Prognostic differences among Grade Group 4 subgroups in robotic‐assisted radical prostatectomy

Abstract Objectives To investigate whether the International Society of Urological Pathology Grade Group 4 (GG 4) subgroups have different oncological outcomes in Japanese prostate cancer (PCa) patients undergoing robotic‐assisted radical prostatectomy (RARP). Patients and Methods We conducted a retrospective multicentre cohort study in PCa patients undergoing RARP at 10 institutions in Japan. Pre‐ and post‐operative variables were collected from enrolled patients. We evaluated biochemical recurrence and clinical and pathological variables in the different GG 4 subgroups. Results A total of 3195 patients were enrolled in the study. Among them, 298 patients with GG 4 tumours (pathological Gleason scores [GSs] of 3 + 5 [N = 37], 4 + 4 [N = 257] and 5 + 3 [N = 4]) based on RARP specimens were analysed. The median follow‐up period was 25.2 months. The 3‐year biochemical recurrence (BCR)‐free survival (BCRFS) rate in the overall population was 74.5%. The 3‐year BCRFS rates in the pathological GS 3 + 5, GS 4 + 4 and GS 5 + 3 subgroups were 93.8%, 71.9% and 50.0%, respectively (P = 0.01). In multivariate analysis, pathological GS based on RARP specimens, PSA levels at surgery, pathological T stage, pathological N stage and surgical margins were independent risk factors significantly associated with BCRFS. In particular, patients with pathological GSs 4 + 4 and 5 + 3 were at higher risk of BCR than patients with pathological GS 3 + 5 (hazard ratio 4.54, P = 0.03 and hazard ratio 11.2, P = 0.01, respectively). The study limitations include the lack of central pathological specimen evaluation. Conclusions For patients with localized PCa undergoing RARP, pathological GS 4 + 4 and GS 5 + 3 were significantly associated with worse BCRFS than pathological GS 3 + 5. Pathological GS 3 + 5 may be overrated in GG 4. This observation emphasizes that primary and secondary GS should be considered to accurately stratify the risk of BCR after RARP.

emphasizes that primary and secondary GS should be considered to accurately stratify the risk of BCR after RARP.

K E Y W O R D S
Gleason score, grade group, prostate cancer, radical prostatectomy, robotic-assisted 1 | INTRODUCTION Several guidelines recommend radical prostatectomy (RP) for localized and some advanced prostate cancers (PCas). 1 Robotic-assisted RP (RARP) is suitable as a treatment modality for localized PCa. Although RARP is associated with less blood loss, a lower transfusion rate and a shorter hospitalization duration than open RP, 2 there is no consensus as to whether RARP has better oncological outcomes than open RP. [2][3][4] Approximately 2.5%-16% of PCa patients undergoing RARP still develop biochemical recurrence (BCR). 2 The International Society of Urological Pathology (ISUP) developed the Gleason grading system for PCa in 2013. 5 The five-tier Grade Group (GG) system was accepted in the 2016 edition of the World Health Organization Classification of Tumours of the Urinary System and Male Genital Organs to be used in conjunction with Gleason score (GS). 6 The five GGs include GG 1 (GS ≤ 6), GG 2 (GS 3 + 4), GG 3 (GS 4 + 3), GG 4 (GS 8) and GG 5 (GS 9-10). However, the definition of ISUP GG 4 is unsettled, 7 as several studies have  8 Additionally, oncological outcomes have been evaluated among localized or metastatic PCa patients within GG 4 treated with RP, radiotherapy (RT) and androgen deprivation therapy (ADT). 8 However, the impact of the GG 4 subgroups on oncological outcomes has never been investigated in Japanese PCa patients undergoing RARP. The ability to predict which patients are more likely to develop BCR after RARP could be helpful when choosing the best treatment strategy. Therefore, in this retrospective study, we investigated whether PCa patients undergoing RARP classified as having pathological GS 3 + 5, GS 4 + 4 and GS 5 + 3 had different clinicopathological variables and outcomes. This retrospective multicentre cohort study included PCa patients undergoing RARP at 10 institutions in Japan.

| Pathological analysis
All prostatectomy specimens were sectioned according to the wholemount staining technique and evaluated according to the ISUP 2005 guidelines. 11 The apex of the prostate was shaved perpendicular to the prostatic urethra. The bladder neck margin was coned from the specimen and sectioned perpendicularly. The remaining prostate tissue was completely sectioned at 3-mm intervals along a plane perpendicular to the urethral axis.

| Follow-up schedule
Following surgery, all patients were assessed at 3-month intervals according to their serum PSA levels. The date of BCR or PSA failure was defined as the date when the serum PSA level exceeded 0.2 ng/mL. If the PSA level did not decrease below 0.2 ng/mL after RARP, the date of BCR was defined as the date of RARP. Times to events were calculated from the day of surgery. Imaging for metastatic disease was left to the physician's judgement based on PSA levels and/or symptoms of recurrent disease. After biochemical progression, salvage RT, hormonal therapy and chemotherapy were performed. Castration-resistant PCa (CRPC) was defined as either progressively rising PSA (two 50% increases over the nadir with a PSA > 2.0 ng/mL), despite a castration level (<50 ng/dL) of testosterone, as previously described. 12

| Statistical analysis
Associations of GS with categorical variables were assessed using the chi-square test, and differences in continuous variables were analysed using the Kruskal-Wallis test. BCR-free survival (BCRFS) after RARP was analysed using the Kaplan-Meier method, and differences between groups were assessed by log-rank test. Multivariate analysis was performed using the Cox proportional hazards regression model. Statistical analysis was performed using SPSS software, Version 22 (IBM Corporation, Armonk, NY, USA), and P < 0.05 was considered to be statistically significant.
In the univariate analysis, pathological GS based on RARP specimens, PSA levels at surgery, pT stage, pN stage, surgical margins and nerve-sparing procedures were associated with BCRFS (Table 2). In the multivariate analysis, pathological GS based on RARP specimens, PSA levels at surgery, pT stage, pN stage and surgical margins were independent risk factors significantly associated with BCRFS (Table 2).
In particular, patients with pathological GS 4 + 4 and GS 5 + 3 tumours were at a higher risk of BCR than patients with pathological GS 3 + 5 tumours (hazard ratio 4.54, P = 0.03 and hazard ratio 11.2, P = 0.01, respectively).

| DISCUSSION
In our study, we evaluated the prognostic differences among PCa tumours within RP specimens in real-world data are similar to or much higher than those in our cohort. 14,16,17 The Epstein group also proposed modifying GGs by incorporating a prognostic grade grouping that accurately reflects prognosis. 5,18 They showed that the 5-year BCRFS rate for GG 4 based on RP pathology was 48%. 18   Conversely, several recent studies evaluated GS 3 + 5 and GS study, although the follow-up period was relatively short, we included only patients undergoing RARP. We found that the metastasis-free survivals and CRPC-free survivals of patients with GS 3 + 5 and GS 4 + 4 disease were similar, but patients with GS 5 + 3 disease had higher risks of metastasis and CRPC ( Figure 1B,C). Further follow-up will provide a more definitive conclusion. The present study has several limitations. First, this was a retrospective, multicentre cohort study and therefore has an inherent potential for bias. Second, we acknowledge the lack of a centralized pathological review involving biopsy and pathological GSs. GS 8 is known to be a very heterogeneous disease including highly variable quantities of Gleason 3, 4 and 5 growth patterns, which may lead to significant inter-observer variability in tumour grading. 17 Third, we had few cases of GS 5 + 3 in our cohort. Last, the follow-up period was relatively short; therefore, it may be insufficient to precisely identify the predictive factors of BCR, and there is a lack of cancer-specific survival data after RARP. However, we believe that further prospective studies with large cohort sizes will enable identification of critical roles of the heterogeneous GG 4 in predicting outcomes.

| CONCLUSIONS
In conclusion, GS  for English language editing.

CONFLICT OF INTEREST
The authors declare no conflict of interest.

ETHICS STATEMENT
The protocol for this research project has been approved by a suitably