR after RP. By incorporating these as well as well-known predictors of BCR, we have developed a nomogram, which provides a highly accurate means for estimating risk of BCR after RP.

References

  1. 1

    Freedland SJ, Moul JW . Prostate specific antigen recurrence after definitive therapy. J Urol 2007; 177: 1985–1991.

  2. 2

    Stephenson AJ, Scardino PT, Eastham JA, Bianco Jr FJ, Dotan ZA, Fearn PA et al. Preoperative nomogram predicting the 10-year probability of prostate cancer recurrence after radical prostatectomy. J Natl Cancer Inst 2006; 98: 715–717.

  3. 3

    Stephenson AJ, Scardino PT, Eastham JA, Bianco Jr FJ, Dotan ZA, DiBlasio CJ et al. Postoperative nomogram predicting the 10-year probability of prostate cancer recurrence after radical prostatectomy. J Clin Oncol 2005; 23: 7005–7012.

  4. 4

    Kattan MW, Wheeler TM, Scardino PT . Postoperative nomogram for disease recurrence after radical prostatectomy for prostate cancer. J Clin Oncol 1999; 17: 1499–1507.

  5. 5

    Mouraviev V, Mayes JM, Sun L, Madden JF, Moul JW, Polascik TJ . Prostate cancer laterality as a rationale of focal ablative therapy for the treatment of clinically localized prostate cancer. Cancer 2007; 110: 906–910.

  6. 6

    R Development Core Team. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria 2007. http://www.R-project.org.

  7. 7

    Harrell FE . Design: Design Package. R package version 2.1-1 2007.

  8. 8

    Harrell FE . Hmisc: Harrell Miscellaneous. R package version 3.4-2 2007.

  9. 9

    Hamilton RJ, Aronson WJ, Presti Jr JC, Terris MK, Kane CJ, Amling CL et al. Race, biochemical disease recurrence, and prostate-specific antigen doubling time after radical prostatectomy: results from the SEARCH database. Cancer 2007; 110: 2202–2209.

  10. 10

    Iselin CE, Box JW, Vollmer RT, Layfield LJ, Robertson JE, Paulson DF . Surgical control of clinically localized prostate carcinoma is equivalent in African-American and white males. Cancer 1998; 83: 2353–2360.

  11. 11

    Eastham JA, Kattan MW . Disease recurrence in black and white men undergoing radical prostatectomy for clinical stage T1-T2 prostate cancer. J Urol 2000; 163: 143–145.

  12. 12

    Cross CK, Shultz D, Malkowicz SB, Huang WC, Whittington R, Tomaszewski JE et al. Impact of race on prostate-specific antigen outcome after radical prostatectomy for clinically localized adenocarcinoma of the prostate. J Clin Oncol 2002; 20: 2863–2868.

  13. 13

    Sohayda CJ, Kupelian PA, Altsman KA, Klein EA . Race as an independent predictor of outcome after treatment for localized prostate cancer. J Urol 1999; 162: 1331–1336.

  14. 14

    Nielsen ME, Han M, Mangold L, Humphreys E, Walsh PC, Partin AW et al. Black race does not independently predict adverse outcome following radical retropubic prostatectomy at a tertiary referral center. J Urol 2006; 176: 515–519.

  15. 15

    Powell IJ, Dey J, Dudley A, Pontes JE, Cher ML, Sakr W et al. Disease-free survival difference between African Americans and whites after radical prostatectomy for local prostate cancer: a multivariable analysis. Urology 2002; 59: 907–912.

  16. 16

    Grossfeld GD, Latini DM, Downs T, Lubeck DP, Mehta SS, Carroll PR . Is ethnicity an independent predictor of prostate cancer recurrence after radical prostatectomy? J Urol 2002; 168: 2510–2515.

  17. 17

    Tarman GJ, Kane CJ, Moul JW, Thrasher JB, Foley JP, Wilhite D et al. Impact of socioeconomic status and race on clinical parameters of patients undergoing radical prostatectomy in an equal access health care system. Urology 2000; 56: 1016–1020.

  18. 18

    Moul JW, Douglas TH, McCarthy WF, McLeod DG . Black race is an adverse prognostic factor for prostate cancer recurrence following radical prostatectomy in an equal access health care setting. J Urol 1996; 155: 1667–1673.

  19. 19

    Powell IJ, Banerjee M, Bianco FJ, Wood Jr DP, Dey J, Lai Z et al. The effect of race/ethnicity on prostate cancer treatment outcome is conditional: a review of Wayne State University data. J Urol 2004; 171: 1508–1512.

  20. 20

    Amling CL, Riffenburgh RH, Sun L, Moul JW, Lance RS, Kusuda L et al. Pathologic variables and recurrence rates as related to obesity and race in men with prostate cancer undergoing radical prostatectomy. J Clin Oncol 2004; 22: 439–445.

  21. 21

    Zorn KC, Orvieto MA, Mikhail AA, Gofrit ON, Lin S, Schaeffer AJ et al. Effect of prostate weight on operative and postoperative outcomes of robotic-assisted laparoscopic prostatectomy. Urology 2007; 69: 300–305.

  22. 22

    Freedland SJ, Isaacs WB, Platz EA, Terris MK, Aronson WJ, Amling CL et al. Prostate size and risk of high-grade, advanced prostate cancer and biochemical progression after radical prostatectomy: a search database study. J Clin Oncol 2005; 23: 7546–7554.

  23. 23

    Stamey TA, Yemoto CM, McNeal JE, Sigal BM, Johnstone IM . Prostate cancer is highly predictable: a prognostic equation based on all morphological variables in radical prostatectomy specimens. J Urol 2000; 163: 1155–1160.

  24. 24

    Foley CL, Bott SR, Thomas K, Parkinson MC, Kirby RS . A large prostate at radical retropubic prostatectomy does not adversely affect cancer control, continence or potency rates. BJU Int 2003; 92: 370–374.

  25. 25

    D’Amico AV, Whittington R, Malkowicz SB, Schultz D, Tomaszewski JE, Wein A . A prostate gland volume of more than 75 cm3 predicts for a favorable outcome after radical prostatectomy for localized prostate cancer. Urology 1998; 52: 631–636.

  26. 26

    Msezane LP, Gofrit ON, Lin S, Shalhav AL, Zagaja GP, Zorn KC . Prostate weight: an independent predictor for positive surgical margins during robotic-assisted laparoscopic radical prostatectomy. Can J Urol 2007; 14: 3697–3701.

  27. 27

    Nixon RG, Meyer GE, Brawer MK . Differences in prostate size between patients from University and Veterans Affairs Medical Center populations. Prostate 1999; 38: 144–150.

  28. 28

    Terris MK, Prestigiacomo AF, Stamey TA . Comparison of prostate size in university and veterans affairs health care system patients with negative prostate biopsies. Urology 1998; 51: 412–414.

  29. 29

    Vickers AJ, Bianco FJ, Serio AM, Eastham JA, Schrag D, Klein EA et al. The surgical learning curve for prostate cancer control after radical prostatectomy. J Natl Cancer Inst 2007; 99: 1171–1177.

  30. 30

    Bolla M, van Poppel H, Collette L, van Cangh P, Vekemans K, Da Pozzo L et al. Postoperative radiotherapy after radical prostatectomy: a randomised controlled trial (EORTC trial 22911). Lancet 2005; 366: 572–578.

  31. 31

    Thompson Jr IM, Tangen CM, Paradelo J, Lucia MS, Miller G, Troyer D et al. Adjuvant radiotherapy for pathologically advanced prostate cancer: a randomized clinical trial. JAMA 2006; 296: 2329–2335.

  32. 32

    Van der Kwast TH, Bolla M, Van Poppel H, Van Cangh P, Vekemans K, Da Pozzo L et al. Identification of patients with prostate cancer who benefit from immediate postoperative radiotherapy: EORTC 22911. J Clin Oncol 2007; 25: 4178–4186.

Download references

Acknowledgements

We thank Donna E Levy for help with the statistical analysis.

Author information

Correspondence to J W Moul.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Schroeck, F., Sun, L., Freedland, S. et al. Race and prostate weight as independent predictors for biochemical recurrence after radical prostatectomy. Prostate Cancer Prostatic Dis 11, 371–376 (2008). https://doi.org/10.1038/pcan.2008.18

Download citation

Keywords

  • race
  • prostate weight
  • biochemical recurrence
  • prediction
  • nomogram

Further reading