- Top of page
- PATIENTS AND METHODS
- CONFLICT OF INTEREST
High grade prostatic intraepithelial neoplasia (HGPIN) is generally accepted as a precursor lesion of prostate cancer (PCa) . The frequency of HGPIN in needle biopsy series ranges from 0% to 24.6% with the mean incidence being 7.6% . The clinical importance of HGPIN is its association with PCa in concurrent or subsequent prostate needle biopsies. HGPIN has been associated with 20%–32% risk of PCa in subsequent biopsies [3–5], although the cancer detection rate in re-biopsy has declined in recent studies due to the contemporary use of an extended sampling scheme . The cancer risk of HGPIN in contemporary studies is 20%–25% [2,4,7].
Many studies have attempted to identify clinical, pathological and molecular markers that may identify patients who are at risk of harbouring or developing PCa. Currently published studies have not found utility of clinical parameters, including age, serum PSA levels and abnormal DRE, in such risk prediction . Among the histopathological features of HGPIN, the number of cores containing HGPIN has been shown to correlate with the risk of PCa in subsequent biopsies, with multifocal HGPIN (involving two cores or more) conferring significantly higher cancer risk than unifocal HGPIN [8,9].
Studies have also explored molecular tumour markers that identify men with HGPIN who may be at risk of PCa. Several potential candidate markers, including α-methylacyl-CoA racemase (AMACR) , prostate tumour overexpressed gene 1 (PTOV1)  and adenomatous polyposis coli (APC) , have been proposed. Investigation of genes involved in the carcinogenic transformation of HGPIN to PCa could be helpful in determining patients at risk for PCa in repeat biopsies.
TMPRSS2-ERG fusion, the most common recurrent genetic alteration in PCa [13,14], is highly specific for PCa and present in approximately 40%–50% of PCa. It is also present in approximately 20% of HGPIN glands that are always intermingled with or immediately adjacent to fusion-positive cancer glands [15,16]. Studies have demonstrated that ERG gene fusion plays an important role in early prostate carcinogenesis, including transition of HGPIN to PCa .
Several recent studies have demonstrated that detection of ERG protein by immunohistochemistry in PCa is highly correlated with the ERG rearrangement status by fluorescence in situ hybridization (FISH) or RT-PCR [18–20]. In the present study, we investigated whether ERG gene fusion determined by ERG immunohistochemistry in HGPIN may be of value in stratifying PCa risk in patients with HGPIN. We addressed two main issues: (i) how often are ERG-positive HGPIN glands identified in prostate biopsies that do not harbour PCa; and (ii) does a positive ERG immunostain in HGPIN correlate with PCa detection in subsequent biopsies.
- Top of page
- PATIENTS AND METHODS
- CONFLICT OF INTEREST
The most important clinical significance of an HGPIN diagnosis is the risk of PCa in repeat biopsies in patients with such a diagnosis. Patients with HGPIN may have a concomitant cancer that was missed in the initial biopsy. Patients with HGPIN without concomitant PCa at the time of the initial biopsy may develop PCa during follow-up. Because not all patients with HGPIN harbour or develop PCa, clinical, pathological or molecular markers that may be predictive of PCa risk have been sought.
TMPRSS2-ERG gene rearrangement is highly specific for PCa [13,22,23]. The gene arrangement can be determined by FISH  and quantitative PCR . Recently, several studies have found that immunohistochemistry for ERG protein is a highly specific surrogate for TMPRSS2-ERG gene fusions with a specificity higher than 95% [18–20,25]. Two different antibodies, one mouse monoclonal and one rabbit monoclonal, have been utilized and both are highly correlated with ERG gene rearrangement status . Both antibodies also stain vascular endothelial cells and vascular neoplasms [27,28]. Immunohistochemistry for ERG has been used in the evaluation of prostate biopsy specimens [21,29,30]. ERG gene status determined by immunohistochemistry has been correlated to other clinicopathological and molecular markers for prostate cancer [31–33].
TMPRSS2-ERG gene rearrangement is also identified in approximately 20% of HGPIN glands; the latter almost always intermingle with or are immediately adjacent to the fusion-positive cancer glands, and very rarely are found in other benign prostatic lesions [15,16]. ERG gene fusion plays an important role in early prostate carcinogenesis, including the transition of HGPIN to PCa . It is therefore hypothesized that positive ERG expression in HGPIN may be able to identify ERG-fusion-positive PCa that is missed by the initial prostate biopsy or PCa that develops from ERG-positive HGPIN glands.
To our knowledge, the present study is the first to investigate ERG protein expression by immunohistochemistry in prostate biopsies containing HGPIN and to explore its utility to stratify the cancer risk associated with HGPIN. Of 94 prostate biopsies with HGPIN, only five (5.3%) contained ERG-positive HGPIN glands. This positive rate is much lower than the 20%–25% of HGPIN reported to harbour TMPRSS2-ERG gene rearrangement detected by FISH or RT-PCR in the literature [15,16]. Such a difference is not due to different detection methods (FISH or RT-PCR vs immunohistochemistry) as positive ERG immunostaining using the same anti-ERG antibody as in the present study highly correlated with the ERG gene rearrangement [18–20]. Rather, previous studies investigated ERG gene rearrangement in HGPIN that was intimately associated with ERG-rearranged PCa [15,16]. In contrast, in the present study we have studied isolated HGPIN not associated with PCa. In our prior study that examined ERG expression in limited PCa in prostate biopsies, we found positive ERG expression in 29% of HGPIN, which were all intermingled with or immediately adjacent to the ERG-positive cancer glands. In contrast, HGPIN glands away from the cancer glands were all negative for ERG . In radical prostatectomy specimens, ERG-positive HGPIN glands were almost always adjacent to ERG-positive cancer glands and were rarely found in HGPIN glands that were not associated with or were distant from ERG-positive cancer glands (He et al., manuscript in preparation).
In addition to a low positive rate in HGPIN, positive ERG immunostain does not stratify the cancer risk associated with HGPIN, as the cancer detection rate in ERG-positive HGPIN (2/5, 40%) is not different from that in ERG-negative HGPIN (34/89, 38.2%). Therefore ERG immunostain has no utility in identifying patients with an HGPIN diagnosis who are at increased risk for having PCa. We do not advocate using ERG immunostain to stratify cancer risk when evaluating HGPIN.
Even though the present study did not show that ERG is useful to predict PCa risk after an HGPIN diagnosis, other molecular markers may be of predictive value for PCa risk in such a setting. For example, AMACR , PTOV1 , APC  and testosterone  have been found to be useful to determine which HGPIN patients are at higher cancer risk. Of course, these molecular markers need to be validated in large studies.
The overall PCa detection rate at repeat biopsy in this series was 38.3%, higher than the 20%–32% in other published reports [3–5]. The difference is probably due to the longer follow-up in our series. The mean follow-up in the present study is 27.4 months and almost half of the patients had at least two follow-up biopsies. Longer follow-up detects not only concomitant PCa but also new PCa that develops from HGPIN; therefore the incidence of PCa detection increases with the number of repeat biopsies and follow-up duration [35,36]. As Lefkowitz et al. [37,38] reported, the risk of PCa after initial unifocal HGPIN increased from 2.3% at 1-year follow-up to 25.8% at 3-year follow-up.
Many studies have found that clinical parameters, including a patient's age, various forms of serum PSA measurement and abnormal DRE, do not correlate with the cancer risk associated with HGPIN in subsequent biopsy [5,36]. In the present study, we did not find that a patient's age and the number of repeat biopsy sessions were associated with different cancer risk. The multifocality of HGPIN seems to confer different cancer risks with cancer detected in 39% of cases with unifocal HGPIN and in 60% of cases with multifocal HGPIN, although this difference is not statistically significant (P= 0.128), probably due to small sample size.
In conclusion, this is the first study to investigate ERG protein expression in prostate biopsies containing HGPIN only and its use to stratify the cancer risk associated with HGPIN. We found that ERG expression in isolated HGPIN is very uncommon (5.3%). In addition, a positive ERG expression is not associated with increased cancer detection in subsequent repeat biopsy. We therefore do not advocate using ERG immunostain in the evaluation and cancer risk stratification of HGPIN.