Transition zone prostate cancer is associated with better clinical outcomes than peripheral zone cancer

Abstract Objectives To determine the biological significance of zonal origins in prostate cancer. Patients and methods Altogether, 270 consecutive radical prostatectomy cases from 2009 to 2012 were adopted. Cases were divided into those having transition zone (TZ) cancer or peripheral zone (PZ) cancer. Cases with indeterminate tumor location and central zone cancers were excluded from the analyses. Prognosis and clinicopathological features were compared between the two tumor locations. Biochemical recurrence (BCR) and clinical progression (CP) were adopted as prognostic outcome measures. Immunohistochemical features of the v‐ets avian erythroblastosis virus E26 oncogene homolog (ERG)/serine peptidase inhibitor, Kazal‐type 1 (SPINK1) status, and loss of phosphatase and tensin homolog (PTEN‐loss), as well as conventional preoperative and postoperative characteristics, were analyzed. Results This cohort comprised 93 cases of TZ cancer and 160 cases of PZ cancer. TZ cancer cases showed significantly higher BCR and CP‐free survival rate than PZ cancer cases. Notably, no TZ cancer cases developed CP during the 7.8 years of median follow‐up time. Tumor location was an independent predictive factor for BCR in the multivariate analysis. Additionally, TZ cancer cases showed a significantly lower prevalence of ERG‐overexpression and PTEN‐loss than PZ cancer cases (3.2% vs 20.1% and 2.2% vs 18.2%, respectively). Conclusion TZ cancer cases showed a better prognosis and different immunohistochemical features. Conservative treatment strategies could be considered for TZ cancer cases.

although Asian countries are known for their high mortality-to-incidence ratios associated with PCa. 2 These differences are partly explained by the delay of the introduction of a large-scale PSA screening program in most Asian countries. Moreover, differences in the anatomical location [eg, peripheral zone (PZ) vs transition zone (TZ)] and genetic alteration among races and regions might affect the difference in PCa incidence and/or mortality-to-incidence ratio.
McNeal proposed the anatomical classification of the prostate in the late 1960s, which has been widely used for clinicopathological studies of the prostate. 3 He divided the prostate gland into three segments as follows: the TZ surrounding the proximal prostatic urethra, the PZ located on the posterolateral to apical prostate, and the central zone (CZ) located from the verumontanum to the posterobasal region along the ejaculatory ducts. 3,4 Several studies from Western countries have shown that the majority of PCa arose in the PZ, and cancer in the TZ was less frequent (4%-20%). [5][6][7][8] However, our previous study showed that the prevalence of TZ cancer was as much as 42% in a Japanese cohort, and subsequent studies from eastern Asian countries also showed a higher prevalence of TZ cancers (30%-46%) than those from Western countries. [9][10][11][12][13] Although the genetic features of PCa remain unclear to date, the racial differences associated with the most frequent genetic alterations of the transmembrane protease, serine 2 (TMPRSS2)v-ets avian erythroblastosis virus E26 oncogene homolog (ERG) gene fusion and phosphatase and tensin homolog (PTEN) inactivation have been reported. The prevalence of the TMPRSS2-ERG gene fusion/ERG-overexpression in PCa was approximately 50% in Western countries, [14][15][16] although the prevalence is less at 10%-20% in east-Asian countries. 16,17 Recent studies reported an association between the TMPRSS2-ERG gene fusion and another well-known prognostic marker, serine peptidase inhibitor, Kazal-type 1 (SPINK1).
Because ERG and SPINK1 show exclusive expression, some studies divided PCa into three categories as follows: ERG-/SPINK1-(double negative), ERG-/SPINK1+, and ERG+/SPINK1-by ERG/SPINK1 status. [18][19][20] The SPINK1-positive cancer cases were reportedly associated with poorer outcomes among ERG-negative cancer cases. A few studies on the prevalence of PTEN inactivation have shown less prevalence in the African American and Asian populations than in the Caucasian population. 2,16 Recent advances in PCa diagnosis, such as multiparametric magnetic resonance imaging (MRI) and MRI-ultrasound fusion targeted biopsy, have enabled clinicians to diagnose PCa location without surgery. Here, the knowledge of the biological features of PCa of different zonal origins is crucial for the daily practice of PCa management, especially for selecting treatment options such as focal therapies. Previous reports including our research have shown that clinicopathological differences exist between TZ and PZ cancers. In these studies, TZ cancer cases showed a lower grade and earlier stage than PZ cancer cases. 7,9,11,21 However, differences in the oncological outcomes and molecular characteristics between both cancers have not been well-elucidated. Here, we conducted further examinations, including prognostic and immunohistochemical analyses, to show the detailed biological differences between both cancers.

| Patients
Altogether, 270 consecutive radical prostatectomy (RP) cases diagnosed by a single genitourinary pathologist (HT) from August 2009 to August 2012 were adopted. Cases with any preoperative treatment, such as endocrine therapy, radiation therapy, and chemotherapy, were excluded. Eighteen cases with indeterminate cancer location or CZ location were excluded from statistical analyses as mentioned below. Hence, 252 cases were ultimately analyzed.

| Pathological examination
The RP specimens were routinely processed, and pathological diagnosis was made as previously reported. 9 Cases were divided into those having TZ or PZ cancers based on their tumor location. The TZ and PZ were defined microscopically according to the McNeal's criteria. 3,4 The zonal origin of the tumor was determined using a tumor mapping picture written by microscopic examination. In the multifocal cases, the index tumor was utilized for analyses. The index tumor was defined as the largest tumor in the case, similar to the definition used in previous studies. 8,11,22 In the current study, tumors extending over the TZ and PZ were treated as cancers of no dominant location (ND) and were excluded from statistical analyses.
Thus, the definition of TZ/PZ cancer is rather different from that in our previous study. 10 The CZ cancer cases were also excluded from the analyses.
In the RP specimens, intraductal carcinoma was examined by one genitourinary pathologist (SS) according to the McNeal's criteria. 23 Immunohistochemical analyses for anti-ERG rabbit monoclonal antibody (clone EPR 3,864, Abcam, Cambridge, UK), anti-PTEN rabbit monoclonal antibody (clone D4.3, Cell Signalling Technology, Denvers, MA, US), and anti-SPINK1 mouse monoclonal antibody (clone 4D4, Abnova, Taipei City, Taiwan) were performed using BenchMark XT automated stainer (Roche Diagnostics K.K., Tokyo, Japan). In the immunohistochemical analyses, the cut-off value was set as 10%. Thus, cases were categorized as ERG/SPINK1 positive or PTEN-loss if 10% or more of tumor cells stained for ERG/SPINK1 antibody or lost the staining for PTEN antibody.

| Acquisition of clinical information
For each patient, clinical information such as age, serum PSA level, clinical T stage, and postoperative follow-up data, were collected from medical records.

| Outcome measures
We adopted biochemical recurrence (BCR) and clinical progression (CP) as the primary and secondary measures of treatment outcome, respectively. The definition of BCR was as previously described. 24 The PSA cut-off value for BCR was set as 0.2 ng/mL. If postoperative treatment (eg, endocrine and radiation therapy) was initiated in the absence of an increase in the serum PSA level to ≥0.2 ng/mL, the day of BCR was defined as the day of initial treatment. CP was defined as metastasis detected by radiological examination including bone scintigraphy. In our institute, treatment after biochemical recurrence is uniformly provided to the patients as follows; salvage radiation therapy with 65 Gy or more with/without androgen deprivation therapy on BCR after RP; androgen deprivation therapy on the second BCR after salvage radiation therapy.

| Statistical analyses
In the first analysis, we compared the clinicopathological features and prognoses between TZ and PZ cancers. The Mann-Whitney U-test was performed for comparing continuous (numerical) variables, and the Fisher's exact test was used for categorical variables. BCR-free and CP-free survival rates were calculated by the Kaplan-Meier method, and the inter-group difference was compared by the log-rank test.
Additionally, multivariate Cox regression analysis for BCR was per- Thereafter, we examined the clinicopathological significance of ERG/SPINK1 expression status in each tumor location. The TZ and PZ cancers were subdivided into three categories by ERG/SPINK1 status as ERG-/SPINK1-, ERG-/SPINK1+, and ERG+/SPINK1-, as previously reported. 20 Clinicopathological and prognostic (BCR) differences were compared between the groups. In the TZ group, ERG+/SPINK1-cases were excluded from statistical analyses because of extremely low prevalence (see Results). Additionally, the Kruskal-Wallis test was used to examine the inter-group difference of continuous variables among the three groups.
In all the analyses, statistical significance was defined as a P-value of <.05. The EZR software package based on R (R Foundation for Statistical Computing, Vienna, Austria) was used for the analyses. 25

| Research ethics
This study received prior approval from the Ethics Committee of the affiliated institution (institutional ID number: 25-312).
Informed consent was not required due to the retrospective study design. Regarding postoperative pathological findings, TZ cancer cases showed significantly lower GG and pT than PZ cancer cases.

| Clinicopathological patient backgrounds
Lymph node metastasis was detected more frequently in PZ cancer cases (8.8%) than in TZ cancer cases (3.2%), although a significant difference was not shown (P = .186). Moreover, no significant difference was shown for tumour volume and positive surgical margin rate. The immunohistochemical profile in the entire cohort was as follows: ERG-overexpression shown in 14.8%, PTEN-loss in 13.0%, and SPINK1 expression in 15.6%. Compared to PZ cancer cases, TZ cancer cases showed an extremely lower prevalence of ERG-overexpression (3.2% vs 20.1%) and PTEN-loss (2.2% vs 18.2%) with a significant difference (P = .0001 and <.0001, respectively). A significant difference was not shown for SPINK1 expression (P = .858).

| Tumor location and biochemical recurrence
The median follow-up period was 93.8 (interquartile range: 59.3-102) months in the entire cohort, 93.4 (60-101.8) months in TZ cancer cases, and 93.6 (55.3-101.9) months in PZ cancer cases without a significant difference (P = .854). TZ cancer cases showed a significantly higher BCR-free survival rate than PZ cancer cases (P < .0001, Figure 1a). The results for univariate and multivariate Cox regression analyses are shown in Tables 3 and 4. In the multivariate model incorporating preoperative factors and tumor location, tumor location was an independent predictive factor of BCR (P = .0008), as well as PSA, cT, and GG in biopsy (Table 3). In the multivariate model incorporating postoperative pathological features and tumor location, tumor location was an independent predictive factor of BCR (P = .036), as well as GG in RP specimen, positive surgical margin, and lymph node metastasis (Table 4).
PTEN-loss was associated with an increased risk of BCR in the univariate analysis (P = .0021), although it was not an independent predictive factor in the multivariate analysis (P = .2585). The ERG/ SPINK1 status was not predictive of BCR in both the univariate and multivariate analyses in the entire cohort.

| Tumor location and clinical progression
TZ cancer cases showed a significantly higher CP-free survival rate than PZ cancer cases (P = .0317, Figure 1b). Notably, no TZ cancer case developed CP during the median follow-up period of 7.8 years, although eight cases of PZ cancer and 11 cases in the entire cohort developed CP.

| Clinicopathological significance of ERG/ SPINK1 status in transition zone cancers
In the TZ cancer group, statistical analyses were conducted only between ERG-/SPINK1-and ERG-/SPINK1 + subgroups because of the extremely lower prevalence of ERG-overexpression in this group.
The ERG-/SPINK1 + cases showed a significantly lower BCR-free survival rate (P = .0472, Figure 2a), although no clinicopathological parameters showed a significant difference between the groups (Table 5).

| Clinicopathological significance of ERG/ SPINK1 status in peripheral zone cancers
In the PZ cancer group, a significant difference was not shown for the BCR-free survival rate among the three subgroups, ERG-/SPINK1-, ERG-/SPINK1+, and ERG+/SPINK1-(P = .39, Figure 2b). Meanwhile, in comparing clinicopathological factors, the distribution of GG showed a significant difference (P = .0119, Table 5). The proportion of high GG (GG3-5) cases was 54.4% in the highest ERG-/SPINK1-subgroup, 41.7% in the second highest ERG-/SPINK1 + subgroup, and 25% in the lowest ERG+/SPINK1-subgroup. Other clinicopathological parameters excluding patient age did not show any significant differences.

| D ISCUSS I ON
We have previously shown the pathological differences in RP specimens between TZ and PZ cancers. 9 In this study, we conducted a comprehensive analysis to examine the association between clinical prognosis and tumor location in a larger cohort adopting stricter definitions of TZ and PZ cancers. We observed that the prevalence of TZ cancer was lower than that noted in our previous report (42%) because of the stricter definition of tumor location and cohort ex-  33 The technical advances in the imaging diagnosis and biopsy method, including MRI-ultrasound fusion targeted biopsy, will allow for more accurate preoperative diagnoses of tumor location. This will lead to a more affirmative adoption of less invasive treatments such as active surveillance and focal therapy for TZ cancers, which have more indolent clinicopathological features and a better prognosis. Moreover, observations could be adopted more commonly to elderly TZ cancer patients as a treatment option. In contrast, this approach should be adopted more carefully for PZ patients.
However, this study has several limitations including its relatively small number of cases due to its single institute design, selection bias for RP as a treatment option, and an old cohort from approximately 10 years ago. Because of the small cohort size, available statistical analysis methods were limited. Considering the recent advances in techniques for PCa diagnosis, cancer biology of this cohort could be different from recent cases at the diagnostic stage.
Further investigations need to be performed with a multi-institutional cohort incorporating novel diagnostic technologies including MRI and targeted biopsy.
In conclusion, TZ cancers showed better outcomes and different clinicopathological features than PZ cancers. Thus, TZ cancers may be treated more conservatively.

ACK N OWLED G M ENTS
We would like to thank Editage (www.edita ge.com) for English language editing.

CO N FLI C T O F I NTE R E S T
None declared.