Molecular characterization of prostate cancer in Middle Eastern population highlights differences with Western populations with prognostic implication

Abstract

Background

To investigate the incidence and prognostication of ERG, PTEN and SPINK1 protein expressions in prostate cancer cohort of Middle Eastern descent in comparison to published data from Western population.

Methods

Immunohistochemistry for ERG, PTEN and SPINK1 was performed in a cohort of localized PCA (n = 340). The data were correlated to pathological and clinical outcomes and compared to Western populations.

Results

ERG expression and PTEN loss were noted in 123/288 (42.7%) and 91/297 (30.6%) of patients, respectively. SPINK1 expression was assessed in a subset of cases, noted in 6/150 (4%) of patients. Only ERG expression was associated with grade groups, being more common in the lower grade groups (1–3 vs 4–5; p = 0.04). In contrast to the Western population, PTEN loss foci were more likely to be ERG negative, observed in 81% of tumor foci and patients with PTEN neg/ERG pos were more likely to exhibit biochemical recurrence (OR 2.831; 95% CI 1.10–726, p = 0.03). This association remained significant in multivariate analysis (OR 2.68; 95% CI 0.98–7.33, p = 0.05), after adjusting for GG, path stage and surgical margin.

Conclusion

This study documents significant differences in key molecular events in PCA in Middle Eastern population compared to Western populations that could explain differences in PCA incidence, progression and prognostication. ERG, PTEN and SPINK1 genomic alteration occur less frequently and the enrichment of ERG for PTEN loss is not observed. Additionally, patients with combined PTEN loss/ERG positive are at highest risk for BCR vs North American Caucasian population where PTEN loss alone seems to be associated with the worst clinical outcome. The data presented here further support differences in clonal evolution between Middle Eastern and Western population in relation to PCA and add further insight to understanding PCA molecular pathways.

Background

The ability to predict disease progression and clinical outcome is a major issue in the management of prostate cancer (PCA) patients. Currently, predicting disease metastasis and clinical progression are one of the main challenges facing physicians in the management of PCA. Currently, several underlying molecular alterations suggested to reflect clonal heterogeneity are well characterized in PCA within the North American and Western populations, including ERG, PTEN and SPINK1 among others (Shoag and Barbieri 2016).

Gene rearrangements between TMPRSS2, which is an androgen receptor-regulated gene located at (21q22.3) and members of ETS family of transcription factors, commonly ERG (21q22.2), are reflective of molecular subtypes of PCA and are proposed to be associated with specific downstream molecular alteration responsible for disease progression (Bismar et al. 2012; Demichelis et al. 2007; Liu et al. 2011; Tomlins et al. 2012). ERG is the most common fusion partner of TMPRSS2 (Perner et al. 2007; Tomlins et al. 2005) detected in approximately 50% of prostate-specific antigen (PSA)-screened prostate cancers (Perner et al. 2006). ETS rearrangements also appear to be more common in peripheral zone than transition zone tumors (Braun et al. 2011). However, the incidence and relation of such molecular alterations in different ethnic groups are not well characterized, yielding conflicting results based on several published reports (Abou-Ouf et al. 2016; Lee et al. 2015; Teng et al. 2013a). ERG has been suggested to be not prognostic in surgical cohorts compared to non-surgically treated cohorts such as those in active surveillance programs or watchful waiting (Pra et al. 2013; Huang et al. 2014a,b; Teng et al. 2013b). The prevalence of ERG expression or gene rearrangements in Western PCA population ranges from 50 to 70% with documented differences between Caucasian-Americans and African-Americans, with ERG being more frequent in Caucasian-Americans (41.9% vs 23.9%) (Mani et al. 2011). In Chinese PCA populations, ERG was noted at much lower frequency of ~ 14% (Jiang et al. 2016). Similar incidence was also observed in patients with Asian descent from Japan, Korea (Furusato et al. 2011) and India (Ateeq et al. 2015).

One previous study performed in Middle East (Jordan) in 2017 with 193 prostatic cancer cases suggested ERG to be present in slightly lower incidence at 33.2% compared to Caucasian–Americans (Aldaoud et al. 2017).

PTEN is one of the most commonly inactivated tumor suppressor genes in prostate cancer (Grasso et al. 2012) and PTEN genomic deletions or protein loss have been associated with several cancer progression including PCA (Bismar et al. 2011; Facher and Law 1998; Hyun et al. 2000). Several studies have documented association between PTEN genomic deletions or protein loss and worse clinical outcome in PCA patients (Ahearn et al. 2016; Guedes et al. 2017; Lotan et al. 2015) which has been gaining significant value in clinical practice given its potential endorsement by the Food and Drug Administration (FDA) as clinically valued biomarker in PCA (Ahearn et al. 2016). Several studies have also documented association between ERG gene rearrangements/expression and increased incidence of PTEN deletions/loss of expression (i.e., there is enrichment of in PTEN deletion in tumor foci with ERG gene rearrangements (Bismar et al. 2011). We also reported significant association of homozygous PTEN genomic deletions and ERG gene rearrangements with lymph node metastasis, hormone-refractory PCA and cancer-specific death (Huang et al. 2016).

The value for ERG/PTEN combined status remains controversial within different ethnicities, with one study from Brazil showing adverse clinical outcome in patients with ERG/PTEN combined genomic aberration, while several others from Western population documented this adverse outcome in patients with only PTEN deletions without ERG gene rearrangements, including one by our group (Reid et al. 2010; Yoshimoto et al. 2008). Furthermore, there is some support that ERG gene rearrangements are more likely observed in lower grade tumors (grade groups 1–3) relative to higher grade ones (grade groups 4–5).

The serine peptidase inhibitor, Kazal type 1 (SPINK1) is a protein, regulated by androgens, overexpressed specifically in a subset of ETS-negative prostate cancers; its overexpression has been associated with decreased biochemical recurrence-free survival (Lippolis et al. 2013). SPINK1 outlier expression has been identified in ~ 10% of prostate cancers and appears to be mutually exclusive from ERG rearrangements (Tomlins et al. 2008). Interestingly, patients harboring these tumors were found to have a shorter time to biochemical recurrence than patients who do not overexpress SPINK1. SPINK1 outlier status, independent of Gleason score, lymph node status, surgical margin status, seminal vesicle invasion, extracapsular extension, and preoperative PSA, has been shown to be a significant predictor of clinical recurrence (Tomlins et al. 2008). SPINK1 is an extracellular secreted protein and therefore is amenable to both therapeutic targeting and non-invasive diagnosis (Ateeq et al. 2011). Only two studies have reported SPINK1 expression in ductal adenocarcinoma of the prostate. They observed that SPINK1 is expressed in about 25% of ductal adenocarcinoma cases and expressed in subset of ETS-negative PCA, and has been associated with aggressiveness (Han et al. 2009; Vinceneux et al. 2017). Another study performed by Pallavi et al. endorsed that SPINK1 is expressed in 22.7% of pure ductal adenocarcinoma and in 43.5% of mixed ductal and adenocarcinoma (Patil et al. 2019). On the other hand, SPINK1 status showed mixed results in relation to patient’s prognosis based on several published reports (Han et al. 2009; Leinonen et al. 2010).

Herein, we comprehensively characterized ERG, PTEN and SPINK1 expression using combined IHC in a cohort of men with PCA patients from Middle Eastern ethnicity population, who were treated by radical prostatectomy as single therapy. We assessed the incidence, overlap and prognostication of individual and combined markers in relation to key clinical and pathological parameter.

Methods

Study population

Our study population consisted of 340 men with PCA from 1998 to 2012 and who were managed by radical prostatectomies. Clinical information on follow-up was available for 329 patients in relation to biochemical recurrence post-radical prostatectomy. Five tissue microarrays (TMAs) were constructed with average four cores (range 2–6) for each patient using manual tissue microarrayer (Beecher Instruments, Silver Spring, MD). The study was approved by the University of Calgary Cumming School of Medicine institutional review board.

ERG, PTEN and SPINK1 status by immunohistochemistry (IHC)

We carried out a dual-color ERG/PTEN and ERG/SPINK1 IHC staining protocol as previously described (Bhalla et al. 2013; Huang et al. 2017). The antibodies used were rabbit monoclonal [ERG at 1:100 dilution (Clone EPR 3864, Epitomics, Burlingame, CA, USA), PTEN monoclonal antibody (138G6, Cell Signaling Technology, Danvers, MA, USA) at 1:25 dilution and SPINK1 antibody (WH0006690M1, Clone 4D4, Sigma-Aldrich, St. Louis, MO, USA) at 1:25 dilution]. Negative control performed by substituting primary antibody with normal mouse pre-diluted serum at 1/200.

Pathological analysis

All cases were re-reviewed and Gleason scores/groups were assessed by the study pathologists based on the 2014 WHO/ISUP criteria. The two common patterns were sampled on the TMAs for evaluation. Both ERG and SPINK1 intensities were assessed as (0; negative; 1, positive) based on previous correlation with ERG rearrangement detected by FISH (data not shown). The expression of ERG in endothelial cells, acted as internal control being highly expressed. PTEN intensity was assessed as (0; negative, 1; weak, 2; moderate and 3; high), with high defined as intensity equal to benign adjacent epithelium. Furthermore, PTEN intensity was further analyzed as intact “positive” vs absent “negative” to avoid smaller numbers within subgroups. Figure 1a–d shows examples of ERG, PTEN and SPINK1 protein expression using combined IHC method.

Fig. 1.

a Immunohistochemistry for ERG positive (brown) and PTEN negative in Gleason grade group 1. b Immunohistochemistry for PTEN positive (purple) and ERG negative in Gleason grade group 1 (note endothelial cells staining brown for ERG acting as internal control). c Immunohistochemistry for ERG positive (brown) and PTEN positive (purple) in Gleason grade group 1. d Immunohistochemistry for SPINK1 positive (red) and ERG negative in Gleason grade group 1 (note endothelial cells staining brown for ERG acting as positive control)

Statistical analysis

Descriptive statistics were used to describe the study variables. Mean and standard deviation were reported for continuous variables and frequency and proportions for the categorical variables. Chi-square tests were to compare two categorical variables. Fisher’s exact test was reported if the cell frequency was less than 5. Binary logistic regression analysis was used for the outcome variable PSA relapse (yes vs. no). Odds ratio (OR) and the corresponding 95% confidence intervals were reported. The association of the combined score of PTEN and ERG with PSA relapse was analyzed using univariate and multivariate models. For the multivariate model, Gleason score, surgical margin and pathological stage were included in the analysis.

SPSS version 25 was used to conduct all the statistical analysis, a p value < 0.05 was used for statistical significance. Two-sided tests were used for all the analysis.

Results

Study population

Clinical features

Median patients’ age was 62 years (range 43–75). 115 of men (34%) were GS = 6, 180 (53%) of GS = 7, 26 (8%) of GS = 8 and 13 (4%) of GS = 9. 243 (73%) of men were pT2 and 91 (27%) of were pT3. 153 patients (45%) had positive surgical margins. Table 1 summarizes clinical and pathological characteristics.

Table 1.

Patients’ demographics of the study cohort

Variables	n (%)
Grade group/Gleason score
GG1 (GS6)	115 (34)
GG2,3 GS7)	180 (53)
GG4 (GS8)	26 (8)
GG5 (GS9–10)	13 (4)
Age	62.1 ± 5.96
Biochemical recurrence
Yes	91 (27)
No	213 (63)
Missing	36 (11)
Stage
pT2	243 (73)
pT3	91 (27)
Surgical margin
Neg	181 (55)
Pos	153 (45)

PTEN, ERG and SPINK1 by immunohistochemistry

ERG positivity was detected in 123/288 (42.7%). PTEN protein expression was positive in 206/297 of cases with (69.3%) ranging from weak to high intensity relative to the adjacent benign prostate epithelium (control).

IHC for SPINK1 was successful for only 2 TMAs (n = 150) due to technical issues, and was expressed in 6/150 cases (4%).

Although there was no significant association between individual grade grouping and each of ERG, PTEN and SPINK1 expression, there was significant correlation between ERG expression and Gleason score when assessed as GG1-3 vs 4–5 with ERG-positive cases occurring in 45% cases of GG1-3 vs. 27% in GG4-5 (p = 0.04). PTEN did not show any significant association within this distribution (p = 0.11). Table 2 summarizes distribution of PTEN, ERG and SPINK1 in various Gleason grade groups and as binary groups.

Table 2.

Distribution of PTEN, ERG and SPINK1 in various grade groups and in binary Gleason groups (GG ≤ 3 vs > 3)

Variables	PTEN negative	PTEN positive	p value
GS 6	27 (29)	35 (71)	0.29
GS 3 + 4	32 (32)	69 (68)
GS 4 + 3	24 (39)	38 (61)
GS 8	3 (14)	19 (86)
GS 9	4 (33)	8 (67)
GS <  = 7	83 (32)	173 (68)	0.11
GS > 7	7 (19)	29 (81)

Variables	ERG negative	ERG positive	p value
GS 6	50 (54)	43 (46)	0.16
GS 3 + 4	49 (53)	44 (47)
GS 4 + 3	35 (61)	22 (39)
GS 8	17 (81)	4 (19)
GS 9	7 (54)	6 (46)
GS <  = 7	135 (55)	110 (45)	0.04
GS > 7	27 (73)	10 (27)

Variables	SPINK negative	SPINK positive	p value
GS 6	40 (28)	3 (50)	0.25
GS 3 + 4	55 (38)	2 (33)
GS 4 + 3	32 (22)	0 (0)
GS 8	12 (8)	0 (0)
GS 9	5 (4)	1 (17)
GS ≤ 7	127 (87)	5 (83)	0.57
GS > 7	19 (13)	1 (17)

Combined makers by immunohistochemistry

To gain insight into the co-expression of markers within same tumor foci, we investigated marker status based on the status within individual core tumor foci. In this cohort, 32/246 (13%) tumor tissue cores were PTEN negative/ERG positive, while 100/246 (41%) of cores showed any PTEN intensity while being ERG negative.

In contrast to published reports from Western population, tumors foci negative for PTEN staining were not enriched for ERG positivity and were more likely to be ERG negative 81% vs 61% ERG-positive tumor foci being PTEN negative (p = 0.06) (Table 3).

Table 3.

Distribution of combined ERG/PTEN in Gleason grade groups

Variables	PTEN Neg/ERG Pos	PTEN Neg/ERG Neg	PTEN Pos/ERG Pos	PTEN Pos/ERG Neg	p value
GS 6	6 (8)	17 (24)	28 (39)	21 (29)	0.06
GS 3 + 4	15 (18)	9 (11)	24 (29)	36 (43)
GS 4 + 3	9 (18)	8 (16)	10 (20)	23 (46)
GS 8	0 (0)	3 (15)	4 (20)	13 (65)
GS 9	2 (17)	2 (17)	4 (33)	4 (33)

Characterizing combined PTEN and SPINK staining status, 29/207 (14%) cores showed PTEN loss, of which 28/29 (96.5%) cores were SPINK1 negative while 1/29 (3.5%) cores was SPINK1 positive. PTEN-positive expression was observed in 178/207 (86%) cores, with 175/178 (98.3%) cores being SPINK1 negative and 3/178 (1.6%) being SPINK1 positive.

As reported before by our group, PTEN-negative tumors were less likely to be SPINK1 positive (3.5% in this cohort) (41).

Investigating combined ERG and SPINK1 staining, 44/217 (20%) cores were ERG positive, none of which was SPINK1 positive. On the other hand, 178/217 (80%) cores were ERG negative, four of which (2%) were SPINK1 positive. These results confirm the mutual exclusivity of ERG and SPINK1.

Key variable distribution of biomarkers stratified by Gleason grade biochemical recurrence

Due to subgroup size and to assess for association with BCR, PTEN was assessed as binary value PTEN negative (absence of staining) vs PTEN positive (presence of any staining). In univariate analysis, PTEN-negative/ERG-positive cases showed higher rates of BCR compared to any other combination (OR 2.831; 95% CI 1.10–7.26, p = 0.03) and remained significant in multivariate analysis after adjusting for Gleason grade grouping, surgical margins and pathological stage (OR 2.68; 95% CI 0.98–7.33), p = 0.05) (Table 4).

Table 4.

Univariate and multivariate analysis of combined status of ERG/PTEN in relation to biochemical recurrence

Variables	OR	95% CI	p value
Univariate analysis of combined status of ERG/PTEN in relation to biochemical recurrence
Combined PTEN and ERG (PTEN pos and ERG neg)
PTEN neg and ERG pos	2.831	1.10–7.26	0.03
PTEN neg and ERG neg	1.17	0.46–2.94	0.74
PTEN pos and ERG pos	1.41	0.68–2.92	0.35
Multivariate analysis of key pathological parameters and combined ERG/PTEN in relation to biochemical recurrence
Combined PTEN and ERG (PTEN pos and ERG neg)
PTEN neg and ERG pos	2.68	0.98–7.33	0.05
PTEN neg and ERG neg	1.34	0.50–3.58	0.56
PTEN pos and ERG pos	1.08	0.49–2.35	0.85
(GG4-5) Gleason score > 7
(GG1-3) Gleason score ≤ 7	0.31	0.14–0.72	0.006
Pathology stage (pT2)
pT3	1.79	0.89–3.60	0.10
Surgical margin neg
Surgical margin pos	2.20	1.13–4.27	0.02

No significant association was found between SPINK1 positive vs negative and BCR (data not shown).

Discussion

The epidemiology of PCA in Middle Eastern Arab populations is not well studied in contrast to North America and Western Europe. This is likely due to the absence of established PSA screening, lack of population-based registries, and existence of prospective oncology databases (Hilal et al. 2015).

In this study, we assessed a large and more homogenous cohort of 340 radical prostatectomies and documented ERG expression to be present in 42% of PCA cases in Middle Eastern Arab population with prognostic implication in association to PTEN loss and SPINK1 expression.

Although the rate of ERG expression detected in this study is at the lower end of that observed in North America and Western Europe, which is documented generally around 45–50% (Darnel et al. 2009), it is still significantly higher than those reported in Asian and African–American populations of about 15% and 27%, respectively (Dong et al. 2014; Khani et al. 2014).

One earlier study by Aldaoud et al. reported on ERG expression in Middle Eastern population suggested the incidence to be lower among Arab population at 33.2% compared to North American and European population and the study failed to document any prognostic value or association to Gleason grade grouping (Aldaoud et al. 2017). The reason for such discrepancy in incidence between our study and that of Aldaoud et al., is that they reported ERG incidence in mixed cases spanning peripheral zone (assessed through needle biopsy) and transition zone (assessed by TURP). This is a significant issue, as it is well known and documented by our group and others that ERG expression is significantly higher in the peripheral zone tumors compared to those in the transition zone (Liu et al. 2011); hence, combining both zones for reporting ERG incidence will artificially suggest lower incidence of ERG.

Loss of PTEN expression is well known to be associated with disease progression and patient’s prognosis in several cancers and has also been implicated as potential targets for PI3K-related therapies (Aldaoud et al. 2017) and PTEN loss, as assessed by a validated and simple IHC protocol, was documented to be associated with a twofold increase in the risk of lethal progression independent of clinical-pathologic parameters (Ahearn et al. 2016). Another study performed by Shan et al. in Arab men with PCA confirmed that PTEN is downregulated and associated with disease progression in PCA (Shan et al. 2017). In the current study, the rate of PTEN loss was noted within the same range compared to Caucasian and African–American cohorts, however, we did not observe significant association between loss of PTEN expression and increased Gleason grade grouping as suggested in North American and Western populations (p = 0.11).

Previous reports in Western and European cohorts suggested significant interplay between ERG positivity and PTEN loss with potential clinical value in various clinical scenarios. However, the prognostic value of combined ERG and PTEN has yielded mixed results. A study from Brazil showed that tumors which are PTEN deleted and ERG gene rearranged were more likely to exhibit biochemical relapse after radical prostatectomy relative to those with no genomic alterations for ERG and PTEN (i.e., not ERG rearranged and PTEN deleted). However, several subsequent studies in North American and European cohorts documented that prostate tumors with PTEN genomic alteration without ERG rearrangements (i.e., those with PTEN deletion or loss of protein expression) are associated with worse clinical outcome (Ahearn et al. 2016; Attard et al. 2009; Bismar et al. 2018). Our results in Middle Eastern cohort document higher BCR in PTEN-neg/ERG-pos tumors similar to the initial data from Brazil.

Similar to earlier reports in Western cohorts, we observed higher frequency of ERG expression in lower grade group 1–3 vs 4–5 (p = 0.04). However, contrary to Western and European cohorts, our results in this Middle Eastern cohort, ERG-positive foci were not enriched for PTEN genomic deletions as reflected by absence of PTEN staining, but showed non-significant trend (p = 0.06).

Most studies reported that SPINK1 protein is detected in about 9% of PCA cases of Caucasian cohorts vs 23% in African–American with some suggestion of prognostic value (Khani et al. 2014; Marrone et al. 2015). In our study, SPINK1 protein expression was noted at much lower incidence of only 4% which is significantly lower than those reported in Western population above. However, we did not observe any prognostic association for SPINK1 which may be due to the fact of lower-case number overall.

Previous studies found SPINK1 protein expression to be inversely associated with PTEN gene deletion and exclusive of loss of PTEN intensity (Huang et al. 2016) and positively associated with higher PTEN protein expression (Flavin et al. 2014). In this study, we also observed notable inverse relation between PTEN-negative intensity and SPINK positivity.

Although some recent studies have reported co-expression of ERG and SPINK1 in up to 4% of prostate cancers (Flavin et al. 2014; Hilal et al. 2015; Huang et al. 2016), the majority of reports confirm ERG and SPINK1 being mutually exclusive (Ateeq et al. 2011), which was also noted in our study.

The above results indicate that molecular landscape of PCA in Middle Eastern population differs from other races. This might be explained by genetic differences, environmental, lifestyle and dietary factors, especially since it has been reported that gut microbiome reflective of diet may have a role on gene expression and cancer risk cancers (Ornish et al. 2008).

The clinical value of our study stems from the fact that pathological staging system and Gleason score do not explain racial difference in PCA outcomes. Therefore, such markers may have a role in the future in personalizing prognosis estimates for patients and guide novel targeted treatment.

Conclusion

ERG, PTEN and SPINK1 status play a significant role in PCA of Middle Eastern population with ERG being expressed at similar rates and SPINK1 expressed at significantly lower rates compared to the Western population. However, in contrast, ERG foci are not enriched for PTEN loss with PTEN-neg/ERG-pos tumor present at a higher frequency in advanced grade grouping and associated with increased risk of BCR. The differences in incidence and correlation between such markers, support different underlying molecular background with respect to ethnic background, and highlight the potential different biology in disease progression which warrant further investigation.

Abbreviations

PCA

Prostate cancer

CRPC

Castrate-resistant prostate cancer

ETS

Erythroblast transformation-specific

PTEN

Phosphatase and tensin homolog

SPINK1

Serine protease inhibitor Kazal-type 1

BCR

Biochemical recurrence

Author contributions

RA, AB, MK, HA performed analysis. IK, TL, MS collected data and performed analysis. SG performed statistical analysis. NP performed, supervised and provided input for manuscript and marker staining. TAB supervised and planned study. All authors have approved the final version of this manuscript.

Funding

This work was supported in part by the Prostate Cancer Foundation, USA. Young Investigator Award and Prostate Cancer Canada, Translational Acceleration Grant and by funds from Calgary Laboratory Services (T.A.B). The Agency provides operating cost funds with no effect on actual research design or outcome.

Compliance with ethical standards

Conflict of interest

All authors have no conflict of interest to declare.

Ethics approval

The study was approved by the University of Calgary Cumming School of Medicine ethics review board and in accordance with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Patient consent was waived by the ethics review board, due to the retrospective nature of the study.

Availability of data and materials

The study is retrospective with samples available from 1998–2008. Data sharing is not applicable.