Combined-Modality Staging for Localized Adenocarcinoma of the Prostate
Combined-Modality Staging for Localized Adenocarcinoma of the Prostate
ABSTRACT: The goal of identifying a set of pretreatment risk-stratifying factors for patients with localized prostate cancer is to be able to individualize treatment and optimize patient selection for clinical trials. Low-risk patients are most likely to remain disease-free following local therapy, whereas high-risk patients are not. By identifying a high-risk group in which approximately two-thirds of the patients fail biochemically within 10 years following local therapy, the ability to detect a difference in outcome as a result of additional therapy (if any difference exists) is maximized. Specifically, small improvements in outcome associated with the use of novel therapies can be measured more quickly in the high-risk population because of the intrinsically higher failure and progression rates. Therefore, novel therapies should be studied first in high-risk populations using a prospective randomized trial design (ie, local therapy with or without novel therapy). If efficacy is established in the high-risk cohort, then testing could be considered in the low-risk population. This review will demonstrate the ability to risk stratify more than 90% of all patients with clinically localized prostate cancer into low- or high-risk groups for biochemical progression following local therapy using four readily available pretreatment clinical parameters.
Combined-modality staging offers a method of predicting prostate-specific antigen (PSA) failure-free survival following either radical prostatectomy or conventional-dose (ie, 70 Gy) three-dimensional (3D) conformal external-beam radiation therapy (CRT) in patients with clinically localized prostate cancer. This methodology identifies the set of independent pretreatment clinical predictors of PSA outcome in order to categorize patients who are likely (ie, at low risk) or unlikely (ie, at high risk) to achieve long-term cancer control after radical prostatectomy or conventional-dose 3D CRT. Using these risk groups as the baseline provides a framework for ascertaining whether a new test can further stratify PSA outcome beyond that already provided by the established predictors. This review will direct particular attention to patients in the intermediate-risk group, who comprise approximately one-third of all patients with localized prostate cancer and for whom better prediction of PSA outcome is most needed.
PSA-era studies[2-17] evaluating men diagnosed with prostate cancer arising in the peripheral zone of the prostate gland have shown that the PSA level, biopsy Gleason score, and 1992 American Joint Committee on Cancer (AJCC) clinical T stage provide independent information on PSA outcome following local therapy. Based on a review of the literature, three risk groups can be defined as follows:
Low-risk group: approximately 80% 10-year PSA failure-free survival, 1992 AJCC clinical stage T1c,2a, and PSA ≤ 10 ng/mL, and biopsy Gleason score ≤ 6;
Intermediate-risk group: approximately 50% 10-year PSA failure-free survival, 1992 AJCC clinical stage T2b, or PSA > 10 and ≤ 20 ng/mL, or biopsy Gleason score 7;
High-risk group: approximately 33% 10-year PSA failure-free survival, 1992 AJCC stage T2c disease, or PSA > 20 ng/mL, or biopsy Gleason score ³ 8.
Figures 1, 2, and 3 illustrate the ability of this stratification to provide three distinct PSA outcome groups following radical prostatectomy or 3D CRT. The 1992 (rather than 1997) AJCC clinical T stage was used for the purpose of this review. Previous investigators have shown a more clinically significant stratification of PSA outcome after radical prostatectomy for patients grouped using the 1992, compared to the 1997, AJCC clinical T category.
Not included in the risk categorization are patients who present with bladder outlet obstructive symptoms (T1a,b) and are diagnosed on the basis of a transurethral resection of the prostate, and men who are diagnosed solely on the basis of a transition zone biopsy. These patients have a less common form of prostate cancer that usually arises in the transition zone and comprises less than 10% of all cases. Typically these cancers are not palpable on digital rectal examination, and patients have no grade 4 or 5 disease in the transurethral resection or transition zone biopsy specimen but often have a PSA that exceeds 20 ng/mL.
Nevertheless, the majority of these cancers are organ-confined and moderately well differentiated. As a result, patients with this type of prostate cancer have a PSA outcome that is similar to that of low-risk patients with peripheral zone cancers following radical prostatectomy or radiation therapy, despite the high pretreatment PSA level.
The pecentage of prostate biopsies found to contain prostate cancer is information that is readily available to all patients with PSA-detected or clinically palpable prostate cancer and is defined as follows:
number of positive biopsies/number of biopsies obtained x 100
Studies investigating the ability of the percentage of positive biopsies to predict pathologic end points after radical prostatectomy suggest a role for this clinical factor in predicting tumor volume, extracapsular extension,[22-25] seminal vesicle invasion, lymph node involvement, and the percentage of Gleason grade 4 and 5 disease in the radical prostatectomy specimen. In one large study (N = 813), Narayan and colleagues performed a logistic regression multivariable analysis and found independent prognostic significance for the clinical T stage, PSA, biopsy Gleason score, and the percentage of positive prostate biopsies in predicting extracapsular extension, seminal vesicle invasion, and positive lymph nodes.
Recognizing the ability of the percentage of positive biopsies to predict pathologic findings at radical prostatectomy, Presti and colleagues asked whether this factor added clinically relevant information that was not provided by the known prognostic factors used for predicting PSA outcome following radical prostatectomy. They performed a Cox regression multivariable analysis evaluating the ability of the percentage of positive biopsies (£ 50% vs > 50%), pretreatment PSA level (£ 20 ng/mL vs > 20 ng/mL), and biopsy Gleason score (2-6 vs 7-10) to predict time to postoperative PSA failure. They included 109 patients in their analysis, spanning all three risk groups, and found that both the Gleason score and percentage of positive biopsies were predictive of PSA outcome following radical prostatectomy.
Subsequently, the question of whether the percentage of positive prostate biopsies provides additional clinical information regarding PSA outcome after definitive local therapy, specifically for men in the intermediate-risk group, was investigated using a radical prostatectomy database from the Hospital of the University of Pennsylvania (HUP). An additional radical prostatectomy database from Brigham and Women’s Hospital (BWH) and a 3D CRT database from the Joint Center for Radiation Therapy (JCRT) were also used to validate or refute the results generated from the HUP database. In this study, a specific categorization of the positive biopsy data was selected for evaluation prior to analysis, in order to be able to apply the results to individual cases. For a standard sextant sampling, the categories selected were 1-2 (< 34% of biopsies positive), 3 (34% to 50% positive), or 4-6 (> 50% positive).
The study found that the majority of patients (214/269 [80%]) in the intermediate-risk group in the HUP radical prostatectomy database could be stratified into either a low- or high-risk cohort using the preoperative prostate biopsy data. These findings were then independently validated using the other radical prostatectomy and 3D CRT databases. Specifically, 251 (or 78%) of the 322 intermediate-risk patients in the BWH radical prostatectomy validation group could be classified as either low or high risk using the preoperative prostate biopsy data. Recently, updated data showing the stratification of PSA outcome obtained when applying the percent positive biopsy information to the intermediate-risk cohort are shown in Figures 4 and 5 for HUP and BWH patients, respectively.
The majority of the 3D CRT-managed patients (158/207 [76%]) in the intermediate-risk group could also be classified as either low or high risk using the preoperative prostate biopsy data, as depicted in Figure 6. Therefore, using the percent positive biopsy categories of < 34% and > 50%, between 76% and 80% of the intermediate-risk patients could be classified as low or high risk across the radical prostatectomy and radiation therapy databases. This improved stratification of PSA outcome in the intermediate-risk patient cohorts resulted in the overall classification of 94%, 91%, and 90% of the HUP, BWH, and JCRT patients, respectively, into low- or high-risk cohorts. Thus, using four readily available pretreatment clinical factors left approximately 10% of all patients with localized prostate cancer without a definitive risk categorization (ie, low or high) following radical prostatectomy or conventional-dose 3D CRT.
Some questions remain, however. First, extended core biopsies (12 to 16 samples) are now often obtained to decrease the known sampling error associated with a sextant sampling, particularly for patients with enlarged prostate glands. (Despite a PSA level greater than 4 and less than 10 ng/mL, approximately two-thirds of patients will have an initial negative sextant prostate biopsy. However, on subsequent biopsy, up to one-half of patients with an initial negative sextant sampling will be found to be positive.[33-37]) Given that 65% of all patients in this study had a sextant sampling and only 7% had 10 or more cores sampled, the validity of these data when extended core biopsies are obtained needs to be studied.
In addition, many pathologists report the percentage of each biopsy core involved with adenocarcinoma. Whether a more complete stratification of PSA outcome in the intermediate-risk patient cohort (ie, > 80%) can be provided by taking into account the percentage of each core involved, as opposed to simply scoring a core as positive or negative, also needs to be studied.