Percentage of Gleason pattern 4 and 5 predicts survival after radical prostatectomy

Authors

  • Liang Cheng MD,

    Corresponding author
    1. Department of Pathology and Laboratory Medicine, Indiana University School of Medicine, Indianapolis, Indiana
    2. Department of Urology, Indiana University School of Medicine, Indianapolis, Indiana
    • Department of Pathology and Laboratory Medicine, Indiana University School of Medicine, 350 West 11th Street, Clarian Pathology Laboratory Room 4010, Indianapolis, IN 46202
    Search for more papers by this author
    • Fax: (317) 491-6419

  • Darrell D. Davidson MD, PhD,

    1. Department of Pathology and Laboratory Medicine, Indiana University School of Medicine, Indianapolis, Indiana
    Search for more papers by this author
  • Haiqun Lin MD, PhD,

    1. Department of Biostatistics, Yale University, New Haven, Connecticut
    Search for more papers by this author
  • Michael O. Koch MD

    1. Department of Urology, Indiana University School of Medicine, Indianapolis, Indiana
    Search for more papers by this author

Abstract

BACKGROUND.

Morphologic and clinical heterogeneity within tumor grades is well recognized in prostate cancer. The objective of the current study was to determine whether the combined percentage of Gleason patterns 4 and 5 in radical prostatectomy specimens is an independent predictor of cancer-specific survival in prostate cancer patients.

METHODS.

The radical prostatectomy specimens were analyzed from 504 consecutive prostate cancer patients who were treated at Indiana University Medical Center between 1990 and 1998. Various clinical and pathologic characteristics were analyzed.

RESULTS.

A higher combined percentage of Gleason patterns 4 and 5 was associated with older age, higher preoperative serum prostate-specific antigen level, higher pathologic stage, positive surgical margins, extraprostatic extension of tumor, higher Gleason score, perineural invasion, and lymph node metastasis. In the multivariate Cox regression model, the combined percentage of Gleason patterns 4 and 5 was found to be an independent predictor of cancer-specific survival (P = .04).

CONCLUSIONS.

The combined percentage of Gleason patterns 4 and 5 is a powerful predictor of prostate cancer-specific survival. Assessment of high-grade cancer amounts may allow for better stratification of patients into appropriate prognostic groups and treatment protocols. Cancer 2007. © 2007 American Cancer Society.

In recent years, many pathologic grading systems have been simplified to separate neoplastic diagnostic categories into either high or low risk for aggressive biologic behavior. Although the 5-pattern, 9-grade rating scale proposed by Gleason has proven its prognostic value for over 40 years,1–4 there is a need for improved tools with which to predict which patients should undergo more vigorous therapy after surgery.5 This retrospective analysis was undertaken to determine whether the proportion of high-grade carcinoma (the combined percentage of Gleason patterns 4 and 5) could be used to predict cancer-specific survival in prostate cancer patients treated with radical prostatectomy.

MATERIALS AND METHODS

Patients

The study cohort consisted of 504 men who underwent radical retropubic prostatectomy for clinically localized prostate cancer between 1990 and 1998 at Indiana University Hospital (Indianapolis, Ind). None of the patients had preoperative radiotherapy or androgen deprivation therapy. Clinical records were reviewed to determine patient age, preoperative serum prostate-specific antigen (PSA) levels, and cancer death. Patients were followed at 6-month intervals for 2 years and annually thereafter. During a mean follow-up period of 44 months (median, 36 months; range, 1.5–144 months), 20 patients died of prostate cancer and 45 patients died of other causes (Fig. 1). All patients who died of prostate cancer had developed distant metastasis.

Figure 1.

Kaplan-Meier survival curves for 504 patients who underwent radical prostatectomy for prostate cancer.

This research was approved by the Indiana University Institutional Review Board.

Radical Prostatectomy Specimens

Pathologic examination of the radical prostatectomy specimens was performed by a single urologic pathologist (L.C.) as described previously.6–8 The prostatectomy specimens were weighed, measured, inked, and fixed in 10% neutral formalin. After fixation, the apex and bladder base were amputated and serially sectioned at 3-mm to 5-mm intervals in the vertical, parasagittal plane. The seminal vesicles were amputated and sectioned with representative sections submitted for microscopic examination. The remaining prostate was serially sectioned at 4-mm intervals perpendicular to the long axis from the apex of the prostate to the bladder base. A partial sampling method was used, and representative sections from each quadrant were submitted for examination. Five-micron sections were then prepared and stained with hematoxylin and eosin. The total number of blocks examined (excluding those submitted from seminal vesicle and lymph nodes) ranged from 12 to 44 (mean, 12 blocks; median, 12 blocks).

All cases were staged according to the 2002 American Joint Committee on Cancer (AJCC) TNM staging system.9 Surgical margins were considered positive when carcinoma cells were in contact with any inked margin.10, 11 All cancers were assigned Gleason scores according to the Gleason grading system.1–3 The Gleason score is obtained by summation of the primary Gleason pattern (most prevalent pattern) and the secondary Gleason pattern (second most prevalent), based on assessment of the entire specimen. The percentage of Gleason pattern 4 or 5 cancer (percent Gleason pattern 4/5) in each case (ie, the combined proportion of the tumor comprised of Gleason pattern 4 or pattern 5 or both) was evaluated separately and semiquantitatively with values ranging from 0 to 100%, as described previously.12 The percent Gleason pattern 4/5 was derived by adding the percentages of Gleason patterns 4 and 5.12 Only malignant epithelial elements were considered in the calculation of percent Gleason pattern 4/5.

Statistical Analysis

SAS software (version 9.1.2; SAS Institute, Inc, Cary, NC) was used to perform the analysis. The Student t-test was used to assess the association between the percent Gleason pattern 4/5 and a variable of interest. Postsurgical duration of cancer-free survival was tested for association with factors such as pathologic stage, surgical margins, etc. Conventional prognostic factors were then analyzed together with the percent Gleason pattern 4/5 in multiple regression analysis by the Cox survival model. The percent Gleason pattern 4/5 was analyzed as a continuous variable. A predictor was considered significant if the P value was <.05. All P values were 2-sided.

RESULTS

The mean patient age was 62 years (range, 34–80 years; median, 63 years). The pathologic stage was T2a or T2b in 104 patients (21%), T2c in 244 patients (48%), T3a in 89 patients (18%), and T3b in 67 patients (13%).

Areas of high-grade carcinoma (either with Gleason pattern 4 or 5 or both) were found in 342 patients (68%). The proportion of high-grade tumor was fairly evenly distributed among the cases in which it was present, with 36% having Gleason pattern 4/5 in ≤30% of the malignant cells, 27% in 30 to 60% of malignant cells, and 37% in 60 to 100% of the malignant cells. In 22% of specimens in which a Gleason pattern 4/5 tumor was present, it constituted ≤20% of the tumor area.

On univariate analysis, the increased percent Gleason 4/5 was associated with older age, higher preoperative serum PSA level, higher pathologic stage, positive surgical margins, extraprostatic tumor extension, higher Gleason score, perineural invasion, and lymph node metastasis (Table 1). The high-grade cancer proportion (the percent Gleason pattern 4/5) demonstrated no association with prostate weight or high-grade prostatic intraepithelial neoplasia (PIN).

Table 1. Correlation Between the Combined Percentage of Gleason Patterns 4 and 5 and Other Clinicopathologic Characteristics in 504 Patients Treated With Radical Prostatectomy Without Adjuvant Therapy
CharacteristicsNo. of patientsMean % Gleason pattern 4/5P*
  • PSA indicates prostate-specific antigen.

  • *

    All P values were 2-sided.

  • Preoperative PSA levels were not available in 80 patients.

  • Prostate weight was not available in 73 patients.

Clinical findings
Age, y  .007
 ≥6323740 
 <6326731 
Preoperative PSA, ng/mL  <.0001
 <1026227 
 10–19.911246 
 ≥205052 
Primary tumor
Prostate weight, g  .65
 <4621434 
 ≥4621736 
Pathologic stage  <.0001
 pT234824 
 pT315662 
Surgical margins  <.0001
 Negative33030 
 Positive17446 
Extraprostatic extension  <.0001
 Present13761 
 Absent36726 
Gleason score  <.0001
 ≤61821 
 722843 
 ≥89485 
Perineural invasion  <.0001
 Absent1139 
 Present39143 
High-grade prostatic intraepithelial neoplasia  .26
 Absent1844 
 Present48635 
Lymph node metastasis  <.0001
 Absent48634 
 Present1872 

During a mean follow-up period of 44 months (range, 1.5–144 months), 20 patients died of prostate cancer. The percent Gleason pattern 4/5 was associated with prostate cancer-specific survival (P = .004) (Fig. 2). No patient died of prostate cancer among those who did not have high-grade tumor components (N = 162). Three patients who had from 1% to ≤20% of percent Gleason pattern 4/5 died of prostate cancer (N = 74); 17 patients who had >20% of the percent Gleason pattern 4/5 died of prostate cancer (N = 268). The 10-year cancer-specific survival was 100% for those without high-grade prostate cancer, compared with 70% for those with the presence of high-grade prostate cancer (P = .0017).

Figure 2.

Prostate cancer-specific survival according to the amount of high-grade prostate cancer (the combined percentage of Gleason patterns 4 and 5). The combined percentage of Gleason patterns 4 and 5 was analyzed as a continuous variable and is presented here for illustrative purposes.

In the multivariate Cox regression model, the percent Gleason pattern 4/5 (analyzed as continuous variable) was found to be an independent predictor of cancer-specific survival (P = .02) (Table 2). The percent Gleason pattern 4/5 was still significant when it was analyzed as a nominal variable (P = .04). The 10-year cancer-specific survival rate was 100%, 85%, and 67%, respectively, for those patients with 0% high-grade prostate cancer, 1% to 20% high-grade cancer, and >20% high-grade cancer (Fig. 2).

Table 2. Results of Multivariate Analysis of the Prediction of Cancer-Specific Survival Using the Cox Model
Variables95% CIHazards ratioP
  1. 95% CI indicates 95% confidence interval; PSA, prostate-specific antigen.

Percent Gleason pattern 4/51.01–1.041.02.02
Lymph node status4.9–77.0619.5<.001
Pathologic stage0.19–4.460.92.92
Gleason score0.13–9.121.10.94
Surgical margin status0.59–4.041.54.38
Age0.95–1.011.02.57
Preoperative PSA level0.93–1.020.97.27

DISCUSSION

In the current study, we identified high-grade carcinoma (either with Gleason pattern 4 or 5 or both) in 68% of the prostatectomy specimens from all men who underwent radical retropubic prostatectomy at our institution during an 8-year period from 1990–1998. The relative proportion of high-grade carcinoma (the percent Gleason pattern 4/5) was found to be strongly associated with established prognostic factors such as higher preoperative PSA levels, advanced pathologic stage, higher Gleason score, positive surgical margins, extraprostatic extension, lymph node metastasis, and perineural invasion. In the multivariate Cox model, the percent Gleason pattern 4/5 and lymph node status were found to be the only parameters that were predictive of cancer-specific survival. In the absence of any components of Gleason pattern 4 or 5 cancer, patients were essentially cured of prostate cancer after radical prostatectomy. The 10-year cancer-specific survival was 100%, 85%, and 67%, respectively, for those patients with 0% high-grade prostate cancer, 1% to 20% high-grade cancer, >20% high-grade cancer (Fig. 2).

The Gleason grading system, an established guideline in the U.S. since its formulation in the 1960s, has recently been adopted by the World Health Organization (WHO) as the international standard for histologic grading of prostate cancer.13 The Gleason score is the sum of the predominant glandular architecture pattern (primary Gleason pattern) with the next most prevalent pattern (secondary Gleason pattern). The original Gleason grading scheme does not address the issue of tertiary (third most prevalent) pattern.2 Because the Veterans Administration Cooperative Urological Research Group studies,2, 4 from which the Gleason grading system was derived, were largely based on transurethral prostatic resection specimens and core needle biopsies, a tertiary pattern was rarely reported. The need for a grading system to account for more than 2 patterns of adenocarcinoma was not present at the time when the Gleason grading system was proposed. However, over the past 30 years, screening techniques have led to the earlier detection of prostate cancer. Radical prostatectomy has emerged as the preferred method of treatment for prostatic carcinoma in selected patients. There is a need to modify our approach to the originally proposed Gleason scoring system as new data emerge.12–15 The worst cancer grade is directly linked to patient outcome. Scoring (or quantification) of the worst cancer grade is relevant for prostate cancer treatment and prognosis. The biology of prostate cancer is dictated by the worst cancer grade.

There have been many clinicopathologic studies focused on radical prostatectomy specimens, but to our knowledge, only a few have investigated the significance of a tertiary component. Pan et al.16 proposed that the Gleason system for radical prostatectomy specimens be modified to take into account small volumes (<5%) of Gleason patterns 4 and 5. Evidence has accumulated supporting the concept that the percent Gleason pattern 4/5 tumor is an important prognostic factor.14, 17–19 Comparing tumors with Gleason conventional score 7, those with >50% of Gleason pattern 4 (ie, Gleason score 4 + 3) portend a significantly worse prognosis than tumors with Gleason score 3 + 4.15, 20–23 McNeal et al.18 analyzed 209 entirely embedded radical prostatectomy specimens and found that the percent Gleason pattern 4/5 was predictive of lymph node metastasis. Stamey et al.14 found that the percent Gleason pattern 4/5 provided additional prognostic information beyond the conventional Gleason score. His group proposed that pathologists should move away from the traditional Gleason scoring system and simply estimate the combined percentage of Gleason patterns 4/5.

Because prostate carcinoma is a multifocal disease, it comes as no surprise that more than 2 Gleason patterns are often found in a single radical prostatectomy specimen. In a detailed study of 115 whole-mount radical prostatectomy specimens, Arora et al.13 found that 87% of all specimens contained 2 or more foci of adenocarcinoma and that the Gleason scores of individual foci often did not correlate with the overall Gleason score. In a series of 364 consecutive radical prostatectomies processed by whole-mount technique, the percent Gleason pattern 4/5 provided the best estimate of risk for cancer progression, regardless of the overall Gleason score.12 Molecular studies suggest that multiple tumors arise independently within a single prostate, most likely due to field effect.24 Tertiary minor high-grade components (Gleason pattern 4/5) are often identified in specimens with overall Gleason scores of 5 or 6.13, 25 Cancer heterogeneity is well recognized in a variety of tumor types. In many cancers, the presence of any high-grade elements dictates the biologic behavior of the tumors.12, 15, 26–29 The findings of the current study uphold the importance of high-grade carcinoma for predicting cancer progression in patients treated by radical prostatectomy. The current study is different from our previous analysis of the percent Gleason pattern 4/5, in which this value predicted cancer progression, as determined by PSA recurrence.12 Our previous study had a smaller sample size and shorter follow-up, and used whole-mount prostatectomy specimens.12 The methodologies of the current study include a larger sample size and handling of the prostatectomy specimens in a manner more applicable to standard practice. One of the weaknesses of the current study is that a partial sampling method was used; therefore, incomplete sampling could alter the results. It is unlikely that in most pathology practices the entire radical prostatectomy specimen would be submitted for pathologic evaluation, therefore making this study correlate more closely with the methodologies of most general pathology practices.

In the absence of Gleason pattern 4/5 cancer, no patients in the current study cohort died of prostate cancer. The excellent prognosis in these patients suggests that our partial sampling method is adequate. If any significant portion of Gleason pattern 4/5 cancer were missed, it is likely that this oversight would have been demonstrated by diminished outcome. We should also emphasize that tumor characteristics have changed since the mid-1990s. Early detection of prostate cancer could have an impact on the assessment of the percent Gleason pattern 4/5 cancer. Nonetheless, our recent study using modern series (from 1999–2003) indicates that evaluation of the percent Gleason pattern 4/5 is still relevant.12 Furthermore, only a small number of patients died of prostate cancer. However, caution is warranted interpreting our results because some variables that were inconclusive due to the limited statistical power may attain statistical significance if the sample size and number of events were increased.

The Gleason score is a powerful predictor of patient outcome. One of the limitations of the Gleason scoring system is that the majority of prostate carcinomas have Gleason scores of 6 or 7. In the current study, 81% of prostatectomy specimens had an overall Gleason score of ≤7. Prostate cancer is a multifocal and heterogeneous disease with diverse histopathologic patterns.13, 30–32 In evaluating a series of whole-mounted radical prostatectomies, Aihara et al.30 found that greater than half contained ≥3 different Gleason patterns. Recent studies have found that the volume of tumor present in biopsies does not correlate with the postoperative radical prostatectomy Gleason score, emphasizing the significance of small-volume, high-grade elements.12, 33–41 Other factors, such as serum PSA levels, clinical stage, the findings of digital rectal examination, and perineural invasion status may help further stratify the risk of biologically aggressive tumor.

In conclusion, the quantification of high-grade prostate cancer provides additional prognostic information beyond the conventional Gleason score and can be readily performed in routine pathology practice. We recommend that the percent Gleason pattern 4/5 be routinely reported for radical prostatectomy specimens.

Ancillary