Methods—Immunostaining with the anti-GST-π antibody was performed on 20 high grade prostatic intraepithelial neoplasia samples of the transition zone, either isolated or associated with prostatic carcinoma (groups 1 and 2, respectively; 10 cases each) and on 20 high grade prostatic intraepithelial neoplasia samples of the non-transition zone, either isolated or associated with prostatic carcinoma (groups 3 and 4, respectively; 10 cases each). This study also included six samples of high grade prostatic intraepithelial neoplasia simultaneously present in the transition and non-transition zones and not associated with prostatic carcinoma (group 5). The presence of immunostaining, staining intensity, and the distribution of immunostaining were evaluated in the high grade prostatic intraepithelial neoplasia lesions and in the normal tissue and cancer areas.

Results—The GST-π antibody stained the cytoplasm of the cells lining the ducts and acini of normal prostate tissue. Staining was stronger and more diffuse in the basal cell layer than in the luminal (or secretory) cell layer. Immunohistochemical staining with anti-GST-π antibodies failed to detect the enzyme in all prostatic carcinoma foci but one. Two patterns were detected in high grade prostatic intraepithelial neoplasia. One was represented by GST-π staining similar to that of the normal tissue (pattern A). The other deviated from it and was characterised by absence of GST-π expression in the secretory cells and abundant expression in scattered basal cells (pattern B). Pattern A staining was seen more frequently in the transition than in the non-transition zone. Pattern B staining was seen mainly in high grade prostatic intraepithelial neoplasia of non-transition zone associated with cancer.

Conclusions—The differential expression of GST-π in the transition and non-transition zones indicates the existence of two populations with the morphological appearance of high grade prostatic intraepithelial neoplasia that might have different associations with carcinoma.

The glutathione S-transferases (GSTs) are a family of multifunctional isoenzymes that catalyse the nucleophilic addition of glutathione to a heterogeneous group of compounds.⁸ GST isoenzymes can be characterised on the basis of substrate specificity, sensitivity to inhibitors, isoelectric points, immunological crossreactivity, and amino acid composition and sequence. A variety of biological functions have been ascribed to this enzyme group, including intracellular binding and transport of lipophilic compounds such as bile products, steroid hormones, drugs, and xenobiotics. By far the most studied function of the GST enzymes is their role in cellular detoxification,⁹ primarily against oxygen free radicals and peroxides produced by cellular physiological processes and exogenous stimuli. GSTs might prevent carcinogenesis through inactivation of reactive electrophiles by conjugation to reduced glutathione.¹⁰ Human GSTs have been classified into five families: α, μ, θ, π, and microsomal.¹¹ The most widely investigated with respect to human malignancy is GST-π.^12–25

Loss of expression of GST-π has been shown in prostatic carcinoma and in high grade prostatic intraepithelial neoplasia.^26–33 This finding supports a role for high grade prostatic intraepithelial neoplasia as a precursor lesion, and might provide insight into the molecular pathogenesis of prostate cancer. To the best of our knowledge there are no previous studies evaluating the expression of this enzyme in prostatic intraepithelial neoplasia from the different prostate zones.

The aim of our study was to compare the immunohistochemical expression of GST-π in prostatic intraepithelial neoplasia of the transition zone with that of the non-transition zone. Because high grade prostatic intraepithelial neoplasia is more closely correlated with prostatic carcinoma than the low grade form, and thus more likely to represent a proximate precursor lesion, we focused on high grade prostatic intraepithelial neoplasia for the evaluation of GST-π expression.³⁴

The samples were selected to represent the following five groups (the age of the patients in all the groups was comparable and ranged from 60 to 70 years): high grade prostatic intraepithelial neoplasia samples of the transition zone, either isolated or associated with prostatic carcinoma (groups 1 and 2, respectively; 10 cases for each group); high grade prostatic intraepithelial neoplasia samples of the non-transition zone, either isolated or associated with prostatic carcinoma (groups 3 and 4, respectively; 10 cases for each group); and samples of high grade prostatic intraepithelial neoplasia simultaneously present in the transition and non-transition zones (group 5; six cases) (table 1 ). The following criteria were adopted in the identification of high grade prostatic intraepithelial neoplasia^1–3: crowding and stratification of the secretory cells with cytological changes mimicking cancer, including enlargement of nuclei and presence of prominent nucleoli, the latter often being multiple and eccentrically positioned. The basal cell layer at the periphery has to be recognisable. In each case, the slide with the largest focus of high grade prostatic intraepithelial neoplasia was selected for immunohistochemistry. The mean diameter of the prostatic intraepithelial neoplasia foci evaluated in our study was 1.8 mm, the individual values ranging from 0.5 to 2.5 mm. The extension of prostatic intraepithelial neoplasia in the five groups was comparable. In those groups where cancer was present, this was always localised to the prostate (that is, pT2).

Table 1 Group composition

All the sections were examined by light microscopy to assess qualitatively the presence or absence of immunostaining, and its distribution. Among the tissue that stained positive, a distinction was made between staining of the basal cell layer and the secretory acinar epithelium. Fisher's exact test was used for statistical analysis of the results.

The anti-GST-π antibody stained the cytoplasm of the cells lining the ducts and acini. In general, staining was stronger and more diffuse in the basal cell layer than in the secretory cell layer (figs 1 and 2 ). Focal and weak nuclear staining was present in the basal epithelial layer of 40% of cases. Short gaps in the immunostained basal cell layer were present. Some variability in the expression of GST-π in the benign ducts and acini was seen within each zone and also between the prostate zones. Sections from five cases selected at random from the five groups were also immunostained for cytokeratin 34βE12. The extension of the basal cell layer staining with the two markers was comparable. In no instance was secretory layer staining for GST-π present in the absence of basal layer staining. There was no appreciable difference in staining intensity between those secretory cells that faced or did not face an intact basal cell layer. The prostatic stroma was not immunoreactive, whereas the urothelium and squamous metaplastic epithelium were.

Figure 1 Normal prostate tissue. The antiglutathione S-transferase π-class (GST-π) antibody stained the cytoplasm and the nuclei of the basal cell layer.

Figure 2 Normal prostate tissue. Glutathione S-transferase π-class (GST-π) staining is stronger in the basal cell layer than in the secretory cell layer.

Prostate cancer was present in the samples included in group 2 (prostatic intraepithelial neoplasia associated with prostatic carcinoma in the transition zone) and group 4 (prostatic intraepithelial neoplasia associated with prostatic carcinoma in the non-transition zone). Prostatic carcinoma with Gleason pattern ranging from 2 to 3 was observed in the former group and with Gleason pattern ranging from 3 to 5 in the latter. Immunohistochemical staining with anti-GST-π antibodies failed to detect the enzyme in all prostatic carcinoma areas apart from one. Weak staining of approximately 15% of cells was observed in a single focus with Gleason pattern 2. It was confirmed to be prostatic carcinoma by the absence of a basal cell layer with cytokeratin 34βE12 immunostaining. Cells undergoing either mitotic division or apoptosis did not express GST-π.

Two discrete immunostaining patterns were seen in high grade prostatic intraepithelial neoplasia. GST-π expression similar to that of the normal tissue; that is, stronger and more diffuse in the basal cell layer than in the luminal cell layer, was defined as pattern A (figs 3 and 4 ). The basal cell layer stained at the periphery was prominent at low magnification, partially and completely encircling acini and ducts containing prostatic intraepithelial neoplasia. The other pattern deviated from the normal and was characterised by complete absence of GST-π expression in the secretory cells and abundant expression in scattered basal cells (pattern B) (figs 5 and 6 ). This pattern was identical to that observed by Cookson et al.²⁷ Coexistence of the two patterns in individual foci was not seen. A pattern intermediate between these two was seldom seen. Bostwick and colleagues³⁴ identified four architectural patterns of high grade prostatic intraepithelial neoplasia: tufting, micropapillary, cribriform, and flat. In our material, epithelial tufting was the most common architectural pattern, being seen in 97% of the prostatic intraepithelial neoplasia foci. There was no relation between the pattern of GST-π expression and prostatic intraepithelial neoplasia architecture.

Figure 3 High grade prostatic intraepithelial neoplasia with diffuse glutathione S-transferase π-class (GST-π) expression in the basal cell layer (pattern A).

Figure 4 High grade prostatic intraepithelial neoplasia with diffuse glutathione S-transferase π-class (GST-π) expression in the basal cell layer. Scattered secretory cells (see arrow) show some positivity, mainly in the supranuclear region (pattern (more ...)

Figure 5 High grade prostatic intraepithelial neoplasia with complete absence of glutathione S-tansferase π-class (GST-π) expression in the secretory cells and abundant expression in scattered basal cells (pattern B).

Figure 6 High grade prostatic intraepithelial neoplasia with complete absence of glutathione S-transferase π-class (GST-π) expression in the secretory cells and abundant expression in scattered basal cells (pattern B).

When the high grade prostatic intraepithelial neoplasia lesions were divided according to whether they were either isolated or associated with prostatic carcinoma, the pattern A of GST-π expression was prevalent in the samples of high grade prostatic intraepithelial neoplasia that were isolated and not associated with prostatic carcinoma. In addition, when considering the zonal distribution, high grade prostatic intraepithelial neoplasia with pattern A was prevalent in the transition zone, the differences being significant. This indicates that pattern B was seen mainly in cases of high grade prostatic intraepithelial neoplasia associated with prostatic carcinoma and present in the non-transition zone. In groups 2 and 4, high grade prostatic intraepithelial neoplasia was either adjacent to or remote from cancer; pattern B was prevalent in high grade prostatic intraepithelial neoplasia adjacent to prostatic carcinoma in the non-transition zone, although the differences were not significant (tables 2 and 3 ). In particular, when pattern B was observed in high grade prostatic intraepithelial neoplasia lesions of the transition zone, the same pattern was also present in the non-transition zone of the same prostates (table 4 ). There was no correlation between the GST-π pattern of staining in high grade prostatic intraepithelial neoplasia and the Gleason pattern. High grade prostatic intraepithelial neoplasia with early invasion—that is, with transition to prostatic carcinoma—was present in one sample from group 4. This lesion showed pattern B staining.

Table 2 Expression of π-class glutathione S-transferase (GST-π) in high grade prostatic intraepithelial neoplasia (PIN)

Table 3 Expression of π-class glutathione S-transferase (GST-π)) in high grade prostatic intraepithelial neoplasia (PIN) subdivided according to the proximity to developing prostatic carcinoma (PCa)

Table 4 Confusion matrix of the expression of π-class glutathione S-transferase (GST-π) in high grade prostatic intraepithelial neoplasia (PIN) in group 5

In some whole mount sections, prostatic intraepithelial neoplasia foci smaller than those evaluated in our study were present. In general, these additional foci showed GST-π expression similar to that of the main foci whose evaluation was detailed above (data not shown).

The expression of GST-π in normal prostate tissue specimens, including those from patients with benign prostatic hyperplasia, has been investigated in previous studies.^{14,27,33–36} Their findings were confirmed in our investigation, where diffuse cytoplasmic immunostaining in the basal cell layer was seen, and only a minority of secretory cells were stained. Although most of the GST-π was cytosolic, nuclear staining was reported by Moskaluk et al,³¹ who found that approximately 50% of cases also exhibited focal nuclear staining, primarily in the basal epithelial layer. Our findings on nuclear staining were similar to those reported by Moskaluk et al.³¹

The benign ducts and acini evaluated in our study showed some variability in GST-π expression and in the proportion of stained cells both within and between the prostate zones. However, the staining of basal and secretory cells in the transition and non-transition zones was qualitatively similar. This is in agreement with a previous study by our group.³⁶ No substantial differences were seen between the samples from patients with benign prostatic hyperplasia and those derived from patients unaffected by this disease. Our findings are in agreement with the data derived from a biochemical study performed by Tew et al.³⁵

De Marzo and colleagues³⁷ proposed that most proliferating cells in normal prostate and benign prostatic hyperplasia tissue reside in the basal cell compartment. Higher amounts of gene products that appear to have genomic protective features, such as GST-π, are found in the basal cells. Other potentially protective molecules are preferentially expressed in basal cells versus secretory cells, including glyceraldehyde-3-phosphate dehydrogenase, which has been shown to be involved in DNA repair, and pp32, which inhibits neoplastic transformation in vitro.³⁷ Therefore, with these protective functions intact, the basal cells are immune from the acquisition of multiple genomic changes and, hence, the extreme rarity of prostatic basal cell carcinoma.

The results of our study and of previous investigations^{14,27,33–36} have shown that only a minority of cells in the secretory compartment stain for GST-π, and only weakly. It has been speculated that in normal prostate tissue the basal cells exert some protective influence on the secretory cell compartment. When basal cells are not present, such as in the gaps present in the basal cell layer, the luminal cells might take up this protective role.³⁶ It is possible that the early loss of GST-π expression in human prostate cells might compromise their electrophilic defences, making them vulnerable to the accumulation of the genetic damage necessary to foster the neoplastic transformation.

Prostatic carcinoma is thought to initiate from an abnormal increase in replication of transiently proliferating cells within the secretory compartment that are poorly protected against DNA damage.³⁷ Although these cells have partial differentiation ability, they abnormally retain or acquire stem cell like features of unlimited self renewal. The immortal nature of the expanding clone, which is proliferating without adequate DNA protection, allows the accumulation of additional genetic damage and genetic instability, thus resulting in the development of prostate cancer. Previous investigations showed that immunohistochemical staining with anti-GST-π antibodies failed to detect the enzyme in most untreated prostatic adenocarcinomas, despite the presence of abundant staining in normal prostatic epithelial cells and in cells making up benign proliferative prostatic lesions.^26,27,31,33 For example, Cookson and colleagues²⁷ reported GST-π expression in only 4% of cases. The almost complete absence of GST-π expression in prostatic carcinoma seen by others was also confirmed in our study. Of the 20 prostatic carcinomas tested, only one was focally positive for GST-π. The lack of GST-π expression among prostate carcinomas appears to be independent of tumour biology and again there appears to be little prognostic information available because even incidental tumours lack staining in almost all cases.

The lack of GST-π expression in prostatic carcinoma has been linked to the fact that the regulatory sequences near the gene GSTP1, which encodes the human π-class GST, often appear to be hypermethylated during prostatic carcinogenesis.^28–30 These studies identify changes in GST-π expression as the most common genetic alteration in the process of prostate cancer development. Total androgen ablation does not restore the expression of this enzyme, indicating that the genetic defect detected in prostate cancer is not reversible under androgen manipulation. A different point of view was expressed by Cookson et al,²⁷ who questioned the hypothesis of the role of GST-π in the malignant transformation of prostatic acinar epithelium. The lack of GST-π expression seen in prostate cancer might be related to the absence of the basal cell layer. If the basal cell is the precursor stem cell of the secretory acinar cell, it is possible that GST-π is necessary to maintain the basal cell phenotype and is lost as it differentiates into an acinar cell. However, this view is not shared by other researchers who favour the role of GST-π in prostate carcinogenesis through an intermediate step defined as high grade prostatic intraepithelial neoplasia.^28–30

GST-π expression in high grade prostatic intraepithelial neoplasia was investigated in previous studies. Cookson and colleagues²⁷ reported GST-π expression in only one of 17 high grade prostatic intraepithelial neoplasia lesions. Similarly, Lee and colleagues³⁰ reported detecting heterogeneity in GST-π expression in a small number of prostatic intraepithelial neoplasia lesions in which occasional cells displayed GST-π expression. Subsequently, Lee's group reviewed the specimens and raised the possibility that the cells expressing GST-π might have been basal cells or normal columnar epithelial cells interspersed in the foci of prostatic intraepithelial neoplasia. Moskaluk and colleagues³¹ found absent GST-π expression in 12 of 12 high grade prostatic intraepithelial neoplasia specimens. Similarly, GST-π expression was completely absent from 60 independent high grade prostatic intraepithelial neoplasia foci analysed by Brooks et al.²⁶ In this last study, lack of GST-π expression in high grade prostatic intraepithelial neoplasia was associated with CG island methylation changes near the GSTP1 gene in seven of 10 high grade prostatic intraepithelial neoplasia lesions. These authors speculated that prostatic intraepithelial neoplasia lesions that lack GSTP1 promoter methylation might never progress to carcinoma. In addition, they suggested that the detection of methylated GSTP1 promoter sequences might serve as a useful biomarker of high grade prostatic intraepithelial neoplasia lesions at risk for progression.²⁶

We found two staining patterns in high grade prostatic intraepithelial neoplasia lesions. One was similar to the pattern of normal tissue, whereas the other differed from it consistently. Our study on GST-π expression in high grade prostatic intraepithelial neoplasia shows some discrepancy with the observations made by Brooks and colleagues²⁶ and Cookson et al,²⁷ whereas it is partially in agreement with the information given by Moskaluk and colleagues³¹ and Lee et al.^29,30 Such differences could be explained on the basis of the type of prostatic intraepithelial neoplasia studied. We selected high grade prostatic intraepithelial neoplasia samples (some of which were associated with cancer whereas others were not) belonging to different prostate zones. The aforementioned authors always selected foci of high grade prostatic intraepithelial neoplasia present in prostates from patients with prostatic carcinoma, and the zonal distribution of these lesions was never considered.^{26,27,29–31}

Even though our study did not include molecular testing of GST-π expression, the immunohistochemical results point towards the identification of two populations with the morphological appearance of high grade prostatic intraepithelial neoplasia. Those prostatic intraepithelial neoplasia lesions whose GST-π expression is substantially different from that seen in normal tissue—namely, the cases of non-transition zone either adjacent to or remote from prostatic carcinoma—are closely associated with cancer. By contrast, enzyme expression in prostatic intraepithelial neoplasia lesions of the transition zone is very similar to that seen in normal tissue, and these lesions have a different association with cancer. This is in keeping with the results of a recent publication by Harvei et al.⁷ These authors investigated prostatic intraepithelial neoplasia in the transition/central zone as a marker for subsequent development of prostatic carcinoma. They did not find any association between the presence of prostatic intraepithelial neoplasia in this zone and subsequent cancer development, thus questioning the precursor nature of prostatic intraepithelial neoplasia at this location. A recent publication by Weinstein gives further support for the existence of two distinct populations of high grade prostatic intraepithelial neoplasia.³⁸ According to this author, the categorisation of prostatic intraepithelial neoplasia based on the degree of proliferation measured by digital image analysis has biological relevance.

In conclusion, the differential expression of GST-π in the transition and non-transition zones indicates the existence of two populations with the morphological appearance of high grade prostatic intraepithelial neoplasia but with different relations to carcinoma. Quantitative analytical studies using image analysis systems are in progress and should provide us with additional information on the activity of GST-π.

This research has been supported by a grant from the University of Ancona and, in part, by a grant from the National Cancer Institute, R 35 CA 53877 (PHB). The content of this paper is solely the responsibility of the authors and does not necessarily represent the official views of the University of Ancona and the National Cancer Institute.