Non‐organ confined stage and upgrading rates in exclusive PSA high‐risk prostate cancer patients

The pathological stage of prostate cancer with high‐risk prostate‐specific antigen (PSA) levels, but otherwise favorable and/or intermediate risk characteristics (clinical T‐stage, Gleason Grade group at biopsy [B‐GGG]) is unknown. We hypothesized that a considerable proportion of such patients will exhibit clinically meaningful GGG upgrading or non‐organ confined (NOC) stage at radical prostatectomy (RP).


| INTRODUCTION
The D'Amico risk stratification system, initially introduced by D'Amico et al. in 1998, still represents a widely accepted and used risk stratification system for patients with clinically localized prostate cancer (PCa). 1,2 Risk stratification is based on prostate-specific antigen (PSA) level, Gleason score at diagnosis and clinical tumor stage (cT). 1 Among patients that qualify as high-risk patients, some will only harbor high-risk PSA levels (>20 ng/ml) that are accompanied by lowrisk clinical stage (cT1c) and low-to intermediate risk Gleason Grade group (GGG) at biopsy (GGG1, GGG2, GGG3). In those specific individuals, dose intensification or treatment assignment according to established high-risk protocols may be questioned based on its excessive intensity, when attempted treatment intensity modifications are based on unknown rates of non-organ confined (NOC) stage and/ or of presence high-risk GGG (GGG4/GGG5) at radical prostatectomy (RP) in this specific PCa patient population with an unusual distribution of risk factors. [3][4][5][6][7][8][9] For example, in radiation therapy protocols, exemptions may be granted for lower duration of androgen deprivation therapy. [10][11][12] Similarly, at RP a more limited or no lymph node dissection may be contemplated. We addressed this knowledge gap and examined rates of upgrading to GGG4/GGG5 and NOC stage in this specific PCa population. Additionally, we tested for potential presurgical eligible clinical variables which were associated with upgrading to GGG4/GGG5 or NOC stage. We addressed this void by relying on a contemporary, North-American cohort of patients within the Surveillance, Epidemiology, and End Results (SEER) database (2010-2015).

| Study population
The current SEER database samples the United States population and approximates it in demographic composition and cancer incidence. 13 Within SEER database 2010-2015, we identified all patients ≥18 years old with histologically confirmed adenocarcinoma of the prostate, diagnosed at biopsy (International Classification of Disease for Oncology [ICD-O-3] code 8140 site code C61.9), as previously reported. 14 We subsequently focused on cT1c-stage patients (cN0/cM0) and GGG1, GGG2, or GGG3 at biopsy, who underwent RP. Moreover, only patients with PSA >20 and ≤50 ng/ml at diagnosis were included in further analyses. Exclusion criteria consisted of less than 10 or more than 14 biopsy cores, unknown pT-stage and unknown GGG at RP. Furthermore, cases identified only at autopsy or death certificate or with unknown histology were excluded. These selection criteria resulted in a cohort of 486 eligible patients, who represent the population of the current study cohort.

| Statistical analyses
Descriptive statistics included frequencies and proportions for categorical variables. Means, medians, and interquartile ranges (IQR) were reported for continuously coded variables. The Chi-square tested the statistical significance in proportions' differences. The t test and Kruskal-Wallis test examined the statistical significance of means' and distributions' differences.
Statistical analyses were based on four steps. First, baseline characteristics were tabulated (GGG1 vs. GGG2 vs. GGG3). Second, rates of upgrading and NOC stage were tabulated for each subgroup specifically. Upgrading was defined as presence of GGG4 or GGG5 at RP, irrespective of the initial GGG at biopsy.
NOC stage was defined as the presence of extracapsular extension (ECE; pT3a) of the tumor and/or seminal vesicle invasion (SVI; pT3b) and/or pT4 and/or pathological lymph-node invasion (LNI; pN1) at RP. Fourth, two separate multivariable logistic regression models tested for independent predictors of (a) upgrading to GGG4/GGG5 and (b) of presence of NOC stage.
Covariates consisted of patient age (per year), baseline PSA (per unit ng/ml), percentage of positive cores for PCa (<33 vs. 33-66 vs. >66%), total numbers of cores obtained at biopsy and GGG at biopsy (GGG1 vs. GGG2 vs. GGG3). Finally, we tested for presence of interaction between GGG at biopsy and percentage of positive cores within each of the two separate logistic regression models.
All tests were two sided with a level of significance set at p < 0.05 and R software environment for statistical computing and graphics (version 3.4.3) was used for all analyses. 15 3 | RESULTS

| Association between clinical variables and upgrading and NOC stage
In multivariable logistic regression models of upgrading to GGG4 or GGG5, only GGG3 at biopsy was an independent predictor (odds ratio Analyses testing for presence of interaction between GGG at biopsy (GGG1 vs. GGG2/GGG3) and percentage of core ratio (<33 vs. 33-66 vs. <66%) revealed insignificant results (p = 0.8; data not shown).
T A B L E 1 Patient and clinical descriptives characteristics of cT1c-stage prostate cancer patients and GGG1/GGG2/GGG3 at biopsy with PSA >20 and ≤50 ng/ml treated radical prostatectomy within the Surveillance, Epidemiology, and End Results database (2010-2015) GGG  These observations including rates and odds ratios may represent a useful indicator of risk stratification within this special population of interest with high-risk PSA, but favorable clinical stage and favorable to intermediate GGG at biopsy. To the best of our knowledge, this special patient population with discordant baseline risk characteristics has not been previously addressed.
Finally, rates of lymph node dissection differed significantly within the study population. Specifically, rates of lymph node dissection were 69%, 88%, and 92% for, respectively, patients with GGG1, GGG2, and GGG3 at biopsy (p < 0.001; Table 1). Despite the fact that the lymph node yield, defined as the number of lymph nodes resected during lymph node dissection, did not differ between GGG1, GGG2, and GGG3 patients (p = 0.6), it may be postulated that GGG at biopsy most likely has influenced the rate of lymph node dissections performed within the current study population. It is of interest that the extent of lymph node removal recorded (Median: 6; IQR: 3-11) in F I G U R E 2 Diagrams depicting concordance and discordance rates between biopsy Gleason grade group (B-GGG) and radical prostatectomy GGG (RP-GGG), stratified by the initial B-GGG of prostate cancer patients with cT1c-stage and PSA >20 and ≤50 ng/ml, subsequently treated with radical prostatectomy within the Surveillance, Epidemiology, and End Results database (2010-2015); All values are frequencies (% The current study is not devoid limitations. First, even though relying on a large-scaled population-based data source, namely the SEER, the current study should be interpreted under the light of a limited sample size. However, it is of note that population-based approaches, such as the current design, represent the most promising approach for addressing these hypotheses due to available data magnitude. Second, potential important differences in baseline characteristics which were not assessable within SEER, may have confounded our findings, such as imaging findings (e.g., magnetic resonance imaging), prostate-associated features (prostate volume, chronic inflammation), and potential bias arising from methodological differences in the process of specimen procurement, fixation and histopathological analyses. [17][18][19][20][21] Moreover, lack of central pathology review may impose a bias that could not be accounted for in the current manuscript. It is of note that this limitation is inherent to all population-based analyses. 22,23 Additionally, the SEER database does not allow to account for the percentage of tumor infiltration within a specific biopsy core. However, since this lack of information is attributable to all subgroups equally and can be seen as a non-differential bias, it is unlikely that the results between those groups were biased in a substantial manner. Finally, exact data regarding potential differences in biopsy mapping templates are not available and might demonstrate a potential bias. To minimize potential biases which are likely to arise from different biopsy templates and consequently different numbers of biopsy cores taken, we included only patients with 10-14 cores harbored at biopsy. By relying on this very strict inclusion criteria, confounding due to a heterogeneity in number of cores was reduced in the best possible approach.

| CONCLUSIONS
In cT1c-stage patients with high-risk PSA baseline, but low-to intermediate risk biopsy-GGG, the rate of upgrading to GGG4 or GGG5 is low (11%). However, NOC stage is found in the majority (51%) and can be independently predicted with percentage of positive cores at biopsy and GGG at biopsy.

ACKNOWLEGMENTS
Benedikt Hoeh was awarded a scholarship by the Stiftung Giersch.
Open Access funding enabled and organized by Projekt DEAL.
T A B L E 2 Separate multivariable logistic regression models testing for independent predictor for (a) upgrading to GGG4/GGG5 and (b) nonorgan confined stage in clinical T1c-stage prostate cancer patients with PSA >20 and ≤50 ng/ml, treated radical prostatectomy within the Surveillance, Epidemiology, and End Results database (2010-2015) Logistic regression model predicting upgrading to GGG4/GGG5