PERSPECTIVE. Abstract. and Glen Kristiansen. Yuri Tolkach*

Transcription

1 Journal of Pathology J Pathol 2018; 244: Published online 7 March 2018 in Wiley Online Library (wileyonlinelibrary.com) DOI: /path.5045 PERSPECTIVE Is high-grade prostatic intraepithelial neoplasia (HGPIN) a reliable precursor for prostate carcinoma? Implications for clonal evolution and early detection strategies Yuri Tolkach* and Glen Kristiansen Institute of Pathology, University Hospital Bonn, Bonn, Germany *Correspondence to: Yuri Tolkach, University Hospital Bonn, Institute of Pathology, Sigmund-Freud-Strasse 25, Bonn, Germany. yuri.tolkach@ukbonn.de Abstract High-grade prostatic intraepithelial neoplasia (HGPIN) is a documented putative precursor lesion for invasive prostate adenocarcinoma. However, the precise mechanisms of the carcinoma s development from HGPIN are unclear. Many studies have attempted a comparative molecular genetic characterisation of HGPIN and its corresponding carcinoma to study this transformation. However, to date, some HGPIN mimickers, such as intraductal carcinoma, which can engage in retrograde colonisation of the prostatic acini in an HGPIN-like manner, have been described. In this work, we hypothesise that the lesion formerly known as HGPIN adjacent to invasive carcinoma does not necessarily represent its respective precursor lesion. This hypothesis stems from recent morphological, experimental, and theoretical evidence on the development of tumour clonality, as well as recent studies outlining the three-dimensional architecture of prostate adenocarcinomas (most importantly, their interconnection with the tumoural glandular system). Copyright 2018 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd. Keywords: prostate cancer; clonality; HGPIN; evolution; progression Received 12 September 2017; Revised 5 January 2018; Accepted 20 January 2018 No conflicts of interest were declared. Cancer is a genetic disease, and its emergence and progression are evolutionally driven by alterations to the genetic material of the cell. A putative precursor lesion of prostate cancer (PCA) is high-grade prostatic intraepithelial neoplasia (HGPIN), a proliferation of the atypical secretory epithelium confined to the prostate s pre-existing glands or ducts. This assumption is based on (a) epidemiological data linking HGPIN to the tumour glands and the later occurrence of invasive carcinoma during surveillance, and (b) the morphological similarities between the epithelia of HGPIN and invasive cancer, as well as the concurrent zonal distribution and co-localisation of HGPIN with invasive PCA and their mutually shared genetic rearrangements and other alterations [1]. Some other lesions (e.g. proliferative inflammatory atrophy, PIA) have been proposed as an even earlier putative precursor lesion of PCA [2 4]. However, the evidence is controversial, with atrophy and PIA even associated with a lower risk of PCA [5 8]. The lack of epidemiological studies and the vague morphological definition of PIA, with inter-observer variability reaching 51%, preclude any further conclusions [9]. Some evidence, however, has suggested that PIA is a putative precursor lesion of HGPIN [2 4], which does not change the essence of the considerations outlined below. Historically, there have been some prominent examples of the concept of evolutional emergence of an invasive carcinoma from precursor lesions, with the adenoma carcinoma sequence of colon cancer probably being the most notorious. The molecular genetic characterisation of a precursor lesion and invasive carcinoma is crucial. To understand the processes underlying the transition between these two will have enormous clinical consequences. The issues of cancer heritability, prevention, screening, and early diagnosis could be addressed in a new way with such information. Multiple studies have tried to establish a causal connection between HGPIN as the precursor lesion and PCA as its consequence. One of the last solid examples is the study of Jung et al [10], where the authors studied somatic mutations and copy-number rearrangements in corresponding samples from invasive carcinoma and adjacent corresponding HGPIN, showing the extremely different extents of shared genetic changes between them. Apart from the technical problems involved in sampling HGPIN (which is only possible using sophisticated microdissection), some other problems should be considered, as addressed by Haffner et al [11,12], namely retrograde colonisation of the acini through invasive carcinoma mimicking the HGPIN lesion; in many cases, this is distinguishable only at the molecular level

2 390 Y Tolkach and G Kristiansen (e.g. via the ERG and PTEN status of the invasive tumour and the intraductal component). The classification of intraductal lesions has been broadened recently by the new entity of intraductal carcinoma, which usually represents the intraductal spread of high-grade invasive carcinoma. Unequivocal criteria to discriminate an intraductal precursor proliferation from intraductally located carcinomas that are stroma-invasive elsewhere have not been established. What was earlier considered as HGPIN could easily have been misclassified, actually representing the retrograde colonisation of the acini by invasive carcinoma (intraductal carcinoma; Figure 1). The striking morphological similarity between the two lesions warrants the reassessment of the HGPIN definition. In addition, the hypothesis of intraductally spreading invasive low-grade carcinoma that may be indistinguishable from HGPIN is gradually finding support [13,14]. All this prevents any lasting conclusions about the evolutionary underpinnings of HGPIN on its way to becoming invasive carcinoma. Despite the issues described above, one can acquire some helpful insights into this field by considering how the carcinoma may arise in view of modern evidence on the clonal development of the tumours. Here, we discuss how the emergence of invasive carcinoma could eventually lead to the extinction of its own precursor HGPIN lesion, making sampling of putative HGPIN and adjacent carcinoma highly doubtful analytes for molecular studies seeking to establish clonal and precursor progeny relationships. The starting point of our considerations is the architecture of normal prostatic tissue and prostatic acinar adenocarcinoma. Recently, a confocal microscopy-based study [15] showed that PCA is a highly interconnected hydra -like system of the tubular/acinar structures. We confirmed this suggestion in our investigation [16] with three-dimensional (3D) reconstructions based on a series of consecutive immunohistochemical sections to provide a higher resolution and greater detail in this context (Figure 2). The architecture of PCA and continuity of the inner epithelial lining have allowed the extrapolation of modern experimental and theoretical findings on the development of clonality in tumours [17 20]. In this respect, PCA is comparable to colon adenomas. Briefly, the process of clonal development can be described as follows: During proliferation, some malignant cells acquire driver molecular alterations that promote hallmarks of cancer characteristics, which make them more viable (changing the balance between cell birth and cell death in favour of cell birth, etc); this enables the new clone to spread in different directions on the surface of the epithelium, conquering and replacing the cells of either less adaptive elderly tumour clones or normal epithelium. In the colon, these clonal waves in colonic adenomas occur at the outer surface of the continual colonic epithelium. We assume that a comparable process of clonal development takes place on the inner surface of the interconnected tubular/acinar system in prostate tumours (supplementary material, Video S1). Although this is difficult to prove, the morphological evidence from the new type of intraductal adenocarcinoma engaging in retrograde colonisation of the acini and mimicking HGPIN, as well as HGPIN itself, illustrates this process unambiguously (Figure 1: both spread on the surface of the epithelium). In addition, the common morphological appearance of HGPIN lesions, with lobular, field-like involvement of many connected nearby glands, is indicative of this pattern of spread. What happens when we apply this hypothesis directly to the time point of prostate cancer emergence from HGPIN? In simple terms, the emergence of invasive PCA occurs when tumour cells access the stroma. Here, we see two main possibilities. The first, in our view, seems most realistic. Here, the HGPIN is an incubator of tumour cells, which stay in situ until they have acquired all necessary genetic rearrangements leading to the ability to invade the stroma (Figure 3A). This implies the possibility of breaking through the natural barriers of the glandular system (basal cells or basal membrane) and the creation of the first tumoural outpouching. The further development of the tumour at this site, however, would take place under one important premise, namely a connection to the normal glandular system of the prostate through the HGPIN lesion (Figure 3B). According to the general principles of clonal evolution mentioned above, the development of the tumour would involve the emergence of new clones, and the colonisation of the precursor HGPIN lesion would take place with each clonal wave in the newly formed tumour (Figure 3C, D). This implies that the precursor HGPIN lesion and its cells will in fact be part of the tumour, genetically equal to the dominating tumour clone in this region. The HGPIN lesion and the tumour build a unified evolutionary system, where the cells of the initial precursor HGPIN lesion would eventually be digested by the emerging tumour clones, practically leading to the formation of low-grade intraductal carcinoma an entity that, to date, has been completely underestimated and possibly mis-studied as HGPIN. All this renders attempts to separate and disentangle the precursor HGPIN from invasive tumour for scientific studies (e.g. using the best available technique of laser microdissection of tissue) challenging, if not impossible. We leave space for a second possibility of how the tumour could arise from the HGPIN, which we call the ejection mechanism. We hypothesise that the destruction of the continuity of the prostatic glandular system by, for example, strong, active inflammation may enable the HGPIN epithelium to gain stromal access, thereby becoming invasive. This would lead to a situation similar to that described above, where the invasive tumour cells would be less aggressive and genetically identical to their (birth-) HGPIN gland, while making their way

3 HGPIN: a precursor of prostate cancer? 391 Figure 1. Microscopic images demonstrating the clonal development of different tumoural lesions in the prostate. Tumour clones proliferate and spread along the surface of the inner epithelial lining of normal or tumoural glands. Here, ERG-negative invasive carcinoma retrogradely colonises the big acinus and mimics a micropapillary HGPIN co-existing with an independent ERG-positive HGPIN lesion. (A) Haematoxylin eosin staining. Red arrowheads show the border between the tumour epithelium and normal epithelium, with the invasive colonising tumour clone spreading on the luminal surface of the acinus. (B) ERG staining. The same type of superficial spread of the ERG-positive HGPIN clone with extension into the nearby normal prostatic acini and extinction of the normal epithelium could be encountered, confirming this unified mechanism of clonal tumour evolution. All lesions here (HGPIN, invasive carcinoma, intraductal carcinoma colonising acini) are PTEN-negative (immunohistochemistry, not shown). into the stroma. Put simply, if the door is already open, it is not necessary to break it to escape. In an extreme situation of inflammation, we could have small viable fragments of the HGPIN lying in the stroma and no longer connected to the main HGPIN lesion (supplementary material, Figure S1). This could be a situation where the HGPIN precursor (main lesion) and tumour (developing from the free fragments of HGPIN in the stroma) are developing independently, and the HGPIN precursor is saved from tumoural colonisation. However, our everyday morphological experience in patients with strong inflammation, with and without PCA, does not broadly support this or provide any morphological correlates for such a mechanism. In addition, inflammation associated with HGPIN is only rarely seen in prostate specimens. One more argument for our hypothesis is the contemporary state of the evidence for biomarkers used to detect PCA through urine analysis (PCA3, TMPRSS2 ERG). In a large-scale study [21] including 1225 patients, 518 (42%) were diagnosed with PCA. Of these, 294 patients (57%) had a Gleason score of = 6 and significantly increased TMPRSS2 ERG and PCA3 scores compared with non-cancer patients, which implies that both substances were in the urine. How would these analyses lead to significant results if the tumours were not connected to the normal acinar system of the prostate? The connection of the tumour to the normal prostatic acini will only enable the detection of the substances produced by tumour cells in the urine. This would be possible in two situations: (1) with a connection through the precursor HGPIN lesion (Figure 3); and (2) with a connection where the tumour has actively invaded the pre-existing glandular system of the prostate (conventional intraductal carcinoma). However, it should be considered that conventional intraductal carcinoma is almost never seen in patients with low-grade cancer (Gleason score: = 6), supporting a connection through the former precursor HGPIN lesion. The spatial relationships between prostate cancer and HGPIN may be as follows: (1) HGPIN within a tumour; (2) at the tumour border; and (3) without any visible

4 392 Y Tolkach and G Kristiansen Figure 2. 3D reconstruction of the prostate adenocarcinoma architecture based on the alignment of consecutive sections of Gleason score = 6 prostate adenocarcinoma (pan-cytokeratin AE1AE3 staining was used for all sections). (A, B) 3D reconstruction of the prostate cancer architecture in two different regions with the evidence of interconnections between seemingly independent (in 2D sections) acini of the adenocarcinoma. Corresponding 2D sections with immunohistochemical staining are shown. Figure 3. Putative mechanism of prostate cancer genesis from the HGPIN. (A) The HGPIN serves as an incubator for the first tumour clone with invasive features, letting it gain access to the stroma. (B) First outpouching initiated by the first invasive tumour clone. (C) Small invasive tumour arising from the HGPIN. Note the connection to the HGPIN and continuity of the luminal epithelial lining of these two lesions. A new, more aggressive clone can be observed arising in the invasive tumour, which would initiate the wave of clonal expansion on the inner surface of the tumour glands, which are all interconnected. (D) Clonal wave reaching the HGPIN lesion with replacement ( digestion ) of HGPIN precursor cells through the emerged tumour clone and formation of low-grade intraductal carcinoma (IDC). The direction of the spread of the new tumour clone is shown with arrows. HGPIN and invasive cancer form a unified evolutionary system, which hampers conservation of the precursor HGPIN lesion. connection to the tumour. PCA extends through long distances in the prostate. At some point, it could be found to embrace the HGPIN lesions (type 1 of the spatial relations) or be situated at their border (type 2). However, according to our arguments, the presence of an HGPIN lesion within a PCA is insufficient for calling the HGPIN a precursor lesion, while a real precursor HGPIN lesion would be digested by the tumour in many cancers in the course of its natural clonal evolution, leading to the formation of low-grade intraductal carcinoma, a still underestimated and understudied lesion. What can be done to prove the hypothesis of the concurrent genetic evolution of an HGPIN lesion and its invasive offspring? One possible approach is a detailed comparative molecular genetic characterisation of HGPIN lesions that are topographically unrelated to invasive cancer from prostates with and without (e.g. prostatectomy in patients with bladder cancer) invasive tumour versus those residing in the prostate adenocarcinoma (HGPIN or low-grade intraductal carcinoma). Comparative analysis would allow the identification of genetic changes that occur before and after the transition

5 HGPIN: a precursor of prostate cancer? 393 to the invasive carcinoma. These studies would warrant precise microdissection and modern approaches to molecular characterisation targeting the common genetic alterations. Single cell studies could also be appropriate in such an interrogation. Evidence of multiple clonal populations in the HGPIN lesion with an evolution vector from the side of the topographically related invasive carcinoma would confirm our hypothesis. This would also support the hypothesis of evolution through clonal expansions unfolding on the surface of the epithelium. To the best of our knowledge, the molecular heterogeneity of HGPIN lesions at a single cell resolution has not yet been a subject of investigation. One further possibility is conducting experimental studies on in vivo and ex vivo induction and lineage tracing of clonal evolution [17]. Further, mathematical modelling of the clonality development could be hypothesis-generating and certainly support, navigate, and complement future molecular-genetic studies. Until such approaches have been used, the relationships between HGPIN and PCA will remain uncertain. As a consequence, the identification of early events for screening and diagnosis, and prevention/treatment strategies based on early changes will require further investigations of the evolutionary profiles of PCA. Acknowledgments No funding was received for this study. Author contributions statement Both authors contributed substantially to this work. References 1. De Marzo AM, Haffner MC, Lotan TL, et al. Premalignancy in prostate cancer: rethinking what we know. Cancer Prev Res 2016; 9: Celma A, Servián P, Planas J, et al. Clinical significance of proliferative inflammatory atrophy in prostate biopsy. Actas Urol Esp 2014; 38: Glover M, Soni S, Ren Q, et al. Influence of chronic inflammation on Bcl-2 and PCNA expression in prostate needle biopsy specimens. Oncol Lett 2017; 14: Iczkowski KA, Torkko KC, Wilson RS, et al. Prostatic atrophy: its spatial proximity to carcinoma and intraepithelial neoplasia based on annotation of digital slides. Hum Pathol 2014; 45: Moreira DM, de O Freitas DM, Nickel JC, et al. The combination of histological prostate atrophy and inflammation is associated with lower risk of prostate cancer in biopsy specimens. Prostate Cancer Prostatic Dis 2017; 20: Freitas DM, Andriole GL, Castro-Santamaria R, et al. Extent of baseline prostate atrophy is associated with lower incidence of lowand high-grade prostate cancer on repeat biopsy. Urology 2017;103: Servian P, Celma A, Planas J, et al. Clinical significance of proliferative inflammatory atrophy in negative prostatic biopsies. Prostate 2016; 76: Servian P, Celma A, Planas J, et al. Clinical significance of proliferative inflammatory atrophy finding in prostatic biopsies. Prostate 2015; 75: Giunchi F, Jordahl K, Bollito E, et al. Interpathologist concordance in the histological diagnosis of focal prostatic atrophy lesions, acute and chronic prostatitis, PIN, and prostate cancer. Virchows Arch 2017; 470: Jung S-H, Shin S, Kim MS, et al. Genetic progression of high grade prostatic intraepithelial neoplasia to prostate cancer. Eur Urol 2016; 69: Haffner MC, Weier C, Xu MM, et al. Molecular evidence that invasive adenocarcinoma can mimic prostatic intraepithelial neoplasia (PIN) and intraductal carcinoma through retrograde glandular colonization. J Pathol 2016; 238: Haffner MC, Barbieri CE. Shifting paradigms for high-grade prostatic intraepithelial neoplasia. Eur Urol 2016; 69: Varma M. Low-grade intraductal carcinoma of the prostate: an idea whose time has not yet come. Histopathology 2017; 71: Shah RB, Yoon J, Liu G, et al. Atypical intraductal proliferation and intraductal carcinoma of the prostate on core needle biopsy: a comparative clinicopathological and molecular study with a proposal to expand the morphological spectrum of intraductal carcinoma. Histopathology 2017; 71: van Royen ME, Verhoef EI, Kweldam CF, et al. Three-dimensional microscopic analysis of clinical prostate specimens. Histopathology 2016; 69: Tolkach Y, Thomann S, Kristiansen G. Three-dimensional reconstruction of prostate cancer architecture with serial immunohistochemical sections: hallmarks of tumour growth, tumour compartmentalization and implications for grading and heterogeneity. Histopathology 2018; DOI: [Epub ahead of print]. 17. Alcolea MP, Greulich P, Wabik A, et al. Differentiation imbalance in single oesophageal progenitor cells causes clonal immortalization and field change. Nat Cell Biol 2014; 16: Baker A-M, Graham TA, Wright NA. Pre-tumour clones, periodic selection and clonal interference in the origin and progression of gastrointestinal cancer: potential for biomarker development. J Pathol 2013; 229: Bozic I, Gerold JM, Nowak MA. Quantifying clonal and subclonal passenger mutations in cancer evolution. PLoS Comput Biol 2016; 12: e Vogelstein B, Papadopoulos N, Velculescu VE, et al. Cancer genome landscapes. Science 2013; 339: Tomlins SA, Day JR, Lonigro RJ, et al. Urine TMPRSS2:ERG plus PCA3 for individualized prostate cancer risk assessment. Eur Urol 2016; 70: SUPPLEMENTARY MATERIAL ONLINE Supplementary figure legend Figure S1. Putative mechanism of prostate cancer genesis from the HGPIN with fragmentation of the main HGPIN lesion due, for example, to massive inflammation (prostatitis) Video S1. Schematic representation of clonality development in prostate adenocarcinoma