- Email: [email protected]
Weighted Gleason Grade Group (WGGG): A new prostate cancer biopsy reporting system with prognostic potential
ARTICLE IN PRESS
Urologic Oncology: Seminars and Original Investigations 000 (2019) 1−7
Clinical-Prostate cancer
Weighted Gleason Grade Group (WGGG): A new prostate cancer biopsy reporting system with prognostic potential Nikhil Waingankar, M.D. M.S.H.P.a,*, Alberto Martini, M.D.a, Luke Griffiths, M.D.b, Paras Shah, M.D.c, David J. Paulucci, M.S.a, Srinath Kotamarti, M.D.d, Zeynep Gul, M.D.a, Matthew Elmasri, B.S.b, Oksana Yaskiv, M.D.b, Kenneth Haines, M.D.a, Seth Lerner, M.D.a,e, Manish Vira, M.D.b, Louis R. Kavoussi, M.D., M.B.A.b, Ashutosh K. Tewari, M.D.a, Deepak A. Kapoor, M.D.a,e, Carl A. Olsson, M.D.a,e,f a
b
Icahn School of Medicine, The Mount Sinai Hospital, New York, NY Hofstra School of Medicine, Smith Institute for Urology, New Hyde Park, NY c The Mayo Clinic, Rochester, MN d Maimonides Medical Center, Brooklyn, NY e Integrated Medical Professionals, New York, NY f Columbia University Medical Center, New York, NY
Received 10 June 2019; received in revised form 9 October 2019; accepted 18 October 2019
Abstract Introduction: Presently, prostate biopsy (PBx) results report the highest Gleason Grade Group (GGG) as a single metric that gauges the overall clinical aggressiveness of cancer and dictates treatment. We hypothesized a PBx showing multiple cores of cancer with more volume cancer per core would represent more aggressive disease. We propose the Weighted Gleason Grade Group (WGGG), a novel scoring system that synthesizes all histopathologic data and cancer volume into a single numeric value representing the entire PBx, allowing for improved prediction of adverse pathology and risk of biochemical recurrence (BCR) following radical prostatectomy (RP). Methods: We studied 171 men who underwent RP after standard PBx. The WGGG was calculated by summing each positive core using the formula: GGG + (GGG x %Ca/core). RP pathology was evaluated for extraprostatic extension (EPE), positive surgical margins (PSM), seminal vesicle invasion (SVI), and lymph node involvement (LNI), and patients were followed for BCR. We compared GGG vs. WGGG receiver operating characteristic curves for each outcome, and determined the predictive capability of GGG and WGGG to identify patients with BCR. Categorized WGGG groups were created based on risk of BCR using classification and regression tree analysis. We then sought to externally validate WGGG in a cohort of 389 patients in a separate institutional dataset. Results: In the development cohort, area under the curves (AUCs) for the WGGG vs. GGG were significantly higher for predicting EPE (0.784 vs. 0.690, P = 0.002), SVI (AUC 0.823 vs. 0.721, P = .014), LNI (AUC 0.862 vs. 0.823, P = 0.039), and PSM (AUC 0.638 vs. 0.575, P = 0.031. Analysis of the validation cohort showed similar findings for EPE (AUC 0.764 vs. 0.729, P = 0.13), SVI (AUC 0.819 vs. 0.749, P = 0.01), LNI (AUC 0.939 vs. 0.867, P = 0.02), and PSM (AUC 0.624 vs. 0.547, P = 0.04). Patients with WGGG >30 (high-risk group) demonstrated »50% failure at 2 years in both cohorts. Conclusions: The WGGG, by providing a metric reflecting the entirety of the PBx, is more informative than conventional single GGG alone in identifying adverse pathologic outcomes and risk of BCR following RP. This superior discriminatory capability has been achieved without any consideration of other commonly available clinical disease characteristics. Ó 2019 Elsevier Inc. All rights reserved.
Keywords: Prostate; Biopsy; Gleason; Grade; GGG; Biochemical recurrence; Prostate cancer Abbreviations: AUC, area under the curve; BCR, biochemical recurrence; EPE, extraprostatic extension; GGG, Gleason Grade Group; LNI, lymph node involvement; PBx, prostate biopsy; PSM, positive surgical margins; ROC, receiver-operating characteristic; RP, radical prostatectomy; SVI, seminal vesicle invasion; WGGG, Weighted Gleason Grade Group
Research funding: None. *Corresponding author. Tel.: 405-590-9346. E-mail address: [email protected] (N. Waingankar). https://doi.org/10.1016/j.urolonc.2019.10.009 1078-1439/Ó 2019 Elsevier Inc. All rights reserved.
ARTICLE IN PRESS 2
N. Waingankar et al. / Urologic Oncology: Seminars and Original Investigations 00 (2019) 1−7
1. Introduction
2.2. Data
Prostate cancer is the most commonly diagnosed cancer in men with an estimated incidence of 164,690 new cases in 2018 [1]. Given the clinically indolent nature of a significant proportion of these cases, concern exists about the potential for over-diagnosis along with the associated harms of increased care intensity [2]. These concerns, however, must be balanced with the risks of under-diagnosis in order to properly classify patients who are appropriate for active surveillance versus those who will benefit from definitive treatment [3]. Indeed, among men who undergo prostate biopsy (PBx), rates of pathologic upgrading from biopsy to prostatectomy are as high as 54%, with up to 26% of men having a discrepancy of 2 or more Gleason grades [4]. In addition to prediction of the final pathology grade based on biopsy data, the ability to predict adverse pathology and risk of recurrence is central to shared decision-making. A number of clinical models and nomograms exist to aid clinicians with risk stratification of newly diagnosed prostate cancer patients [5−10]. While many of these tools are informative, they may be limited by a fundamental flaw in contemporary reporting of PBx data. Currently, biopsy reports are communicated and incorporated into risk stratification models by utilizing only the highest Gleason grade or Gleason Grade Group (GGG) of cancer present [11]. This ubiquitous approach neglects both the grade and volume of cancer present in other cores within the specimen, and therefore may be ignoring valuable data that may have an impact on outcomes. To address this potential limitation in pathology reporting, we developed the Weighted Gleason Grade Group (WGGG)—a novel scoring system that incorporates all grade and volume data into a single metric representative of the entire PBx specimen. The objective of our study is to describe the WGGG and to characterize its ability to predict adverse pathologic outcomes and biochemical recurrence among patients who undergo radical prostatectomy (RP) relative to that of the standard GGG.
Baseline demographic and clinical data included age, prostate specific antigen (PSA), prostate size, number of cores biopsied, number of positive cancer cores, and percent cancer per core. Final RP specimen primary covariates of interest included presence of adverse pathologic features such as extraprostatic extension (EPE), positive surgical margins (PSM), lymph node involvement (LNI), and seminal vesicle invasion (SVI). Patients were followed postoperatively for biochemical recurrence (BCR), defined as a PSA > 0.2 ng/ml.
2. Methods 2.1. Patients In the WGGG development cohort, we retrospectively reviewed data on patients from a single academic institution that were managed from January 2012 to January 2016. Patients were included if they had prostate cancer that was diagnosed by standard PBx that included 11 or more cores, and ultimately underwent treatment with RP. The validation cohort was composed of patients that were managed at a separate single academic institution from June 2014 to July 2017 with identical inclusion criteria.
2.3. Calculation of WGGG Each Gleason score was converted to a GGG using the method described by Epstein et al [11]. The WGGG was then calculated by summing the values of each positive core, normalized for a 12-core standard biopsy. Our calculation was performed using the formula: X ½GGG þ ðGGG x % Ca=coreÞ
2.4. Value of WGGG In a 12-core biopsy, the theoretical range of scores using this system is 0 (no cancer) to 120 (12 cores with 100% GGG 5). The utility of WGGG scoring is demonstrated in Fig. 1 which shows 2 separate patients, both of whom are described as having GGG 3 disease according to the new ISUP standards [11]. By using the above formula to derive our proposed WGGG score, the patient on the left has a WGGG score of 3.15, whereas the patient on the right has a WGGG score of 62.55. 2.5. Statistical analysis We created receiver-operating characteristic curves to assess the abilities of the GGG and WGGG to predict adverse pathologic features. Separate areas under the curve (AUC) were calculated for each grading system to predict EPE, SVI, PSM, and LNI. AUCs for GGG and WGGG were compared using DeLong testing. BCR-free survival was calculated using Kaplan Meier analysis for the entire cohort. To assess BCR-free survival function by WGGG score, we employed a classification and regression tree approach to identify the most meaningful cut points for the WGGG for predicting BCR. We then calculated BCR-free survival by GGG and categorized WGGG. Similar methods were employed for both the development and validation cohorts, and calibration plots were assessed to compare both cohorts in terms of risk of adverse pathology. All calculations were performed using Stata version 14 (StataCorp LLC, College Station, TX). A P value of <0.05 was considered to be statistically significant.
ARTICLE IN PRESS N. Waingankar et al. / Urologic Oncology: Seminars and Original Investigations 00 (2019) 1−7
3
Fig. 1. Sample calculation comparing GGG vs. WGGG. GGG = Gleason Grade Group; WGGG = Weighted Gleason Grade Group.
3. Results For the development cohort, our final analytic dataset included 171 patients. The median age was 61 years (interquartile range (IQR) 56−66). Median PSA was 5.9 ng/ml (IQR 4.6−9.0) and prostate size was 46.7 cc (IQR 40−59). On biopsy, 49 (29%), 68 (40%), 21 (12%), and 33 (19%) patients had GGG 1, 2, 3, and 4 to 5, respectively. Median WGGG was 10.9 (IQR 5.6−18.6). On final pathology, 67 (39%) patients had EPE, 19 (11%) had SVI, 10 (6%) had LNI, and 42 (25%) had PSM. The validation cohort included 389 patients with a median age of 62 years (IQR 57−67). Median PSA was 6.0 ng/ml (IQR 4.6−9.0) and prostate size was 36 cc (IQR 27−47). On biopsy, 89 (23%), 153 (39%), 65 (17%), and 82 (21%) had GGG 1, 2, 3, and 4 to 5, respectively. Median WGGG was 8.1 (IQR 4.0−16.6). On final pathology, 122 (31%) of patients had EPE, 47 (12%) had SVI, 12 (3%) had LNI, and 30 (8%) had PSM. Complete baseline clinical and pathologic data for both cohorts are listed in Table 1. Analysis and comparison of receiver-operating characteristic curves in the development cohort indicated that WGGG outperformed GGG for all outcomes of interest. The AUC for WGGG vs. GGG was significantly higher for predicting EPE (AUC 0.784 vs. 0.690, P = 0.002), SVI (AUC 0.823 vs. 0.721, P = 0.014), LNI (AUC 0.862 vs 0.823, P = 0.039), and PSM (AUC 0.638 vs. 0.575, P = 0.031; Fig. 2). The predictive capability of WGGG was confirmed in the validation cohort for EPE (AUC 0.764 vs 0.729, P = 0.13), SVI (AUC 0.819 vs. 0.749, P = 0.01), LNI (AUC 0.939 vs 0.867, P = 0.02), and PSM (AUC 0.624 vs 0.547, P = 0.04; Fig. 3). Calibration plots demonstrated shared similarities between the development and validation cohorts in terms of risk for having EPE, SVI, and LNI, but not PSM (Supplemental Fig. 1).
Median follow-up for the development cohort was 19.0 months (IQR 9−28), during which 32/171 patients (18.7%) had BCR. BCR-free survival at 24 months was 81% (95% CI 73−87%). Nodes from the classification and regression tree model (Supplemental Fig. 2) indicated 2 distinct cut points resulting in 3 WGGG risk groups, with 57 patients (33.3%) in group 1 (≤7, low risk), 92 (53.8%) in group 2 (8−29, intermediate risk), and 22 (12.9%) in group 3 (≥30, high risk). For both GGG and WGGG models, incremental increases in the Table 1 Baseline clinical and pathologic data for development and validation cohorts Development cohort (n = 171) Age at surgery, y 60.9 (IQR 55.9−66.3) PSA 5.9 (IQR 4.6−9.0) Prostate size 46.7cc (IQR 40−59) Biopsy GGG, n (%) 1 49 (29%) 2 68 (40%) 3 21 (12%) 4−5 33 (19%) Weighted GGG 10.9 (IQR 5.6−18.6) Extraprostatic extension, n (%) Absent 104 (61%) Present 67 (39%) Seminal vesicle invasion, n (%) Absent 152 (89%) Present 19 (11%) Lymph node involvement, n (%) Absent 161 (94%) Present 10 (6%) Surgical margins status, n (%) Negative 129 (75%) Positive 42 (25%) GGG = Gleason Grade Group.
Validation cohort (n = 389) 62 (IQR 57−67) 6.0 (IQR 4.6−9.0) 36cc (IQR 27−47) 89 (23%) 153 (39%) 65 (17%) 82 (21%) 8.1 (IQR 4.0−16.6) 267 (69%) 122 (31%) 342 (88%) 47 (12%) 377 (97%) 12 (3%) 359 (92%) 30 (8%)
ARTICLE IN PRESS N. Waingankar et al. / Urologic Oncology: Seminars and Original Investigations 00 (2019) 1−7
4
Fig. 2. ROC analysis of the development cohort. ROC = receiver operating characteristics.
Fig. 3. ROC analysis of the validation cohort. ROC = receiver operating characteristics.
grade and weighted grade of cancer were associated with increased hazard of BCR (Table 2, Fig. 4). In the development cohort, when stratifying patients according to the WGGG risk groups, we found that the 2-year BCR-free
survival rates were: 96%, 78%, and 53% for the low, intermediate, and high-risk group, respectively. Median follow-up in the validation cohort was 12.3 months (IQR 6−23) during which 46/334 (13.4%; 55 with
Table 2 Impact of incremental increases in GGG and WGGG groups on hazard of BCR Development cohort Covariate GGG 1 2 3 4−5 WGGG groups 1 2 3
HR
95% CI
Validation cohort P value
Ref. 2.26 3.59 7.33
0.7−7.21 0.88−14.65 2.41−22.34
0.17 0.075 <0.001
Ref. 5.05 16.96
1.49−17.11 4.57−62.99
0.009 <0.001
Covariate GGG 1 2 3 4−5 WGGG groups 1 2 3
BCR = biochemical recurrence; GGG = Gleason Grade Group; WGGG = Weighted Gleason Grade Group.
HR
95% CI
P value
Ref. 3.00 9.82 15.49
0.65−13.89 2.23−43.24 3.63−66.14
0.16 0.003 <0.001
Ref. 1.95 8.43
0.93−4.11 3.69−19.28
0.078 <0.001
ARTICLE IN PRESS N. Waingankar et al. / Urologic Oncology: Seminars and Original Investigations 00 (2019) 1−7
Fig. 4. BCR-free survival by WGGG risk group in the development cohort. BCR = biochemical recurrence; WGGG = Weighted Gleason Grade Group.
missing data) had BCR. BCR-free survival at 24 months was 83% (95% CI 77%−87%). The same 3 WGGG cutpoints were applied in this population: 166 (42.7%) patients were in group 1, 184 (47.3%) in group 2, and 49 (10%) in group 3; these groups maintained their prognostic discrimination for BCR (Table 2, Fig. 5). In this cohort, when stratifying patients according to the WGGG risk groups, we found the 2-year BCR-free survival rates to be: 91%, 82%, and 47% for the low, intermediate, and high-risk group, respectively. C-indices of the WGGG and GGG models for BCR were 71 and 68, respectively, for the development cohort, and not calculated for the validation group due to paucity of mature follow-up data. 4. Discussion The Gleason score, while ubiquitously used to grade prostate cancer, has been limited by 3 factors: First, the communication of “low risk” disease is often unclear to patients when using a number that starts in the middle of a
Fig. 5. BCR-free survival by WGGG risk group in the validation cohort. BCR = biochemical recurrence; WGGG = Weighted Gleason Grade Group.
5
2 to 10 scale (i.e., Gleason 6 cancer), especially when initiating discussion on appropriateness of management with active surveillance [12]. Second, certain patterns historically regarded as Gleason 6 are now more commonly diagnosed as Gleason 7, leading to heterogeneity in prognoses among patients diagnosed over time with Gleason 6 cancers [13]. Finally, further heterogeneity in prognoses exists within the commonly stratified groups of low (Gleason 6), intermediate (Gleason 3 + 4 and 4 + 3), and high (Gleason 8, 9, and 10) risk patients [14]. In 2016, the World Health Organization adopted a new grading system that converted the classic Gleason score from a 2 to 10 scale, to a more easily interpreted and prognostically accurate 1 to 5 scale [11,15]. While the updated GGG system can further enhance prognostic capability when incorporated into clinical models, it is still limited by our incomplete, inefficient utilization of this data: by focusing only on the highest grade and highest volume core, we are ultimately leaving potentially relevant information “on the table.” Our technique incorporates all available pathology data from biopsy samples. Fig. 1 reflects this and clearly distinguishes GGG from our proposed WGGG, suggesting that a numerical scoring system may be used as a surrogate to reflect the aggressiveness of individual prostate cancer cases. This inclusion of all available data is consistent with the growing trend in other fields of medicine where previously bypassed data points are being combed for added diagnostic or prognostic value. For example, this has been the case in the field of Radiology, where CT scans are being re-evaluated using analytic morphometric techniques to identify specific pathology as well as to provide an objective metric to supplement the gestalt patient “eyeball test” (e.g., Psoas muscle area as a metric for frailty), thereby giving clinicians a more accurate global assessment of health and patient risk [16]. The same computational techniques are also being applied to the field of pathology to transform histologic patterns on a slide into pixels, lines, angles, and shapes with reproducible features that reliably identify histopathology, thus allowing subjective reads to be enhanced by objective analyses that utilize the entire slide. Until this so-called “precision pathology” becomes the standard of care, it is paramount that urologists utilize all readily available histopathologic data in our management of patients with prostate cancer. Currently, every PBx, regardless of scoring system used, provides 3 forms of data: (1) Gleason score or grade group, (2) number of cores positive for cancer, and (3) tumor volume per positive core. Failure to incorporate all 3 risks limiting prognostic capability and can potentially impact patient care. We demonstrated that by providing a metric reflecting the entirety of the PBx—including volume involved and grade of all positive cores—we can improve upon the prognostic value offered with utilization of a single metric as in the standard GGG. In our study, AUCs for identification of adverse pathology with GGG ranged from 0.575 to 0.823 in the development group and 0.547 to 0.867 in the validation
ARTICLE IN PRESS 6
N. Waingankar et al. / Urologic Oncology: Seminars and Original Investigations 00 (2019) 1−7
group. However, by considering the grades and volumes of all positive cores using the WGGG, these AUCs improved to 0.638 to 0.862 (development) and 0.624 to 0.929 (validation). Indeed, WGGG significantly outperformed GGG for identification of all adverse pathologic outcomes. These findings were externally validated in a well-calibrated, separate set of institutional data. Beyond the predictive association between increasing WGGG and risk of adverse pathology, we showed that WGGG alone carries prognostic value as well. Our study identified 3 distinct cutpoints in WGGG that stratify patients into risk of BCR. In the development cohort, we even identified a group of patients (group 3) for which a urologist could use WGGG to predict definitive biochemical failure with prostatectomy alone, and therefore potentially proactively augment management with a multimodal approach. These WGGG cutpoints were confirmed in the validation cohort, and while we did not identify a cutpoint above which we could definitively predict biochemical failure, this may have been secondary to the shorter follow-up time in the validation cohort. With these encouraging results, work is underway for identification of incremental net benefit of adding WGGG to clinical models. The primary strengths of our study are that patients were drawn from a large network of urologists at an institution that serves a major metropolitan area, and that our findings were externally validated. Patients were heterogeneous in demographic and clinical factors, and providers likely exhibited some degree of uncaptured variation of practice patterns. Finally, as the WGGG is derived from the GGG, it shares identical expected levels of precision, accuracy, and bias with the current gold standard, and therefore achieves criterion validity. Our study is limited by its retrospective design that uses data from a single institution for development and single institution for validation, along with lack of standardized biopsy protocol at both sites. All WGGGs were normalized for a 12-core biopsy to attempt to account for this. Other potential limitations include lack of centralized pathology review and standardized pathology reporting. However, this lack of standardization may more accurately capture the natural variation in pathology review and reporting experienced across various practices. Additionally, while WGGG and GGG did not have significantly different AUCs for identification of EPE in the validation set, this is likely because GGG performed better in the validation set (AUC 0.729) than it did in the development cohort (AUC 0.690); meanwhile, the AUCs for WGGG displayed similar predictive capability in both datasets. Furthermore, our use of 2year BCR as an outcome of interest is based on the recency of the dataset available to us. The necessary time horizon for identifying post-treatment outcomes should ideally be much longer given the typical duration of the prostate cancer disease course. Furthermore, while the validation dataset was well calibrated for most outcomes of interest, this was not the case for PSM; this is likely because PSM
reflects surgeon variation in addition to patient variation between the 2 groups. Nonetheless, discrimination was similar for PSM in both cohorts. As in any study of a novel diagnostic tool, applicability beyond the characteristics of the patient population studied must necessarily be subject to further validation. This includes the applicability of WGGG to patients who undergo MRI-TRUS fusion biopsy; our analysis included only patients with standard 12-core transrectal ultrasound-guided biopsy, and as such, our conclusions may not apply to those patients who undergo MRIguided fusion biopsy. A separate analysis is underway to determine the predictive and prognostic capability of WGGG among these patients. Early criticism of our proposed WGGG metric has emphasized that aggressiveness of cancer cannot accurately be characterized on a linear scale; yet, nearly all cancer grading is based on linear scales. This is particularly true in prostate cancer where the Gleason Grading system had traditionally been a scale of 2 to 10 based on pathologic and clinical behavior going hand-in-glove together. It evolved to a 2 to 6, 7, and 8 to 10 system, and further evolved to Gleason 7 being divided into 3 + 4 vs. 4 + 3. Finally, now we have a 1 to 5 scale, which combines grades 2 to 6 into 3 + 3 or GGG1, maintains the 3 + 4 (GGG2) and 4 + 3 (GGG3) distinction, and separates 4 + 4 (GGG4) from ≥9 (GGG5). With each evolution we have tried to reflect the Gleason score numerically; so it is with WGGG. Is it perfect? In terms of test discrimination, we have demonstrated that it performs well without the use of any clinical data like PSA, PSA density, or clinical stage. Most importantly, our proposed WGGG costs nothing except the time and effort for the urologist to collect and assess the data that has already been provided. 5. Conclusions The WGGG provides a single metric that captures the true disease potential of the entire prostate. Our externally validated findings indicate that WGGG carries more accurate predictive capability relative to the standard GGG for identification of adverse pathology and prognostic capability for identifying patients at risk for BCR. Conflicts of interest None. Supplementary materials Supplementary material associated with this article can be found in the online version at https://doi.org/10.1016/j. urolonc.2019.10.009. References [1] American Cancer Society. Cancer facts and figures 2018. Atlanta, GA: American Cancer Society; 2018.
ARTICLE IN PRESS N. Waingankar et al. / Urologic Oncology: Seminars and Original Investigations 00 (2019) 1−7 [2] United States Preventative Services Task Force Draft Recommendation Statement. Prostate Cancer Screening. 2017. [3] Graif T, Loeb S, Roehl KA, et al. Under diagnosis and over diagnosis of prostate cancer. J Urol 2007;178:88–92. [4] Cookson MS, Fleshner NE, Soloway SM, et al. Correlation between Gleason score of needle biopsy and radical prostatectomy specimen: accuracy and clinical implications. J Urol 1997;157: 559–62. [5] Baccala A Jr., Reuther AM, Bianco FJ Jr., et al. Complete resection of seminal vesicles at radical prostatectomy results in substantial long-term disease-free survival: multi-institutional study of 6740 patients. Urology 2007;69:536–40. [6] Cagiannos I, Karakiewicz P, Eastham JA, et al. A preoperative nomogram identifying decreased risk of positive pelvic lymph nodes in patients with prostate cancer. J Urol 2003;170:1798–803. [7] Chun FK, Steuber T, Erbersdobler A, et al. Development and internal validation of a nomogram predicting the probability of prostate cancer Gleason sum upgrading between biopsy and radical prostatectomy pathology. Eur Urol 2006;49:820–6. [8] Kattan MW, Eastham JA, Wheeler TM, et al. Counseling men with prostate cancer: a nomogram for predicting the presence of small, moderately differentiated, confined tumors. J Urol 2003;170:1792–7.
7
[9] Partin AW, Yoo J, Carter HB, et al. The use of prostate specific antigen, clinical stage and Gleason score to predict pathological stage in men with localized prostate cancer. J Urol 1993;150:110–4. [10] Steuber T, Graefen M, Haese A, et al. Validation of a nomogram for prediction of side specific extracapsular extension at radical prostatectomy. J Urol 2006;175:939–44:discussion 944. [11] Epstein JI, Zelefsky MJ, Sjoberg DD, et al. A Contemporary prostate cancer grading system: a validated alternative to the Gleason Score. Eur Urol 2016;69:428–35. [12] Nelson JB. Observation for clinically localized prostate cancer. J Clin Oncol 2014;32:1295–8. [13] Dong F, Wang C, Farris AB, et al. Impact on the clinical outcome of prostate cancer by the 2005 international society of urological pathology modified Gleason grading system. Am J Surg Pathol 2012;36:838–43. [14] Tsao CK, Gray KP, Nakabayashi M, et al. Patients with biopsy Gleason 9 and 10 prostate cancer have significantly worse outcomes compared to patients with Gleason 8 disease. J Urol 2015;194:91–7. [15] Pierorazio PM, Walsh PC, Partin AW, et al. Prognostic Gleason grade grouping: data based on the modified Gleason scoring system. BJU Int 2013;111:753–60. [16] Englesbe MJ, Lee JS, He K, et al. Analytic morphomics, core muscle size, and surgical outcomes. Ann Surg 2012;256:255–61.
Urologic Oncology: Seminars and Original Investigations 000 (2019) 1−7
Clinical-Prostate cancer
Weighted Gleason Grade Group (WGGG): A new prostate cancer biopsy reporting system with prognostic potential Nikhil Waingankar, M.D. M.S.H.P.a,*, Alberto Martini, M.D.a, Luke Griffiths, M.D.b, Paras Shah, M.D.c, David J. Paulucci, M.S.a, Srinath Kotamarti, M.D.d, Zeynep Gul, M.D.a, Matthew Elmasri, B.S.b, Oksana Yaskiv, M.D.b, Kenneth Haines, M.D.a, Seth Lerner, M.D.a,e, Manish Vira, M.D.b, Louis R. Kavoussi, M.D., M.B.A.b, Ashutosh K. Tewari, M.D.a, Deepak A. Kapoor, M.D.a,e, Carl A. Olsson, M.D.a,e,f a
b
Icahn School of Medicine, The Mount Sinai Hospital, New York, NY Hofstra School of Medicine, Smith Institute for Urology, New Hyde Park, NY c The Mayo Clinic, Rochester, MN d Maimonides Medical Center, Brooklyn, NY e Integrated Medical Professionals, New York, NY f Columbia University Medical Center, New York, NY
Received 10 June 2019; received in revised form 9 October 2019; accepted 18 October 2019
Abstract Introduction: Presently, prostate biopsy (PBx) results report the highest Gleason Grade Group (GGG) as a single metric that gauges the overall clinical aggressiveness of cancer and dictates treatment. We hypothesized a PBx showing multiple cores of cancer with more volume cancer per core would represent more aggressive disease. We propose the Weighted Gleason Grade Group (WGGG), a novel scoring system that synthesizes all histopathologic data and cancer volume into a single numeric value representing the entire PBx, allowing for improved prediction of adverse pathology and risk of biochemical recurrence (BCR) following radical prostatectomy (RP). Methods: We studied 171 men who underwent RP after standard PBx. The WGGG was calculated by summing each positive core using the formula: GGG + (GGG x %Ca/core). RP pathology was evaluated for extraprostatic extension (EPE), positive surgical margins (PSM), seminal vesicle invasion (SVI), and lymph node involvement (LNI), and patients were followed for BCR. We compared GGG vs. WGGG receiver operating characteristic curves for each outcome, and determined the predictive capability of GGG and WGGG to identify patients with BCR. Categorized WGGG groups were created based on risk of BCR using classification and regression tree analysis. We then sought to externally validate WGGG in a cohort of 389 patients in a separate institutional dataset. Results: In the development cohort, area under the curves (AUCs) for the WGGG vs. GGG were significantly higher for predicting EPE (0.784 vs. 0.690, P = 0.002), SVI (AUC 0.823 vs. 0.721, P = .014), LNI (AUC 0.862 vs. 0.823, P = 0.039), and PSM (AUC 0.638 vs. 0.575, P = 0.031. Analysis of the validation cohort showed similar findings for EPE (AUC 0.764 vs. 0.729, P = 0.13), SVI (AUC 0.819 vs. 0.749, P = 0.01), LNI (AUC 0.939 vs. 0.867, P = 0.02), and PSM (AUC 0.624 vs. 0.547, P = 0.04). Patients with WGGG >30 (high-risk group) demonstrated »50% failure at 2 years in both cohorts. Conclusions: The WGGG, by providing a metric reflecting the entirety of the PBx, is more informative than conventional single GGG alone in identifying adverse pathologic outcomes and risk of BCR following RP. This superior discriminatory capability has been achieved without any consideration of other commonly available clinical disease characteristics. Ó 2019 Elsevier Inc. All rights reserved.
Keywords: Prostate; Biopsy; Gleason; Grade; GGG; Biochemical recurrence; Prostate cancer Abbreviations: AUC, area under the curve; BCR, biochemical recurrence; EPE, extraprostatic extension; GGG, Gleason Grade Group; LNI, lymph node involvement; PBx, prostate biopsy; PSM, positive surgical margins; ROC, receiver-operating characteristic; RP, radical prostatectomy; SVI, seminal vesicle invasion; WGGG, Weighted Gleason Grade Group
Research funding: None. *Corresponding author. Tel.: 405-590-9346. E-mail address: [email protected] (N. Waingankar). https://doi.org/10.1016/j.urolonc.2019.10.009 1078-1439/Ó 2019 Elsevier Inc. All rights reserved.
ARTICLE IN PRESS 2
N. Waingankar et al. / Urologic Oncology: Seminars and Original Investigations 00 (2019) 1−7
1. Introduction
2.2. Data
Prostate cancer is the most commonly diagnosed cancer in men with an estimated incidence of 164,690 new cases in 2018 [1]. Given the clinically indolent nature of a significant proportion of these cases, concern exists about the potential for over-diagnosis along with the associated harms of increased care intensity [2]. These concerns, however, must be balanced with the risks of under-diagnosis in order to properly classify patients who are appropriate for active surveillance versus those who will benefit from definitive treatment [3]. Indeed, among men who undergo prostate biopsy (PBx), rates of pathologic upgrading from biopsy to prostatectomy are as high as 54%, with up to 26% of men having a discrepancy of 2 or more Gleason grades [4]. In addition to prediction of the final pathology grade based on biopsy data, the ability to predict adverse pathology and risk of recurrence is central to shared decision-making. A number of clinical models and nomograms exist to aid clinicians with risk stratification of newly diagnosed prostate cancer patients [5−10]. While many of these tools are informative, they may be limited by a fundamental flaw in contemporary reporting of PBx data. Currently, biopsy reports are communicated and incorporated into risk stratification models by utilizing only the highest Gleason grade or Gleason Grade Group (GGG) of cancer present [11]. This ubiquitous approach neglects both the grade and volume of cancer present in other cores within the specimen, and therefore may be ignoring valuable data that may have an impact on outcomes. To address this potential limitation in pathology reporting, we developed the Weighted Gleason Grade Group (WGGG)—a novel scoring system that incorporates all grade and volume data into a single metric representative of the entire PBx specimen. The objective of our study is to describe the WGGG and to characterize its ability to predict adverse pathologic outcomes and biochemical recurrence among patients who undergo radical prostatectomy (RP) relative to that of the standard GGG.
Baseline demographic and clinical data included age, prostate specific antigen (PSA), prostate size, number of cores biopsied, number of positive cancer cores, and percent cancer per core. Final RP specimen primary covariates of interest included presence of adverse pathologic features such as extraprostatic extension (EPE), positive surgical margins (PSM), lymph node involvement (LNI), and seminal vesicle invasion (SVI). Patients were followed postoperatively for biochemical recurrence (BCR), defined as a PSA > 0.2 ng/ml.
2. Methods 2.1. Patients In the WGGG development cohort, we retrospectively reviewed data on patients from a single academic institution that were managed from January 2012 to January 2016. Patients were included if they had prostate cancer that was diagnosed by standard PBx that included 11 or more cores, and ultimately underwent treatment with RP. The validation cohort was composed of patients that were managed at a separate single academic institution from June 2014 to July 2017 with identical inclusion criteria.
2.3. Calculation of WGGG Each Gleason score was converted to a GGG using the method described by Epstein et al [11]. The WGGG was then calculated by summing the values of each positive core, normalized for a 12-core standard biopsy. Our calculation was performed using the formula: X ½GGG þ ðGGG x % Ca=coreÞ
2.4. Value of WGGG In a 12-core biopsy, the theoretical range of scores using this system is 0 (no cancer) to 120 (12 cores with 100% GGG 5). The utility of WGGG scoring is demonstrated in Fig. 1 which shows 2 separate patients, both of whom are described as having GGG 3 disease according to the new ISUP standards [11]. By using the above formula to derive our proposed WGGG score, the patient on the left has a WGGG score of 3.15, whereas the patient on the right has a WGGG score of 62.55. 2.5. Statistical analysis We created receiver-operating characteristic curves to assess the abilities of the GGG and WGGG to predict adverse pathologic features. Separate areas under the curve (AUC) were calculated for each grading system to predict EPE, SVI, PSM, and LNI. AUCs for GGG and WGGG were compared using DeLong testing. BCR-free survival was calculated using Kaplan Meier analysis for the entire cohort. To assess BCR-free survival function by WGGG score, we employed a classification and regression tree approach to identify the most meaningful cut points for the WGGG for predicting BCR. We then calculated BCR-free survival by GGG and categorized WGGG. Similar methods were employed for both the development and validation cohorts, and calibration plots were assessed to compare both cohorts in terms of risk of adverse pathology. All calculations were performed using Stata version 14 (StataCorp LLC, College Station, TX). A P value of <0.05 was considered to be statistically significant.
ARTICLE IN PRESS N. Waingankar et al. / Urologic Oncology: Seminars and Original Investigations 00 (2019) 1−7
3
Fig. 1. Sample calculation comparing GGG vs. WGGG. GGG = Gleason Grade Group; WGGG = Weighted Gleason Grade Group.
3. Results For the development cohort, our final analytic dataset included 171 patients. The median age was 61 years (interquartile range (IQR) 56−66). Median PSA was 5.9 ng/ml (IQR 4.6−9.0) and prostate size was 46.7 cc (IQR 40−59). On biopsy, 49 (29%), 68 (40%), 21 (12%), and 33 (19%) patients had GGG 1, 2, 3, and 4 to 5, respectively. Median WGGG was 10.9 (IQR 5.6−18.6). On final pathology, 67 (39%) patients had EPE, 19 (11%) had SVI, 10 (6%) had LNI, and 42 (25%) had PSM. The validation cohort included 389 patients with a median age of 62 years (IQR 57−67). Median PSA was 6.0 ng/ml (IQR 4.6−9.0) and prostate size was 36 cc (IQR 27−47). On biopsy, 89 (23%), 153 (39%), 65 (17%), and 82 (21%) had GGG 1, 2, 3, and 4 to 5, respectively. Median WGGG was 8.1 (IQR 4.0−16.6). On final pathology, 122 (31%) of patients had EPE, 47 (12%) had SVI, 12 (3%) had LNI, and 30 (8%) had PSM. Complete baseline clinical and pathologic data for both cohorts are listed in Table 1. Analysis and comparison of receiver-operating characteristic curves in the development cohort indicated that WGGG outperformed GGG for all outcomes of interest. The AUC for WGGG vs. GGG was significantly higher for predicting EPE (AUC 0.784 vs. 0.690, P = 0.002), SVI (AUC 0.823 vs. 0.721, P = 0.014), LNI (AUC 0.862 vs 0.823, P = 0.039), and PSM (AUC 0.638 vs. 0.575, P = 0.031; Fig. 2). The predictive capability of WGGG was confirmed in the validation cohort for EPE (AUC 0.764 vs 0.729, P = 0.13), SVI (AUC 0.819 vs. 0.749, P = 0.01), LNI (AUC 0.939 vs 0.867, P = 0.02), and PSM (AUC 0.624 vs 0.547, P = 0.04; Fig. 3). Calibration plots demonstrated shared similarities between the development and validation cohorts in terms of risk for having EPE, SVI, and LNI, but not PSM (Supplemental Fig. 1).
Median follow-up for the development cohort was 19.0 months (IQR 9−28), during which 32/171 patients (18.7%) had BCR. BCR-free survival at 24 months was 81% (95% CI 73−87%). Nodes from the classification and regression tree model (Supplemental Fig. 2) indicated 2 distinct cut points resulting in 3 WGGG risk groups, with 57 patients (33.3%) in group 1 (≤7, low risk), 92 (53.8%) in group 2 (8−29, intermediate risk), and 22 (12.9%) in group 3 (≥30, high risk). For both GGG and WGGG models, incremental increases in the Table 1 Baseline clinical and pathologic data for development and validation cohorts Development cohort (n = 171) Age at surgery, y 60.9 (IQR 55.9−66.3) PSA 5.9 (IQR 4.6−9.0) Prostate size 46.7cc (IQR 40−59) Biopsy GGG, n (%) 1 49 (29%) 2 68 (40%) 3 21 (12%) 4−5 33 (19%) Weighted GGG 10.9 (IQR 5.6−18.6) Extraprostatic extension, n (%) Absent 104 (61%) Present 67 (39%) Seminal vesicle invasion, n (%) Absent 152 (89%) Present 19 (11%) Lymph node involvement, n (%) Absent 161 (94%) Present 10 (6%) Surgical margins status, n (%) Negative 129 (75%) Positive 42 (25%) GGG = Gleason Grade Group.
Validation cohort (n = 389) 62 (IQR 57−67) 6.0 (IQR 4.6−9.0) 36cc (IQR 27−47) 89 (23%) 153 (39%) 65 (17%) 82 (21%) 8.1 (IQR 4.0−16.6) 267 (69%) 122 (31%) 342 (88%) 47 (12%) 377 (97%) 12 (3%) 359 (92%) 30 (8%)
ARTICLE IN PRESS N. Waingankar et al. / Urologic Oncology: Seminars and Original Investigations 00 (2019) 1−7
4
Fig. 2. ROC analysis of the development cohort. ROC = receiver operating characteristics.
Fig. 3. ROC analysis of the validation cohort. ROC = receiver operating characteristics.
grade and weighted grade of cancer were associated with increased hazard of BCR (Table 2, Fig. 4). In the development cohort, when stratifying patients according to the WGGG risk groups, we found that the 2-year BCR-free
survival rates were: 96%, 78%, and 53% for the low, intermediate, and high-risk group, respectively. Median follow-up in the validation cohort was 12.3 months (IQR 6−23) during which 46/334 (13.4%; 55 with
Table 2 Impact of incremental increases in GGG and WGGG groups on hazard of BCR Development cohort Covariate GGG 1 2 3 4−5 WGGG groups 1 2 3
HR
95% CI
Validation cohort P value
Ref. 2.26 3.59 7.33
0.7−7.21 0.88−14.65 2.41−22.34
0.17 0.075 <0.001
Ref. 5.05 16.96
1.49−17.11 4.57−62.99
0.009 <0.001
Covariate GGG 1 2 3 4−5 WGGG groups 1 2 3
BCR = biochemical recurrence; GGG = Gleason Grade Group; WGGG = Weighted Gleason Grade Group.
HR
95% CI
P value
Ref. 3.00 9.82 15.49
0.65−13.89 2.23−43.24 3.63−66.14
0.16 0.003 <0.001
Ref. 1.95 8.43
0.93−4.11 3.69−19.28
0.078 <0.001
ARTICLE IN PRESS N. Waingankar et al. / Urologic Oncology: Seminars and Original Investigations 00 (2019) 1−7
Fig. 4. BCR-free survival by WGGG risk group in the development cohort. BCR = biochemical recurrence; WGGG = Weighted Gleason Grade Group.
missing data) had BCR. BCR-free survival at 24 months was 83% (95% CI 77%−87%). The same 3 WGGG cutpoints were applied in this population: 166 (42.7%) patients were in group 1, 184 (47.3%) in group 2, and 49 (10%) in group 3; these groups maintained their prognostic discrimination for BCR (Table 2, Fig. 5). In this cohort, when stratifying patients according to the WGGG risk groups, we found the 2-year BCR-free survival rates to be: 91%, 82%, and 47% for the low, intermediate, and high-risk group, respectively. C-indices of the WGGG and GGG models for BCR were 71 and 68, respectively, for the development cohort, and not calculated for the validation group due to paucity of mature follow-up data. 4. Discussion The Gleason score, while ubiquitously used to grade prostate cancer, has been limited by 3 factors: First, the communication of “low risk” disease is often unclear to patients when using a number that starts in the middle of a
Fig. 5. BCR-free survival by WGGG risk group in the validation cohort. BCR = biochemical recurrence; WGGG = Weighted Gleason Grade Group.
5
2 to 10 scale (i.e., Gleason 6 cancer), especially when initiating discussion on appropriateness of management with active surveillance [12]. Second, certain patterns historically regarded as Gleason 6 are now more commonly diagnosed as Gleason 7, leading to heterogeneity in prognoses among patients diagnosed over time with Gleason 6 cancers [13]. Finally, further heterogeneity in prognoses exists within the commonly stratified groups of low (Gleason 6), intermediate (Gleason 3 + 4 and 4 + 3), and high (Gleason 8, 9, and 10) risk patients [14]. In 2016, the World Health Organization adopted a new grading system that converted the classic Gleason score from a 2 to 10 scale, to a more easily interpreted and prognostically accurate 1 to 5 scale [11,15]. While the updated GGG system can further enhance prognostic capability when incorporated into clinical models, it is still limited by our incomplete, inefficient utilization of this data: by focusing only on the highest grade and highest volume core, we are ultimately leaving potentially relevant information “on the table.” Our technique incorporates all available pathology data from biopsy samples. Fig. 1 reflects this and clearly distinguishes GGG from our proposed WGGG, suggesting that a numerical scoring system may be used as a surrogate to reflect the aggressiveness of individual prostate cancer cases. This inclusion of all available data is consistent with the growing trend in other fields of medicine where previously bypassed data points are being combed for added diagnostic or prognostic value. For example, this has been the case in the field of Radiology, where CT scans are being re-evaluated using analytic morphometric techniques to identify specific pathology as well as to provide an objective metric to supplement the gestalt patient “eyeball test” (e.g., Psoas muscle area as a metric for frailty), thereby giving clinicians a more accurate global assessment of health and patient risk [16]. The same computational techniques are also being applied to the field of pathology to transform histologic patterns on a slide into pixels, lines, angles, and shapes with reproducible features that reliably identify histopathology, thus allowing subjective reads to be enhanced by objective analyses that utilize the entire slide. Until this so-called “precision pathology” becomes the standard of care, it is paramount that urologists utilize all readily available histopathologic data in our management of patients with prostate cancer. Currently, every PBx, regardless of scoring system used, provides 3 forms of data: (1) Gleason score or grade group, (2) number of cores positive for cancer, and (3) tumor volume per positive core. Failure to incorporate all 3 risks limiting prognostic capability and can potentially impact patient care. We demonstrated that by providing a metric reflecting the entirety of the PBx—including volume involved and grade of all positive cores—we can improve upon the prognostic value offered with utilization of a single metric as in the standard GGG. In our study, AUCs for identification of adverse pathology with GGG ranged from 0.575 to 0.823 in the development group and 0.547 to 0.867 in the validation
ARTICLE IN PRESS 6
N. Waingankar et al. / Urologic Oncology: Seminars and Original Investigations 00 (2019) 1−7
group. However, by considering the grades and volumes of all positive cores using the WGGG, these AUCs improved to 0.638 to 0.862 (development) and 0.624 to 0.929 (validation). Indeed, WGGG significantly outperformed GGG for identification of all adverse pathologic outcomes. These findings were externally validated in a well-calibrated, separate set of institutional data. Beyond the predictive association between increasing WGGG and risk of adverse pathology, we showed that WGGG alone carries prognostic value as well. Our study identified 3 distinct cutpoints in WGGG that stratify patients into risk of BCR. In the development cohort, we even identified a group of patients (group 3) for which a urologist could use WGGG to predict definitive biochemical failure with prostatectomy alone, and therefore potentially proactively augment management with a multimodal approach. These WGGG cutpoints were confirmed in the validation cohort, and while we did not identify a cutpoint above which we could definitively predict biochemical failure, this may have been secondary to the shorter follow-up time in the validation cohort. With these encouraging results, work is underway for identification of incremental net benefit of adding WGGG to clinical models. The primary strengths of our study are that patients were drawn from a large network of urologists at an institution that serves a major metropolitan area, and that our findings were externally validated. Patients were heterogeneous in demographic and clinical factors, and providers likely exhibited some degree of uncaptured variation of practice patterns. Finally, as the WGGG is derived from the GGG, it shares identical expected levels of precision, accuracy, and bias with the current gold standard, and therefore achieves criterion validity. Our study is limited by its retrospective design that uses data from a single institution for development and single institution for validation, along with lack of standardized biopsy protocol at both sites. All WGGGs were normalized for a 12-core biopsy to attempt to account for this. Other potential limitations include lack of centralized pathology review and standardized pathology reporting. However, this lack of standardization may more accurately capture the natural variation in pathology review and reporting experienced across various practices. Additionally, while WGGG and GGG did not have significantly different AUCs for identification of EPE in the validation set, this is likely because GGG performed better in the validation set (AUC 0.729) than it did in the development cohort (AUC 0.690); meanwhile, the AUCs for WGGG displayed similar predictive capability in both datasets. Furthermore, our use of 2year BCR as an outcome of interest is based on the recency of the dataset available to us. The necessary time horizon for identifying post-treatment outcomes should ideally be much longer given the typical duration of the prostate cancer disease course. Furthermore, while the validation dataset was well calibrated for most outcomes of interest, this was not the case for PSM; this is likely because PSM
reflects surgeon variation in addition to patient variation between the 2 groups. Nonetheless, discrimination was similar for PSM in both cohorts. As in any study of a novel diagnostic tool, applicability beyond the characteristics of the patient population studied must necessarily be subject to further validation. This includes the applicability of WGGG to patients who undergo MRI-TRUS fusion biopsy; our analysis included only patients with standard 12-core transrectal ultrasound-guided biopsy, and as such, our conclusions may not apply to those patients who undergo MRIguided fusion biopsy. A separate analysis is underway to determine the predictive and prognostic capability of WGGG among these patients. Early criticism of our proposed WGGG metric has emphasized that aggressiveness of cancer cannot accurately be characterized on a linear scale; yet, nearly all cancer grading is based on linear scales. This is particularly true in prostate cancer where the Gleason Grading system had traditionally been a scale of 2 to 10 based on pathologic and clinical behavior going hand-in-glove together. It evolved to a 2 to 6, 7, and 8 to 10 system, and further evolved to Gleason 7 being divided into 3 + 4 vs. 4 + 3. Finally, now we have a 1 to 5 scale, which combines grades 2 to 6 into 3 + 3 or GGG1, maintains the 3 + 4 (GGG2) and 4 + 3 (GGG3) distinction, and separates 4 + 4 (GGG4) from ≥9 (GGG5). With each evolution we have tried to reflect the Gleason score numerically; so it is with WGGG. Is it perfect? In terms of test discrimination, we have demonstrated that it performs well without the use of any clinical data like PSA, PSA density, or clinical stage. Most importantly, our proposed WGGG costs nothing except the time and effort for the urologist to collect and assess the data that has already been provided. 5. Conclusions The WGGG provides a single metric that captures the true disease potential of the entire prostate. Our externally validated findings indicate that WGGG carries more accurate predictive capability relative to the standard GGG for identification of adverse pathology and prognostic capability for identifying patients at risk for BCR. Conflicts of interest None. Supplementary materials Supplementary material associated with this article can be found in the online version at https://doi.org/10.1016/j. urolonc.2019.10.009. References [1] American Cancer Society. Cancer facts and figures 2018. Atlanta, GA: American Cancer Society; 2018.
ARTICLE IN PRESS N. Waingankar et al. / Urologic Oncology: Seminars and Original Investigations 00 (2019) 1−7 [2] United States Preventative Services Task Force Draft Recommendation Statement. Prostate Cancer Screening. 2017. [3] Graif T, Loeb S, Roehl KA, et al. Under diagnosis and over diagnosis of prostate cancer. J Urol 2007;178:88–92. [4] Cookson MS, Fleshner NE, Soloway SM, et al. Correlation between Gleason score of needle biopsy and radical prostatectomy specimen: accuracy and clinical implications. J Urol 1997;157: 559–62. [5] Baccala A Jr., Reuther AM, Bianco FJ Jr., et al. Complete resection of seminal vesicles at radical prostatectomy results in substantial long-term disease-free survival: multi-institutional study of 6740 patients. Urology 2007;69:536–40. [6] Cagiannos I, Karakiewicz P, Eastham JA, et al. A preoperative nomogram identifying decreased risk of positive pelvic lymph nodes in patients with prostate cancer. J Urol 2003;170:1798–803. [7] Chun FK, Steuber T, Erbersdobler A, et al. Development and internal validation of a nomogram predicting the probability of prostate cancer Gleason sum upgrading between biopsy and radical prostatectomy pathology. Eur Urol 2006;49:820–6. [8] Kattan MW, Eastham JA, Wheeler TM, et al. Counseling men with prostate cancer: a nomogram for predicting the presence of small, moderately differentiated, confined tumors. J Urol 2003;170:1792–7.
7
[9] Partin AW, Yoo J, Carter HB, et al. The use of prostate specific antigen, clinical stage and Gleason score to predict pathological stage in men with localized prostate cancer. J Urol 1993;150:110–4. [10] Steuber T, Graefen M, Haese A, et al. Validation of a nomogram for prediction of side specific extracapsular extension at radical prostatectomy. J Urol 2006;175:939–44:discussion 944. [11] Epstein JI, Zelefsky MJ, Sjoberg DD, et al. A Contemporary prostate cancer grading system: a validated alternative to the Gleason Score. Eur Urol 2016;69:428–35. [12] Nelson JB. Observation for clinically localized prostate cancer. J Clin Oncol 2014;32:1295–8. [13] Dong F, Wang C, Farris AB, et al. Impact on the clinical outcome of prostate cancer by the 2005 international society of urological pathology modified Gleason grading system. Am J Surg Pathol 2012;36:838–43. [14] Tsao CK, Gray KP, Nakabayashi M, et al. Patients with biopsy Gleason 9 and 10 prostate cancer have significantly worse outcomes compared to patients with Gleason 8 disease. J Urol 2015;194:91–7. [15] Pierorazio PM, Walsh PC, Partin AW, et al. Prognostic Gleason grade grouping: data based on the modified Gleason scoring system. BJU Int 2013;111:753–60. [16] Englesbe MJ, Lee JS, He K, et al. Analytic morphomics, core muscle size, and surgical outcomes. Ann Surg 2012;256:255–61.