Rodolfo Montironi, Uropathology Section, Institute of Pathological Anatomy and Histopathology, Polytechnic University of the Marche Region (Ancona), School of Medicine, United Hospitals, I-60126 Torrette, Ancona, Italy. e-mail: email@example.com
What's known on the subject? and What does the study add?
High grade prostatic intraepithelial neoplasia (HGPIN) is the precursor lesion to some forms of adenocarcinoma.
HGPIN should not be confused with intraductal carcinoma. The latter, even when isolated in a prostate biopsy, carries a predictive value of 100% for cancer in a repeat biopsy. Multifocal HGPIN, when isolated in a prostate biopsy, still carries a high predictive value for carcinoma in repeat biopsy. Adenosis and PIA are not considered as precursor lesions.
The aim of the present paper was to review the morphological spectrum of prostatic intraepithelial neoplasia (PIN), its relationship to carcinoma of the prostate (PCa) and its clinical significance. We reviewed the literature on premalignant lesions of the prostate, with an emphasis on high grade prostatic intraepithelial neoplasia (HGPIN). HGPIN is the most likely precursor of PCa, according to almost all available evidence. HGPIN is characterized by cellular proliferations within pre-existing ducts and acini, with nuclear and nucleolar enlargement similar to PCa. The clinical importance of recognizing HGPIN is based on its association with PCa. In recent years, a significant decline from 36% to 22% in the predictive value of cancer after an initial diagnosis of HGPIN. A major factor contributing to the decreased incidence of cancer after a diagnosis of HGPIN on needle biopsy in the contemporary era is related to increased needle biopsy core sampling, which detects many associated cancers on initial biopsy. Some recent studies have suggested that molecular findings associated with HGPIN might be able to predict which men are more likely to have cancer on re-biopsy.
high grade PIN with adjacent small atypical glands
Prostatic intraepithelial neoplasia (PIN), as a precursor to some prostatic carcinomas, was first described in the 1960s by McNeal under the name of ‘intraductal dysplasia’, and was more precisely characterized in 1986 by McNeal and Bostwick . PIN consists of pre-existing prostatic ducts and acini lined by cytologically atypical cells and is subdivided into low grade and high grade PIN (LGPIN and HGPIN). In its original description, PIN was subcategorized into three grades, with grade 1 equating to LGPIN and grades 2–3 combined into HGPIN.
Prostatic intraepithelial neoplasia is found predominantly in the peripheral zone of the prostate (75–80%), rarely in the transition zone (10–15%), and very rarely in the central zone (5%). This distribution mirrors the frequency of the zonal predilection for carcinoma of the prostate (PCa). The frequency of HGPIN in needle biopsy series ranges from 5 to 16%. HGPIN is relatively uncommon in TURP, with studies reporting a rate of 2.3% and 2.8%, respectively [2,3]. The prevalence of HGPIN in radical prostatectomy specimens is remarkably high; it was present in 85–100% of specimens, reflecting the strong association between the lesion and PCa .
There are two other possible findings in the prostate that may be premalignant, adenosis (atypical adenomatous hyperplasia) and proliferative inflammatory atrophy (PIA). The data for these are much less convincing than those for HGPIN. If adenosis and PIA are removed from the list of precursor lesions of PCa, there remains only one well established precursor to PCa, that is HGPIN [3–5].
MORPHOLOGY OF PIN
Prostatic intraepithelial neoplasia can be identified at low magnification by three important characteristics: (i) a darker lining of the ductal structures; (ii) a lining thicker than the surrounding normal ducts and acini, and (iii) a complex intraluminal pattern of growth. At high magnification, there are three cytological characteristics: (i) varying degrees of nuclear enlargement with nuclear stratification; (ii) hyperchromasia; and (iii) nucleolar prominence [3–5].
The classification of PIN into low grade and high grade is chiefly based on the cytological characteristics of the secretory cells. The nuclei of the cells that make up LGPIN are enlarged, vary in size, have a normal or slightly increased chromatin content, and possess small or inconspicuous nucleoli. HGPIN is characterized by cells with large nuclei of relatively uniform size, an increased chromatin content, which may be irregularly distributed, and prominent nucleoli that are similar to those of carcinoma cells (Fig. 1). Mitotic figures are rare in HGPIN and are not included in the grading criteria for PIN.
The basal cell layer, as best shown using immunohistochemical investigations with antibodies directed against the nuclear p63 and against the cytoplasmic high molecular weight cytokeratin 34betaE12, is intact or rarely interrupted in LGPIN, but may have frequent disruptions in HGPIN. Similarly to PCa, the cytoplasm in most of the HGPIN cases is positively stained with an antibody directed against alpha-methyl acyl-Co A racemase (AMACR) [2,4,6].
There is inversion of the normal orientation of epithelial proliferation with HGPIN. Proliferation (evaluated using the Ki67 antibody) occurs in the basal cell compartment in the benign epithelium, whereas in HGPIN proliferation predominantly occurs on the luminal side of the ducts and acini .
Although the cytological features of LGPIN and HGPIN are fairly constant, the architecture shows a spectrum varying from a flattened epithelium to a florid cribriform proliferation. There are four main patterns of HGPIN (Fig. 2): a tufting pattern (87%); a micropapillary pattern (85%); a cribriform pattern (32%); and a flat pattern (28%). There are no known clinically important differences between the architectural patterns of HGPIN, and their recognition appears to be useful only for diagnostic purposes.
Variants of HGPIN include signet ring-cell, small cell neuroendocrine, mucinous, foamy, inverted, and a variant with squamous differentiation (Fig. 3). The presence of HGPIN with various histological patterns further supports the hypothesis that there is a close relationship between HGPIN and the variants of PCa [2,4].
The central zone glands (Fig. 4) are architecturally more complex than the peripheral and transition zone glands and exhibit a certain degree of nuclear stratification that may be interpreted as PIN. In addition, bridging, papillary formation with a central vascular core, and focal tubular or cribriform patterns may be present. The central zone is frequently found in core biopsies from the base of the prostate. The site of the biopsy as well as the presence of the other histological features of the central zone, including the presence of abundant compact muscular stroma, should confirm that the ‘epithelial atypia’ is a normal finding in this location.
Clear cell cribriform hyperplasia (CCCH) is typically located in the transition zone and typically, although not always, sampled on TURP, whereas HGPIN predominates in the peripheral zone. CCCH consists of crowded cribriform glands with pale cytoplasm sometimes growing as a nodule and in other instances more diffusely. The key feature distinguishing CCCH from HGPIN is the lack of nuclear atypia. Furthermore, in a nodule of CCCH at least some of the cribriform glands show an obvious basal cell layer.
Basal cell hyperplasia (BCH) may be mistaken for HGPIN as both can have prominent nucleoli and mitotic activity. This unfavourable cytology may be seen in several architectural patterns of BCH, including: (i) proliferation of small round crowded basaloid glands with atrophic cytoplasm or even solid nests; (ii) proliferation of basal cells in pre-existing larger open benign glands; and (iii) cribriform and pseudo-cribriform BCH. In contrast to BCH with small crowded glands, HGPIN resembles usual prostate glands in their larger size, open lumina, abundant cytoplasm, and distribution. BCH reveals high molecular weight cytokeratin or p63 positivity in multilayered nuclei, although in some cases the more centrally located cells are not immunoreactive. In HGPIN, basal cell markers label only flattened cytologically benign basal cells beneath negatively stained atypical PIN cells. Finally, most cases of BCH are found in TURP specimens, indicating growth in the transition zone, in contrast to HGPIN's preferential location in the periphery.
Seminal vesicle or ejaculatory duct epithelium may be found in TURP specimens or needle biopsies. Helpful features in distinguishing these normal structures from HGPIN include the presence of variably sized nuclei (round to oval), often with large ‘monstrous’ cells containing intranuclear inclusions, and the presence of cytoplasmic pigment (lipofuscin). It should be noted that lipofuscin is also present in prostatic cells. In case of doubt, the ejaculatory duct/seminal vesicle epithelium is not immunoreactive to PSA or prostatic acid phosphatase.
Transitional cell metaplasia of ducts and acini is characterized by elongated, oval-shaped nuclei that often contain longitudinal nuclear grooves. When transitional cell metaplasia is more immature than usual, some nucleoli may be present; it may therefore also be confused with HGPIN. In general, there is no nuclear enlargement [2,4,8].
Intraductal carcinoma is characterized by a proliferation of malignant epithelial cells filling large acini and ducts with complete or incomplete retention of basal cells. Marked pleomorphism is present with large hyperchromatic nuclei that are six times larger than those of normal ductal epithelium. Intraductal carcinoma of the prostate may show solid, dense cribriform, loose cribriform and micropapillary patterns. Intraductal carcinoma of the prostate generally occurs in conjunction with invasive cribriform or ductal adenocarcinoma of the prostate.
While the most common forms of invasive ductal adenocarcinoma mimic micropapillary and cribriform HGPIN, ductal adenocarcinoma may be composed of simple glands lined by stratified columnar epithelium with cytological and architectural features of flat and tufting HGPIN, i.e. PIN-like ductal cancers. PIN-like ductal cancers are distinguished from HGPIN either because the atypical glands are too crowded to represent HGPIN or there are too many atypical glands that are negative for basal cell markers to be consistent with HGPIN. In some cases, there are only a few atypical glands with papillary fronds that are negative for basal cell markers that are highly suspicious for ductal adenocarcinoma, yet because of the limited number of atypical glands HGPIN cannot be ruled out with certainty. Repeat biopsy is recommended in these cases [9,10].
Features helpful in distinguishing TCC involving ducts and acini from HGPIN include the occasional presence of a residual luminal cell layer and cytological characteristics. The cells of TCC usually vary significantly in size and shape and have a very coarse chromatin pattern, significant mitotic activity and frequent tumour necrosis. In lower grade TCC, the cells may show longitudinal nuclear grooves. Because TCC in situ of the bladder frequently shows pagetoid spread, prostatic ducts involved by TCC frequently have a basal cell layer, as in HGPIN. Mitotic figures are frequently present in high grade TCC and in prostate carcinoma of ductal origin, and are rare in cribriform carcinoma and in HGPIN .
RELATIONSHIP TO CARCINOMA
There is epidemiological, morphological, and molecular evidence that HGPIN is a precursor lesion to some carcinomas of the prostate (Tables 1,2). (See Montironi et al.  for an extensive review of the relationship of HGPIN to PCa.)
Table 1. Epidemiological and morphological evidence linking HGPIN and prostate cancer
• Incidence and extent of HGPIN and prostate increase with advancing age
• The frequency, severity and extent of HGPIN increases in the presence of prostate cancer
• HGPIN and prostate cancer are both multifocal and share similar locations in the prostate zones
• Transition of HGPIN to prostate cancer can be observed from the morphological point of view
• HGPIN has several features similar to prostate cancer
Table 2. Molecular evidence linking HGPIN and prostate cancer
• Rates of cell proliferation and death are higher in HGPIN and prostate cancer than those in normal prostates
• HGPIN and prostate cancer are phenotypically similar
• HGPIN and prostate are morphometrically similar
• Some genetic and molecular alterations are common to HGPIN and prostate cancer
• The basal cell layer is disrupted in HGPIN and is absent in prostate cancer
• The rate of neovascularization is higher in HGPIN and prostate cancer than that in normal prostates
Frequent changes in both PCa and HGPIN are losses of chromosome 8p and gains of 8q. Chromosomal losses in HGPIN and PCa are also seen with some frequency in 10q, 16q, and 18q with gains of chromosomes 7, 10, 12, and Y. Overall, HGPIN has greater aneuploidy than benign prostate tissue, although somewhat less than seen in some invasive carcinomas. Telomere shortening and increased telomerase activity are both seen in HGPIN and PCa. Glutathionine S-transferase P1 is hypermethylated in many HGPIN glands, which mirrors the process seen with PCa. Overexpression in some cases of both HGPIN and PCa are seen in p16, p53, Bcl-2, MYC, AMACR, as well as many other genes. Decreased expression of genes is also found in some cases of both HGPIN and PCa, including NKX3.1 and p27. About 20% of HGPIN lesions harbour a TMPRSS2-ERG fusion gene, which is a common molecular abnormality detectable in about 50% of prostate cancers . Proliferative and apoptotic rates are higher in HGPIN, analogous to that seen in adenocarcinoma compared with benign glands.
LGPIN ON BIOPSY
Low grade PIN on biopsy should not be listed in pathology reports because there is a lack of reproducibility in its diagnosis, even by uropathologists, and it is not associated with a higher risk of cancer on re-biopsy than the risk after a benign diagnosis on initial biopsy.
HGPIN ON BIOPSY
There is marked variation in the literature on the incidence of isolated HGPIN on needle biopsy, ranging from 0 to 24.6%, the mean incidence being 7.6% with a median value of 4.7%. The most likely explanation for the observed variation in the incidence of HGPIN relates to the vague definition of HGPIN. For example, there are no criteria as to how prominent or how frequent the nucleoli must be to diagnose HGPIN. Different thresholds for the diagnosis of HGPIN include: (i) any visible nucleoli; (ii) nucleoli visible in at least 10% of the cells in the gland; (iii) complete involvement of a gland with cells having nucleoli; and (iv) nucleoli visible at 20× magnification. Technical factors relating to the processing of needle biopsy specimens can also contribute to the reported variability in the incidence of HGPIN on biopsy. Fixatives that enhance nuclear detail and nucleolar prominence can increase the diagnosis of HGPIN, whereas thick sections and increased uptake of dyes can obscure fine nuclear detail. Although one study has reported that African-American men have a higher incidence of HGPIN than Caucasian men, this by itself is an unlikely explanation for the marked variation seen in the literature [2,4].
Risk of cancer on re-biopsy
Studies from the early 1990s on relatively few cases reported a 50% risk of cancer after the diagnosis of HGPIN. These earlier studies used sextant biopsies on the initial biopsy that missed associated cancers, resulting in only HGPIN on the initial biopsy. On re-biopsy, some of these initially missed cancers were detected, yielding a high re-biopsy risk of cancer. More contemporary data report that the median risk of cancer after a diagnosis of HGPIN on biopsy is only 22%. This is similar to the median risk of finding cancer in a repeat biopsy after a benign diagnosis, which is 15–19%. Of 12 publications that have examined the risk of cancer on re-biopsy after a needle biopsy diagnosis of HGPIN compared with that after a benign diagnosis, nine showed no statistically significant difference. In these contemporary studies, sampling more extensively on the initial biopsy detects many associated cancers, such that when only HGPIN is found they often truly represent isolated HGPIN; re-biopsy, therefore, even with good sampling, does not detect many additional cancers. The importance of sampling can be seen in the study by Eskicorapci et al.  where the risk of cancer after an initial sextant biopsy showing HGPIN was 56.5% and was significantly more than that after a benign diagnosis. By contrast, Eskicorapci found that the risk of cancer after an extended biopsy (10 cores) showing HGPIN was only 22.9% and was not statistically different from that seen after a benign diagnosis .
The number of cores involved in HGPIN is the one pathological factor that predicts a higher risk of subsequent carcinoma on re-biopsy. HGPIN on >3 cores is associated with a sufficiently high risk of subsequent cancer to warrant re-biopsy within a year of the initial PIN diagnosis. For cases with one or two cores of HGPIN on needle biopsy it is recommended that men do not have a routine repeat needle biopsy within the first year after the diagnosis of HGPIN, in the absence of other clinical indicators of cancer . In cases with HGPIN and adjacent small atypical glands, where the differential diagnosis of the small glands is adjacent cancer or outpouchings off the HGPIN (PINATYP), the risk of cancer is equivalent to that after a diagnosis of ‘atypical glands suspicious for carcinoma’; all these men need re-biopsy within 3–6 months of their PINATYP diagnosis [14–16].
The morphology of HGPIN (flat vs tufting vs micropapillary vs cribriform) does not determine which HGPIN lesions are at greater risk of being associated with carcinoma on repeat biopsy.
Recently, some studies have suggested that molecular findings associated with HGPIN might be able to predict which men are more likely to have cancer on re-biopsy [5,11,17]. In one study using radical prostatectomy specimens, HGPIN lesions adjacent to carcinoma had more AMACR overexpression (56%) than HGPIN lesions away from carcinoma (14%). In a study using needle biopsy cores, patients with at least one AMACR-positive HGPIN gland were 5.2 times more likely to have a subsequent diagnosis of PCa on repeat biopsy than those without any AMACR positive HGPIN glands. Other markers detected by immunohistochemistry, such as PTOV1, may be useful in the future to determine which PIN patients are more likely be diagnosed with PCa on re-biopsy, but all these markers are currently experimental and are not in routine use . Fusion-positive HGPIN is associated with concurrent TMPRSS2-ERG PCa. The detection of isolated TMPRSS2-ERG fusion HGPIN would improve the positive predictive value of finding TMPRSS2-ERG fusion PCa in subsequent biopsies .
Clinical parameters (various serum PSA measurements, DRE, imaging studies) do not identify which men with isolated HGPIN on needle biopsy are more likely to have PCa on re-biopsy .
Long-term risk of PCa
There is scant data on the long-term risk of PCa after a diagnosis of HGPIN. Lefkowitz et al.  reported in a small series that, after an initial diagnosis of HGPIN on 12-core biopsy, the rate of PCa on repeat 12-core biopsy at 3 years was higher than if the re-biopsy was performed within 1 year of the HGPIN diagnosis. They hypothesized that the 3-year interval either allowed unsampled small cancers that were associated with the HGPIN at the time of initial biopsy to grow to a size where repeat biopsy could detect them, or alternatively some of the HGPIN lesions progressed to PCa over this 3-year interval. In an unpublished update from this group, they performed a 3-year delayed-interval biopsy in 101 men with isolated HGPIN and a 6-year delayed-interval biopsy, which yielded PCa detection rates of 23.2% and 28.6%, respectively. Izawa et al.  studied 21 men with isolated HGPIN with a mean follow-up of 72.2 months. A total of seven men were diagnosed with cancer. Although these few studies lack a control population to assess the risk in men without HGPIN, because of the potential medico-legal consequences of not following up on a HGPIN diagnosis, we believe a reasonable approach would be to perform repeat biopsy 2–3 years after a HGPIN diagnosis on needle biopsy until more data is gathered. Re-biopsy should be performed in the region of the original HGPIN site and in adjacent sites, although the entire prostate should be sampled.
High grade PIN is characterized by cellular proliferations within pre-existing ducts and acini, with nuclear and nucleolar enlargement similar to that found in prostate cancer. There is epidemiological, morphological and molecular evidence that HGPIN is a precursor lesion to some carcinomas of the prostate. The clinical importance of recognizing HGPIN is based on its association with PCa. In recent years, a significant decline from 36% to 22% in the predictive value of PCa after an initial diagnosis of HGPIN has been observed. A major factor contributing to the decreased incidence of PCa after a diagnosis of HGPIN on needle biopsy in the contemporary era is related to increased needle biopsy core sampling, which detects many associated cancers on initial biopsy.