Prediction of Pathologic Stage Based on Clinical Stage, Serum PSA, and Biopsy Gleason Score: Partin Tables in the Contemporary Era

Jeffrey J. Tosoian, Meera Chappidi, Zhaoyong Feng, Elizabeth B. Humphreys, Misop Han, Christian P. Pavlovich, Jonathan I. Epstein, Alan W. Partin, Bruce J. Trock
The James Buchanan Brady Urological Institute and Department of Urology at the Johns Hopkins University School of Medicine, Baltimore, MD, USA

The Partin Tables use clinical features of prostate cancer – Gleason score, serum PSA, and clinical stage – to predict whether the tumor will be confined to the prostate. The tables are based on the accumulated experience of urologists performing radical prostatectomy at the James Buchanan Brady Urological Institute. For decades, urologists around the world have relied on the tables for counseling patients preoperatively and for surgical planning.

This website provides our updated findings based on 4459 patients treated with radical prostatectomy at Johns Hopkins from January 2010 through October 2015. A summary of our methodology and findings are contained in the abstract below.

Abstract

Objective

• To update the Partin Tables for prediction of pathological stage in the contemporary setting and examine trends in patients treated with radical prostatectomy (RP) over the past three decades. Patients and Methods

• From January 2010 through October 2015, a total of 4459 men meeting inclusion criteria underwent RP and pelvic lymphadenectomy for histologically-confirmed prostate cancer at the Johns Hopkins Hospital.

• Preoperative clinical stage, serum prostate-specific antigen (PSA) level, and biopsy Gleason score (i.e. prognostic Grade Group) were utilized in polychotomous logistic regression to predict the probability of pathological outcomes categorized as: organ-confined (OC), extraprostatic extension (EPE), seminal vesicle involvement (SV+), or lymph node involvement (LN+).

• Preoperative characteristics and pathological findings in men treated with RP since 1983 were collected and clinical-pathological trends were described.

Results

• Median age at surgery was 60 years (range 34-77) and median PSA was 4.9 ng/ml (0.1-125.0).

• The observed probabilities of pathological outcomes were: OC disease in 74%, EPE in 20%, SV+ in 4%, and LN+ in 2%.

• The probability of EPE increased substantially when biopsy Gleason score increased from 6 (Grade Group 1) to 3+4 (Grade Group 2). The probability of LN+ was substantially higher for biopsy Gleason score 9-10 (Grade Group 5) as compared to lower Gleason scores.

• Area under the receiver operating characteristic curves for binary logistic models predicting EPE, SV+, and LN+ versus OC were 0.724, 0.856, and 0.918, respectively.

• The proportion of men treated with biopsy Gleason score ≤6 cancer (Grade Group 1) was 47%, representing a substantial decrease from 63% in the previous cohort and 77% in 2000-2005. The proportion of men with organ-confined cancer has remained similar during that time, equaling 73% to 74% overall.

• The proportions of men with SV+ (4.1% from 3.4%) and LN+ (2.3% from 1.4%) increased relative to the preceding era for the first time since the Partin tables were introduced in 1993. Conclusions

• The Partin Tables remain a straightforward and accurate approach for projecting pathological outcomes based on readily available clinical data.

• At our tertiary care referral center, the proportion of men with OC disease has remained stable since 2000, despite a substantial decline in the proportion of men with biopsy Gleason score 6 (Grade Group 1). This is consistent with the notion that many men with Gleason score 6 (Grade Group 1) disease were overtreated in previous eras.

*Note: "n" refers to the number of patients analyzed in each group. The numbers in the table represent the probability estimate of being in a particular pathologic stage based on clinical stage, PSA, and Gleason score. The numbers in parentheses represent the 95% confidence interval, which shows the range of uncertainty in each estimate. The uncertainty arises because we only analyze a sample of men who underwent prostatectomy (i.e. those who underwent surgery at Johns Hopkins), rather than all prostatectomy patients in the US. The true probability could be any value between the lower and higher value of the confidence interval.