Abstract
High-grade prostatic intraepithelial neoplasia (PIN) has been accepted as the main precursor lesion to invasive adenocarcinoma of the prostate, and this is likely to be the case. However, in an unknown number of cases, lesions fulfilling the diagnostic criteria for high-grade PIN may actually represent intra-acinar or intraductal spread of invasive carcinoma. Intriguingly, this possibility would not contradict many of the findings of previous epidemiologic studies linking high-grade PIN to carcinoma or molecular pathologic studies showing similar genomic (e.g., TMPRSS2-ERG gene fusion) as well as epigenomic and molecular phenotypic alterations between high-grade PIN and carcinoma. Also, this possibility would be consistent with previous anatomic studies in prostate specimens linking high-grade PIN and carcinoma in autopsy and other whole prostate specimens. In addition, if some cases meeting morphologic criteria for PIN actually represent intra-acinar spread of invasive carcinoma, this could be an important potential confounder of the interpretation of past clinical trials enrolling patients presumed to be without carcinoma, who are at high risk of invasive carcinoma. Thus, in order to reduce possible bias in future study/trial designs, novel molecular pathology approaches are needed to decipher when an apparent PIN lesion may be intra-acinar/intra-ductal spread of an invasive cancer and when it truly represents a precursor state. Similar approaches are needed for lesions known as intraductal carcinoma to facilitate better classification of them as true intra-ductal/acinar spread on one hand or as precursor high-grade PIN (cribriform type) on the other hand; a number of such molecular approaches (e.g., coevaluating TMPRSS-ERG fusion and PTEN loss) are already showing excellent promise. Cancer Prev Res; 9(8); 648–56. ©2016 AACR.
Introduction
Prostate cancer is the most common noncutaneous cancer and the second leading cause of cancer-related death in men in the United States. While the death rate has been decreasing in some countries, including the United States, the aging population and the increasing “westernization” in a number of parts of the world is projected to produce an increasing burden of cases worldwide for the foreseeable future. Early detection, through serum prostate-specific antigen (PSA) screening coupled with improved treatment of localized disease, is likely responsible for much of the decrease in death rate (1). Nevertheless, as a result of the potential harmful side effects associated with radical treatment of men who do not necessarily need treatment, PSA screening is controversial (2). Thus, in addition to early detection with serum PSA screening and treatment of established disease, there has been great interest in developing strategies to prevent the disease altogether or to intercept the disease before it can progress to aggressive forms.
Attempts to prevent prostate cancer were conducted in two different large prospective randomized clinical trials (RCT) employing the 5-α reductase inhibitors finasteride (the Prostate Cancer Prevention Trial or PCPT; ref. 3) and dutasteride (Reduction by Dutasteride of Prostate Cancer Events or REDUCE; each vs. placebo; ref. 4). Both trials showed a reduction in the period-prevalence of prostate cancer of approximately 25%. However, in both trials, there was a small increase in the total number of patients diagnosed with higher grade disease; and, with long-term follow-up in the PCPT, there was no apparent survival advantage in patients in the treatment arm (5). Thus, neither of these agents are currently being employed widely for the chemoprevention of prostate cancer. A third large RCT (Selenium and Vitamin E Cancer Prevention Trial or SELECT) consisted of treatment with the antioxidants vitamin E and/or selenium which showed a lack of efficacy (6). A number of chemoprevention trials employed enrichment of men with high risk of cancer development because they were diagnosed with high-grade PIN, in which attempts were made to reduce the development of invasive disease with a variety of agents (reviewed in refs. 7, 8); such trials have not resulted in strong candidates for use as chemopreventative agents in any population to date (Table 1).
Selected clinical trials for prostate cancer prevention
Proposed Alternative Prostate Cancer Precursor Lesions
In addition to the well-accepted lesion of high-grade PIN, a number of different histologic lesions have been proposed as potential precursor lesions for prostate cancer, such as adenosis (atypical adenomatous hyperplasia) and proliferative inflammatory atrophy (PIA). Adenosis may be a potential precursor to carcinomas arising in the transition zone, lesions which are often low grade and not generally thought to possess strong malignant potential (9). PIA, consisting of simple atrophy and postatrophic hyperplasia, that is often associated with inflammation, has been found to merge directly with small adenocarcinoma lesions in the peripheral zone, but this appears to be relatively rare (although very few systematic studies have been carried out to date; refs. 10–12). If PIA is a precursor (or “risk factor lesion”) for prostate cancer, it may do so indirectly by leading to carcinoma via PIN as PIA merging with PIN is quite common (10–12).
Diagnosis and Subtypes of PIN
High-grade PIN is characterized by atypical secretory luminal cells present within preexisting ducts and acini, with nuclear features similar to invasive adenocarcinoma (Fig. 1) (7, 9, 13). PIN has been proposed to arise with progressive morphologic abnormalities along a continuum from normal epithelium, to low-grade PIN to high-grade PIN and then to invasive carcinoma (7). However, as low-grade PIN has been deemed to be a diagnostic entity with poor interobserver reproducibility, and its clinical utility has not been demonstrated (7, 9), many studies of prostate cancer pathogenesis as well as randomized trials have not attempted to measure it.
Models of prostate cancer development based on latest evidence. A, model of progression from normal to low-grade PIN (LGPIN) to high-grade PIN (HGPIN) to carcinoma. Thicker solid lines denote standard/canonical view. Note that there are only very limited data showing normal epithelium goes to low-grade PIN and that high-grade PIN arises from low-grade PIN. A subset of low- or high-grade PIN or invasive carcinoma may also arise from PIA, with the most likely route going from PIA to low- or high-grade PIN more commonly than going directly to carcinoma. Intraductal carcinoma is generally considered to arise after invasion, but has now been found in some cases to be present in prostates without invasive carcinoma and thus may arise directly from high-grade PIN at times or as a de novo lesion, which does occur in some mouse models of prostate cancer. The canonical view has been that lesions morphologically identifiable as high-grade PIN arise prior to invasive carcinoma development, but new evidence suggests that, at times, lesions morphologically identifiable as high-grade PIN may also arise as retrograde spread from invasive carcinoma. B, H&E images of morphologic entities indicated in A. B, normal (A), focal atrophy with inflammation/PIA (B), low-grade PIN (C), high-grade PIN (D), invasive adenocarcinoma (Gleason pattern 4 + 3 = 7; E), intraductal carcinoma and adjacent invasive adenocarcinoma (F). Lu, lumens of glands.
High-grade PIN shows a variety of architectural patterns, with the “tufted” pattern being the most common, occurring in >95% of cases (7). Other patterns include micropapillary and less commonly cribriform and flat. The diagnosis of high-grade PIN can be difficult at times, which is related to the fact that there are a number of both benign and malignant mimickers as well as borderline lesions (9). This has led to the fact that the diagnosis of high-grade PIN shows only moderate to good interobserver reproducibility (κ statistic of ∼0.41 through 0.6) among experts in the field (9).
Key Data Supporting High-Grade PIN as a Prostate Cancer Precursor
There is abundant data supporting the concept that high-grade PIN is a precursor to adenocarcinoma of the prostate (7, 9, 13, 14). For example, high-grade PIN cells resemble invasive adenocarcinoma cells morphologically, including having the key feature of nucleolar enlargement. Further evidence comes from zonal colocalization between high-grade PIN and carcinoma, the frequent multifocal occurrence of high-grade PIN and carcinoma, morphologic transitions between high-grade PIN and invasive “microcarcinomas,” morphometric measurements, phenotypic features (e.g., abnormal luminal cells expressing AR etc.), and a number of shared somatic genomic changes between high-grade PIN and carcinoma.
Perhaps the most cited finding for implicating high-grade PIN as the principal prostate cancer precursor comes from studies showing that fully embedded prostates containing carcinoma show a significantly increased prevalence (∼2 fold) and extent of high-grade PIN when compared with prostates without carcinoma (from autopsies, prostatectomies, or cystoprostatectomies; refs. 7, 9). While African-Americans tend to show increased rates of clinically diagnosed carcinoma, as well as increased death rates due to prostate cancer, they do not harbor more invasive carcinomas or high-grade carcinoma in autopsy studies when compared with age-matched men of non-African ancestry. However, they do tend to show an increase in the fraction of cases showing extensive high-grade PIN lesions (7, 9, 15).
A number of autopsy studies from different parts of the world indicate clearly that, similar to invasive carcinoma, the rates of high-grade PIN increase significantly with age. For example, a prevalence of 7% to 8% has been seen in the third decade of life and upwards of 60% to 86% in the eighth decade of life (9). Furthermore, the rates of high-grade PIN, like invasive carcinoma, appear to vary significantly throughout the world. For example, a relatively recent study of Caucasian Mediterranean men showed that the prevalence of high-grade PIN and carcinoma was lower than those from age-matched Caucasian North American and African-American men (16).
If high-grade PIN is a precursor lesion, one would expect that in autopsy studies PIN lesions would predate carcinoma. While this is the cases in some studies (e.g., ref. 16) it is not so in all, including a recent study from Hungary (17) and the most well-known studies conducted in North American men, where the increase in high-grade PIN with age was either delayed somewhat in relation to the increase in carcinoma or occurred at nearly the same time as invasive carcinoma (refs. 15, 18, 19; Fig. 2). In two highly cited studies (18, 19), only a single time period (in the third decade) was found in which the prevalence of PIN was higher than carcinoma, and, this predating of carcinoma was nearly exclusively due to low-grade PIN; high-grade PIN was first identified in the fifth decade of life (19).
Prevalence of PIN and carcinoma in autopsy studies. A and B, prevalence of high-grade PIN and carcinoma across age groups in men from different ancestry (A, Whites; B, African American). Graphs were plotted using frequencies obtained as reported by Sakr and colleagues in the largest study to date of this kind (n = 525 autopsied med; ref. 15). C and D, similar plots obtained from frequencies reported by Sakr and colleagues (ref. 18 = 249 autopsied men).
In prostate needle biopsies from various populations, the prevalence of isolated high-grade PIN (no carcinoma present) ranges between 4% and 24% (mean of 9%; ref. 7). The rate of high-grade PIN in the placebo arm of the PCPT trial (men who never had an elevated PSA or abnormal digital rectal exam) was 7% (20).
The clinical significance of isolated high-grade PIN on prostate biopsy stems from its association with cancer on subsequent biopsies. However, this association has been decreasing over the years from approximately 30%–50% to approximately 20%. This decrease has been largely attributed to increased sampling from 6 to 12 or more cores which results in fewer cancers missed on the initial biopsy (7,21). In any case, focal high-grade PIN on needle biopsy no longer dictates the need for immediate repeat biopsy and most authors suggest a repeat biopsy at 1 year if there are no other clinical indications. The presence of multifocal high-grade PIN on prostate needle biopsy (involving 2 or more cores) still appears to represent a risk factor for detection of cancer on subsequent biopsies (9).
Molecular Biology of PIN
A large number of molecular alterations have been identified in high-grade PIN lesions, many of which are also found in adenocarcinoma (reviewed in ref. 7, 22; Table 2 shows some well-known key changes in various lesions). Early examples include a study from Emmert-Buck and colleagues, showing that microdissected high-grade PIN had a similar rate and pattern of loss of one allele on chromosome 8p12–21 (90% in carcinoma and 63% in high-grade PIN) as compared with invasive adenocarcinoma; and that a region on 8p22 showed less frequent LOH in PIN than in carcinoma, implying that PIN is molecularly intermediate between normal and carcinoma (23). Others reported similar findings (24). Bostwick and colleagues reported that multifocal PIN often harbored similar loss of heterozygosity by alleotyping to that seen in concurrent invasive carcinoma lesions (25). Quin and colleagues used centromere-specific enumeration probes as well as a probe encompassing MYC at 8q24 to show that PIN frequently harbors similar chromosomal alterations as invasive carcinoma (26, 27). Not all studies, however, supported the concept that PIN lesions always behave as precursors as some PIN lesions near carcinoma actually harbor additional or different genetic changes not found in the adjacent carcinoma (23, 28, 29). Furthermore, not all studies have found evidence of high rates of chromosomal changes in PIN or found much lower rates than the earlier studies. For example, Bethel and colleagues found no allelic loss at chromosome 8p22 (LPL locus) in 19 isolated high-grade PIN lesions (30). Similarly, PTEN protein and allelic loss were recently found to be very common in intraductal carcinoma and atypical cribriform lesions, but not in isolated high-grade PIN (31, 32).
Selected molecular alterations in atrophy, PIN, and intraductal carcinoma
ETS gene rearrangements, which are mostly characterized by the TMPRSS2-ERG fusion, can be detected in high-grade PIN in 5%–20% of men of European decent. This contrasts with a much higher frequency in carcinoma (∼50%; ref. 7). Furthermore, PIN lesions near carcinoma are much more likely to be TMPRSS2-ERG fusion positive than isolated high-grade PIN away from carcinoma, in which the frequency is very low (22, 33, 34), leading the authors to propose that TMPRSS2-ERG–positive PIN lesions are associated with clonal progression to carcinoma. In addition, some (albeit not all) studies have shown that ERG-positive high-grade PIN show a higher rate of cancer on repeat biopsy than ERG-negative cases (7, 35).
MYC is located on chromosome 8q24, a locus that frequently shows increased copy number in prostate cancer. Qian and colleagues have shown that 8q24 gain is commonly seen in at least some forms of high-grade PIN (27), although a separate study using a similar approach by Bethel and colleagues did not see 8q24 gain in any of 19 high-grade PIN lesions away from carcinoma (30). Despite the lack of 8q24 gain, Gurel and colleagues found a clear stepwise increase in MYC protein levels from normal epithelium to low-grade and then to high-grade PIN lesions; high-grade PIN was similar to invasive adenocarcinoma (30, 36).
Somatic DNA hypermethylation is a common occurrence in prostate cancer and a number of genes shown to be selectively hypermethylated in prostatic carcinomas, and not in benign tissues, are also frequently methylated in high-grade PIN. For example, the CpG island upstream of the GSTP1 gene is methylated in approximately 90%–95% of carcinomas, and in approximately 70% of high-grade PIN lesions (37, 38). Also Henrique and colleagues found a high prevalence and extent of hypermethylation of GSTPT1, RARB, and APC in high-grade PIN and invasive adenocarcinoma (39).
Telomeres have been shown in multiple studies to be paradoxically shortened in prostate tumor cells, compared with their normal counterparts, despite the fact that most of them have high levels of telomerase activity. In high-grade PIN, telomeres are also short in the vast majority of cases (40, 41). Interestingly, Vukovic and colleagues found that PIN lesions near carcinoma often showed shorter telomeres than PIN lesions away from carcinoma (41).
Phenotypically, a large number of immunohistochemical biomarkers have been shown to be altered in PIN. The most relevant are often similarly altered in carcinoma and these include MYC, as mentioned above, as well as proliferation markers such as Ki67, and differentiation markers such as PSA, and NKX3.1, which are somewhat decreased in PIN and carcinoma as compared with normal luminal cells (7, 9).
Intraductal Carcinoma
Intraductal carcinoma consists of malignant appearing epithelial cells expanding prostatic acini and/or ducts with retention of the basal cell layer. Intraductal carcinoma has been postulated to represent, in most cases, retrograde glandular colonization (intra-acinar/ductal spread), of preexisting invasive carcinoma that is usually high grade (Gleason grade 4 + 3 = 7 or higher). A number of studies have also shown that the presence of intraductal carcinoma is associated with adverse pathology at prostatectomy and worse outcomes after treatment (42). There are two main patterns of intraductal carcinoma: (i) solid or dense cribriform or (ii) lose cribriform or micropapillary. Guo and Epstein (2006) introduced strict morphologic criteria for the diagnosis of intraductal carcinoma on needle biopsy such that the pattern had to be either solid or dense cribriform or loose cribriform or micropapillary with either marked nuclear atypia (nuclear size 6 times normal or larger) or nonfocal necrosis (43). Isolated intraductal carcinoma is very rare on needle biopsy, but if found there is a general consensus that it should be treated similarly to high-grade invasive carcinoma (42, 44).
In addition to lesions that fit the Guo and Epstein criteria, pathologists have reported on the presence of atypical cribriform lesions with features intermediate between high-grade PIN and intraductal carcinoma. In a study that attempted to distinguish high-grade cribriform PIN from intraductal carcinoma, Han and colleagues observed that isolated atypical cribriform lesions were uncommon such that the overwhelming majority of them were associated with high-grade (Gleason score ≥ 7) and high-volume prostate cancer supporting the concept that they represent a spectrum within intraductal carcinoma (45).
Molecularly, intraductal carcinoma shows frequent loss of heterozygosity, indicative of genomic instability (46). Han and colleagues used FISH for ERG rearrangements and categorized the lesions into 2 groups, group A (meeting the Guo and Epstein criteria) and group B (atypical cribriform lesions nearby invasive high-grade carcinoma but that did not meet the Guo and Epstein criteria); they also studied cribriform high-grade PIN lesions that were isolated away from carcinoma (45). ERG rearrangements were not detected in isolated cribriform high-grade PIN but were present in 47% of group A and 48% of group B lesions (45). PTEN loss (usually by large-scale deletion and rarely by mutation) is common in prostate cancer and is enriched in high-grade aggressive disease. Lotan and colleagues found common loss of PTEN in both intraductal carcinoma (∼80%) as well as in atypical cribriform lesions that fall short of criteria for intraductal carcinoma (analogous to Han and colleagues' group B cases as described above; refs. 31, 32). In these studies, a total of 59 isolated high-grade PIN lesions were evaluated and none showed PTEN loss. Taken together, the data indicate that intraductal carcinoma and atypical cribriform lesions adjacent to carcinoma are quite distinct from isolated high-grade PIN, even cribriform high-grade PIN that is away from carcinoma.
While pathologists have largely accepted that intraductal carcinoma is usually the result of retrograde spread of carcinoma colonizing benign ducts and acini, until very recently, there was no unequivocal genetic evidence to support this. Recently, we used a novel molecular pathology approach to show this is indeed true, at least in some cases. The method involved the use of capture sequencing to determine case-specific TMPRSS2-ERG genomic breakpoints (geBACS) coupled with ERG IHC to identify the precise genomic breakpoint in ERG-rearranged cases to establish clonality in a given adenocarcinoma lesion (47). Then, we used the PTEN genomic status determined by IHC and FISH to show that only part of the invasive tumor lost PTEN, yet all of the intraductal tumor lost PTEN and had the identical TMPRSS2-ERG breakpoint, which established that the intraductal component arouse temporally after the invasive component (48). In another recent study, Lindberg and colleagues, recently published on a single case using whole-genome sequencing to show that an intraductal carcinoma was phylogenetically closer to lymph node metastases than were most areas of an adjacent carcinoma, also strongly indicating that the intraductal carcinoma likely arose after the invasive component (49). On the other hand, it has also become apparent recently that occasionally, lesions that are histologically diagnosed as intraductal carcinoma can be present in prostates without any invasive carcinoma (44), or, can be present away from invasive carcinoma (50). It appears, therefore, that histologic intraductal carcinoma may at times be a precursor lesion or occur in the absence of invasive carcinoma.
What about high-grade PIN? Interestingly, in the study by Haffner and colleagues (48), lesions meeting morphologic criteria for high-grade PIN (both cribriform and tufting) that were adjacent to carcinoma also showed evidence that they arose temporally after invasion commenced, suggesting that not all lesions regarded as PIN represent precursors, and at times, may represent invasive carcinoma colonizing benign glands in a retrograde manner (Fig. 1). Another frequent finding was that in cases of ERG-positive carcinoma, adjacent benign glands that were only partially involved with ERG-positive carcinoma cells were present (Fig. 3). This observation provides further morphologic evidence of common intraepithelial spread of carcinoma, at least near invasive lesions.
ERG-positive carcinoma cells present in benign prostatic acini/ducts. An ERG-positive carcinoma (brown) is shown invading the prostatic stroma. Note 4 benign glands (numbers 1–4) with partial involvement with malignant ERG-positive cells.
A Way Forward
What are the implications of these findings? These types of molecular timing or “order of event” studies have been performed on a limited number of cases so far. Nevertheless, many of the same histologic, molecular, and epidemiologic findings used to support high-grade PIN as a bone fide precursor lesion are equally consistent with a scenario in which a significant fraction of lesions that are currently classified histologically as high-grade PIN actually represent invasive adenocarcinoma invading into the otherwise benign duct/acinar system. In addition, in an unknown number of cases, the results of clinical trials where patients are enriched for high risk of developing cancer by having high-grade PIN could be significantly confounded as many of the patients considered to have precursor lesions only would actually have unsampled invasive carcinoma. This further compounds the already well-known confounder of these trials that, due to the sampling problem occurring with blinded needle biopsies, many patients deemed to be negative for cancer (e.g., with entirely negative biopsies) actually have invasive carcinomas.
To address this confounding problem of carcinoma potentially mimicking PIN, additional studies need to be performed using molecular pathology approaches similar to those recently employed (48, 49) to obtain estimates of the fraction of high-grade PIN lesions that represent retrograde colonization by carcinoma. These combinatorial approaches include the integration of H&E histology of well-characterized specimens with IHC (including multiplex assays) and in situ hybridization–based methods (both FISH and RNA in situ), along with solution-based molecular approaches applied after laser capture microdissection, that are required to solve this issue. Such solution-based molecular approaches that employ high-throughput next-generation sequencing can be combined with molecular phylogenetic analyses to comprehensively decipher the temporal “order of events” to better understand precursor–progeny relations. In addition, if many of the cases of high-grade PIN studied with molecular tools to date actually represent invasive carcinoma mimicking PIN, it will be important to move beyond our reluctance for studying low-grade PIN and more systematically employ molecular pathologic tools to determine whether key molecular driver alterations indeed begin in these lesions. Table 2 shows some selected molecular lesions thought to be involved as key events in early prostate cancer formation. While the table is not exhaustive, it is quite clear that very limited data exist regarding molecular studies of low-grade PIN. Given the diagnostic challenges involved with low-grade PIN, we submit that the time is right for genitourinary pathologists to revisit the reproducibility of diagnosing low-grade PIN to facilitate future studies.
Improvements in our ability to understand which PIN lesions represent true precursors, together with improvements in imaging localized disease, could ultimately aid patient management as approximately 50,000–70,000 new cases of isolated high-grade PIN are diagnosed per year in the United States. This combination of improved molecular diagnostics and imaging could also aid in terms of future chemoprevention clinical trials by enhancing the confidence in the classification of patients who do not have carcinoma at the start of the trial and truly following them for the effect of the selected agent on cancer development (this is likely a conservative estimate. This assumes the biopsy rate has been reduced from >1,000,000 per year to ∼800,000 per year after the USPTF recommendations in 2013, and an isolated high-grade PIN rate of 7% = 56,000 cases per year).
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed
- Received December 14, 2015.
- Accepted January 18, 2016.
- ©2016 American Association for Cancer Research.
References
Volume 9, Issue 8
