The performance of the new prognostic grade and stage groups in conservatively treated prostate cancer

Abstract

We evaluated the prognosis of the new grade groups and American Joint Committee on Cancer (AJCC) stage groups in men with prostate cancer (PCa) who were treated conservatively. A total of 13 798 eligible men were chosen from the Surveillance Epidemiology and End Results database. The new grade and AJCC stage groups were investigated on prostate biopsy specimens. Kaplan–Meier survival analysis and multivariable hazards models were applied to estimate the association of new grade and stage groups with overall survival (OS) and PCa-specific survival (CSS). Mean follow-up was 42.65 months (95% confidence interval: 42.47–42.84) in the entire cohort. The 3-year OS and CSS rates stepped down for grade groups 1–5 and AJCC stage groups I–IVB, respectively. After adjusting for clinical and pathological characteristics, all grade groups and AJCC stage groups were associated with higher all-cause and PCa-specific mortality compared to the reference group (all P ≤ 0.003). In conclusion, we evaluated the oncological outcome of the new grade and AJCC stage groups on biopsy specimens of conservatively treated PCa. These two novel clinically relevant classifications can assist physicians to determine different therapeutic strategies for PCa patients.

INTRODUCTION

Prostate cancer (PCa) is the second most common cancer of men in the world.1 Men with newly diagnosed PCa are often assigned a clinical staging, which includes physical examination, prostate-specific antigen (PSA), the biopsy Gleason score as well as imaging studies. For patients who undergo radical prostatectomy (RP), additional information is obtained from histological examination of the surgically resected specimen, which forms the basis for pathologic staging. Men who do not undergo prostatectomy are not assigned a pathologic stage and treatment decisions are based on the clinical stage.

The Gleason grade for the two most prevalent differentiation patterns is combined to create the Gleason score, which is now incorporated into a newly adopted grade group system. The revised system was endorsed and accepted by the International Society of Urological Pathology (ISUP) in 2014 and then included in the World Health Organization (WHO) in 2016.2,3 In new grade groups, tumors are separated into five categories based on the primary and secondary Gleason patterns: group 1 (Gleason score 3 + 3), group 2 (3 + 4), group 3 (4 + 3), group 4 (4 + 4, 3 + 5, or 5 + 3), and group 5 (4 + 5, 5 + 4, or 5 + 5). Previous studies have substantially proven the prognostic performance of this grading system in men who underwent radical prostatectomy4,5,6,7,8 using biochemical recurrence (BCR), clinical recurrence, and overall survival (OS). Several studies also discussed the differences in the new grade groups in radiotherapy-treated4,5,9 or conservatively treated patients.10 A recent study published in European Urology demonstrated that new grade groups are associated with the risk of PCa-specific survival (CSS) and bone metastasis in nondefinitive therapy and RP-only cohort.11 The eighth (2017) American Joint Committee on Cancer (AJCC) staging system12,13 added the new grade groups and pretreatment PSA to the anatomic tumor-node-metastasis (TNM) staging to create prognostic stage groups for PCa, which to our knowledge, has not been validated up to now. Therefore, in our study, all-cause and PCa-specific mortality were used as the outcome to evaluate the prognostic performance of both the new grading system and AJCC staging system in conservatively treated PCa cohort.

PATIENTS AND METHODS

Study population

All patients were selected in the Surveillance, Epidemiology, and End Results (SEER) database with prostate adenocarcinoma in 2010. All men had an explicit Gleason Score of primary and secondary pattern values on needle core biopsy. Patients who underwent initial primary management with RP or radiation therapy were excluded. As more experience has been gained with Gleason grading, pathologists generally will not diagnose PCa with composite Gleason scores of 2–5.2 Therefore, the range of composite Gleason scores was chosen as 6 (3 + 3) to 10 (5 + 5) on prostate biopsies. The permission to access the research data was obtained by reference number 12027-Nov2016.

Covariates

Demographic characteristics included age, race, and marital status. Clinical characteristics included preoperative PSA value, clinical T and N stages, distant metastasis, and Gleason score at biopsy. We defined prostate needle biopsy-positive rate as the number of positive cores over total needle biopsy cores. The new AJCC stage group was determined according to the AJCC Eighth Edition Cancer Staging Manual.12,13

Statistical analysis

Variables of Table 1 presented as medians (interquartile range, IQR) were analyzed by Mann–Whitney U-test and percentages were analyzed by Chi-squared test. The Kaplan–Meier method was used to calculate survival in PCa patients and the log-rank test was used to estimate the differences in OS and CSS probabilities. The association of the new grade groups and AJCC stage groups with OS and CSS was analyzed using the Cox proportional hazards model. Predictive accuracy calculations of the model were assessed using the Harrell's concordance index (C-index).14 Statistical analyses were performed using SPSS version 21 (IBM, Chicago, IL, USA) and R (version 3.0.0, The R Foundation for Statistical Computing, Vienna, Austria), and a two-sided P < 0.05 was considered statistically significant.

Table 1.

Association of new grade group with clinicopathological variables in the entire cohort

RESULTS

Patient characteristics

Data pertaining to 13 798 patients diagnosed with prostate adenocarcinoma as well as information on biopsy Gleason score and TNM stage in 2010 were extracted. A total of 2180 (15.8%) men experienced all-cause death and 771 (5.6%) experienced PCa-specific death. The baseline characteristics are summarized in Table 1.

Among the 13 798 patients, 7604 (55.1%) patients were assigned to grade group 1, 2347 (17.0%) were allocated to grade group 2, 1170 (8.5%) were assigned to grade group 3, 1237 (9.0%) were assigned to grade group 4, and the remaining 1440 (10.4%) were designated as grade group 5. The new grade groups were associated with age, race, and marital status, and higher grade groups were related to higher preoperative PSA, higher biopsy-positive rate, higher T stage, more lymph node metastases, and more distant visceral metastasis (P < 0.001). A total of 8762 (63.5%) cases could be assigned to specific AJCC stage group according to the definition.12 Among them, 3482 (25.2%) belonged to stage group I, 548 (4.0%) belonged to IIA, 1315 (9.5%) to IIB, 879 (6.4%) to IIC, 678 (4.9%) to IIIA, 93 (0.7%) to IIIB, 618 (4.5%) to IIIC, 124 (0.9%) to IVA, and 1025 (7.4%) to IVB.

Correlation of new grade groups with OS and CSS

The mean follow-up time was 42.65 months (95% confidence interval [CI]: 42.47–42.84) and 45.40 months (95% CI: 45.28–45.52) for patients with OS and CSS, respectively. Remarkable differences concerning OS and CSS were found between each grade group (P < 0.001, Figure 1). Pairwise comparisons of prognostic grade groups were significantly different using the log-rank test (all P < 0.001, Supplementary Table 1 ^{(119.9KB, tif)} and 2 ^{(120.8KB, tif)} ). The 1, 2, and 3-year estimated OS rates, CSS rates, and the mean survival time for grade groups 1–5 were declining step by step, respectively (Supplementary Table 3 ^{(205.8KB, tif)} and 4 ^{(205.8KB, tif)} ).

Figure 1.

The Kaplan–Meier plot of (a) overall survival and (b) prostate cancer-specific survival stratified by the new grade groups.

Supplementary Table 1

Pairwise comparison of new grade groups in prostate overall survival using the log-rank test

Supplementary Table 2

Pairwise comparison of new grade groups in prostate cancer-specific survival using the log-rank test

Supplementary Table 3

Survival rates of new grade groups in the follow-up years and the mean overall survival time

Supplementary Table 4

Survival rates of new grade groups in the follow-up years and the mean cancer-specific survival time

Relative to grade group 1, all the other grade groups were closely correlated with an increased hazard of OS and CSS on univariate analyses (all P < 0.001, Table 2 and 3). After adjusting for age, preoperative PSA, needle core biopsy-positive rate, race, marital status, and clinical TNM stage, all the other grade groups remained associated with an elevated higher risk of OS and CSS (all P < 0.001, Table 2 and 3). However, the multivariable prognostic model based on the new five-tier grade groups failed to provide a better predictive accuracy than the old classification (improvements of C-index <0.01, Table 4).

Table 2.

Uni- and multivariate analyses of overall survival in the entire cohort

Table 3.

Uni- and multivariate analyses of prostate cancer-specific survival in the entire cohort

Table 4.

Results of Harrell's C-index for the entire cohort with the standard three-tier Gleason groups and the new five-tier grade groups

Correlation of new AJCC stage groups with OS and CSS

The mean follow-up time was 42.06 months (95% CI: 41.82–42.30) and 44.87 months (95% CI: 44.71–45.04) for patients with OS and CSS, respectively. Remarkable differences on OS and CSS were observed between the new AJCC stage groups (P < 0.001, Figure 2). Pairwise comparisons of prognostic AJCC stage groups were significantly different in OS with the exception of stage group IIIA versus Group IIIB (P = 0.058), IIIA versus IVA (P = 0.296), IIIB versus IIIC (P = 0.832), IIIB versus IVA (P = 0.455), and IIIC versus IVA (P = 0.200) (Supplementary Table 5 ^{(363.9KB, tif)} ) and in CSS with the exception of stage group IIA versus Group IIB (P = 0.512), IIA versus IIC (P = 0.185), IIIA versus IIIB (P = 0.655), and IIIC versus IVA (P = 0.860) (Supplementary Table 6 ^{(363.9KB, tif)} ). The 1, 2, and 3-year estimated OS rates, CSS rates, and the mean survival time for AJCC stage groups I–IVB were declining step by step, respectively (Supplementary Table 7 ^{(322.4KB, tif)} and 8 ^{(322.4KB, tif)} ).

Figure 2.

The Kaplan–Meier plot of (a) overall survival and (b) prostate cancer-specific survival stratified by the new AJCC stage groups. AJCC: The American Joint Committee on Cancer.

Supplementary Table 5

Pairwise comparison of new American Joint Committee on Cancer Stage groups in prostate overall survival using the log-rank test

Supplementary Table 6

Pairwise comparison of new American Joint Committee on Cancer stage groups in prostate cancer-specific survival using the log-rank test

Supplementary Table 7

Survival rates of new American Joint Committee on Cancer stage groups in the follow-up years and the mean overall survival time

Supplementary Table 8

Survival rates of new American Joint Committee on Cancer stage groups in the follow-up years and the mean cancer-specific survival time

Compared to stage group I, all the other stage groups were in good correlation with an incremental hazard of OS and CSS on univariate analyses (all P ≤ 0.001, Table 2 and 3). After adjusting for age, race, and marital status, relative to stage group I, all the other groups were still in good correlation with an increased hazard of OS and CSS (all P ≤ 0.003, Table 2 and 3).

DISCUSSION

Our study shows that, in PCa men who did not receive definitive treatment for their cancer, both the new grade grouping and AJCC stage grouping system can effectively predict the risk of OS and CSS as the end point.

The Gleason grade for the two most prevalent differentiation patterns has been used to create the Gleason score and is now being used in the newly adopted grade group system, which provides a more accurate risk stratification than the current composite Gleason score.4 Tumors are separated into five categories based on the primary and secondary Gleason patterns. Gleason 3 + 4 and 4 + 3 tumors were formally grouped as Gleason score 7. However, as these groups differ substantially in prognosis, they have now been divided into groups 2 and 3, respectively. Men with Gleason 4 + 3 tumors (where grade 4 is more prevalent than grade 3) have a less favorable outcome than those with Gleason 3 + 4 disease.15,16,17,18,19,20 For example, a multivariate analysis of PCa men with Gleason 7 found a remarkable incremental risk of seminal vesicle invasion in those with Gleason 4 + 3 disease.21,22 The new group grading system, as discussed above, separates Gleason score 7 into grade group 2 (Gleason 3 + 4) and grade group 3 (Gleason 4 + 3) to address the different risks associated with each group. The grade group system was comprehensively validated in RP patients between 2005 and 2014. There was an increasing risk of BCR and PCa mortality with increasing grade.10 Relative to grade group 1, hazard ratio (HR) was 1.9, 5.4, 8.0, and 11.7 in grade groups 2, 3, 4, and 5, respectively. In our study, the prognostic grade grouping system offered incremental prediction of the risk of OS and CSS with adjusted HR ranging from 1.77 to 3.05 and 2.68 to 7.74, respectively. A higher histologic grade group indicated a greater likelihood of having nonorgan-confined disease as well as a worse outcome. These results are in line with those of previous studies.

It remains unclear whether the new grade groups provide a greater predictive accuracy than the old ones. A multi-institutional validation study from European PCa men proved that the new grade groups do not improve the accuracies of prognostic models compared to the current three-tier classification.8 Similar to previous studies,8,23 we found that, despite an independent association with OS and CSS, the new grading system failed to add more prognostic value than the old classification using concordance index, which may be explained by a short follow-up time.

The serum PSA level provides important prognostic information that is useful for determining the degree of workup required after the initial biopsy and for planning subsequent treatment. A higher baseline serum PSA is linked to an increased risk of more advanced disease as well as subsequent disease progression. The use of a combination of serum PSA level, the new grade groups, and clinical T stage allows a more reliable prediction of pathologic stage and treatment outcome. These three parameters have been incorporated into prognostic stage groups in the eighth (2017) edition of the AJCC TNM staging system. The new AJCC prognostic stage group was also evaluated in PCa men in our study for the first time, which offered increased prediction of the risk of OS and CSS with adjusted HR ranging from 1.82 to 15.85 and 4.75 to 181.48, respectively.

Limitations do exist in this study. First, the retrospective nature results in selection bias in this study. Second, the clinical staging based on needle biopsy can under- or over-estimate the extent and aggressiveness of the disease when compared to results based on a resection specimen after radical prostatectomy.24,25 Third, a central pathologic review is lacking in SEER data, which has been proven to be important when interpreting pathology results.26 Forth, inaccurate PSA data within SEER27 could have a negative impact on further AJCC stage classification. Fifth, as the grade and stage of PCa patients increase, their comorbidity is also increasingly severe so as to be considered for “conservative” or no treatment. Hence, untreated patients tend to experience more noncancer-specific mortality than treated men. The lack of reporting comorbidity and competing risk reduces the credibility of this study. Sixth, PSA biochemical recurrence and metastatic recurrence rates were unavailable in this study, which would be more informative as the end point than OS and CSS, subject to a too short follow-up time (<4 years). Moreover, the effect of adjuvant hormone therapy was also not factored in this analysis, where many PCa patients would have likely received hormone therapy, especially for grade groups 4–5. Finally, there was no access to data on positive surgical margin or prostate gland volume, which are all important prognostic factors for PCa patients and may have added further clarification to this analysis.

CONCLUSION

There is a strong correlation between the new grading system and AJCC staging system with all-cause and prostate cancer-specific mortality in biopsy specimens obtained from the conservatively treated population. However, more prospective and multicenter studies are necessary to confirm and validate these findings.

AUTHOR CONTRIBUTIONS

XZH and RFX designed the study. YC collected the data. CCJ and LKH performed the statistical analysis. CC prepared the manuscript. RFX and XZH reviewed the manuscript. All authors read and approved the final manuscript.

COMPETING INTERESTS

All authors declare no competing interests.

Supplementary Information is linked to the online version of the paper at Asian Journal of Andrology website.