The Prostate 43:11–19 (2000) Patterns of Differentiation and Proliferation in Intraductal Carcinoma of the Prostate: Significance for Cancer Progression Ronald J. Cohen,1* John E. McNeal,2 and Tina Baillie3 1 Urological Research Centre, University of Western Australia, Perth, Western Australia, Australia 2 Department of Urology, Stanford University Medical Center, Stanford, California 3 Department of Pathology, University of Auckland, Auckland, New Zealand BACKGROUND. Cribriform prostatic intraepithelial neoplasia (C-PIN) identifies a unique histological pattern: dysplastic cells line ductal/acinar walls but also span across gland lumens. C-PIN is distinct from other forms of dysplasia; it is seldom seen except within invasive cancer, it is more frequent in larger/higher-grade cancers; and it appears to contribute independently to aggressive behavior. Hence, C-PIN may represent a separate, more aggressive entity: intraductal carcinoma of the prostate (IDC-P). Here, support for that distinction stems from a histologic/biologic subclassification of IDC-P, whose elements are linked to features of invasive cancer. METHODS. Histologic criteria were tested against 26 radical prostatectomies, using immunostains for prostate-specific antigen, MUC-2, androgen receptor (differentiation), and Ki-67 (proliferation). Invasive cancer grade, stage, and follow-up were compared. RESULTS. Architecture of the central (luminal) cell compartment defined three subclasses of IDC-P: A (trabecular), B (cribriform), and C (solid/comedo), which represented progressive dedifferentiation with a reciprocal increase in proliferation. The IDC-P subpattern correlated with cancer stage, grade, and clinical course. CONCLUSIONS. IDC-P is a separate entity, distinct from PIN; cancers associated with IDC-P are more aggressive than those associated with only PIN. It comprises a spectrum of histological patterns which appear to be determined in concert with invasive cancer, whose prognosis it worsens. Prostate 43:11–19, 2000. © 2000 Wiley-Liss, Inc. KEY WORDS: classification; intraductal prostate cancer; proliferation; differentiation INTRODUCTION Prostatic ducts and acini whose lumens are filled by malignant-appearing glandular epithelial cells represent a well-recognized histologic pattern [1–3], whose precise histologic criteria and biologic significance have been variously defined. In the Gleason grading system , this histologic pattern was classified with invasive cancers and was designated a cribriform subpattern of grade 3 carcinoma. More recently, the identification of basal cells surrounding cribriform nests has stimulated the reclassification  of this lesion as cribriform prostatic intraepithelial neoplasia (cribriform PIN). Though referring to the same histologic © 2000 Wiley-Liss, Inc. pattern, the implications for biologic behavior between these two definitions are vastly different. PIN or duct-acinar dysplasia refers to alterations of the epithelial layer lining preexisting benign lumens by histologic features resembling carcinoma. These include cell proliferation with nuclear multilayering, nuclear enlargement with nonuniformity of size and Grant sponsor: Richard M. Lucas Foundation; Grant sponsor: Research Committee, Sir Charles Gairdner Hospital. *Correspondence to: Dr. R.J. Cohen, Urological Research Centre, Level 2, M Block, QE II Medical Centre, Nedlands, Western Australia 6009, Australia. E-mail: email@example.com Received 5 April 1999; Accepted 3 November 1999 12 Cohen et al. contour, hyperchromasia, and enlarged nucleoli. Dysplasia is probably the main precursor lesion for invasive cancer in the prostate, and hence it is often recognized in prostate glands that have not yet developed invasive carcinoma. Unique among dysplasia variants, cribriform PIN has atypical cells completely spanning the lumen and often filling it . In order to shed light on disparities in histological classification, we recently studied this lesion in a multivariate morphologic analysis of 136 serially blocked radical prostatectomy specimens . We found other histologic patterns of severe dysplasia (PIN) to be extremely common outside the boundaries of invasive cancer areas, while the cribriform pattern (and other lumen-spanning patterns) was rare except inside the boundaries of invasive carcinoma. In addition, duct-spanning lesions were very uncommon in small tumors, but showed a linear increase in frequency parallel to the volume of invasive carcinoma, and they reached a peak of 40% in carcinomas over 4 cm3. It was concluded that the above features were unlikely for a cancer precursor, but they were quite consistent with the emergence of cribriform duct involvement within an already established invasive carcinoma. The frequency of intraductal, duct-spanning lesions also correlated highly with percent of high-grade (grades 4/5) invasive cancer, but multivariate analysis indicated an additional independent role for such luminal cell masses in predicting poor prognosis [6–8]. Further, the close resemblance of these duct-spanning lesions with the histologic and biologic features of the concurrent invasive cancer was inconsistent with the features of dysplasia (PIN); accordingly, they were designated as a separate entity, intraductal carcinoma  of the prostate (IDC-P). Most of the above findings have been confirmed by two subsequent studies [7,8]. However, these authors chose to retain the “cribriform PIN” terminology, despite the declaration in one paper  that this “represents intraductal spread of carcinoma within preexisting ducts and acini and should not be categorized as PIN.” As a common entity and one closely related to invasive cancer progression, intraductal carcinoma of the prostate (IDC-P) warrants further investigation. Accordingly, in this current study we investigated the different histologic patterns of IDC-P and thoroughly describe their features. With immunohistochemical staining for markers of differentiation and proliferation, we have been able to place these along a histologic continuum, which correlates with biologic aggressiveness. MATERIALS AND METHODS Tissues A series of 26 prostatic adenocarcinomas with areas of intraductal carcinoma (IDC-P) was randomly selected from two previous published series [6,9]. The first 19 cases were collected from prostatectomies performed at the Stanford University Medical Center, and each was represented by two selected slides containing the largest IDC-P areas. In all 19 cases, the entire prostate was blocked and each 3-mm transverse slice was divided into quarters for ease of processing. Seven prostatectomies from the Department of Pathology at Auckland University were added and represented by serial whole-mount sections. From both centers, cases had been excluded which showed poor fixation or had a history of hormonal or irradiation therapy, and all tumors were peripheral-zone cancers. All specimens were fixed whole for 18–23 hr in 10% buffered formalin. Constant circulation of formalin around the specimen as well as perfusion through the urethra, using a small pump, assured adequate and prompt fixation. Fixed prostates were serially blocked at 3-mm intervals and transferred to 70% ethanol. There was no further contact with formalin, and all tissues were processed through solvents to paraffin wax. Immunohistochemistry Eight 5-m sections were cut from blocks of all 26 cases and subjected to the following stains: hematoxylin and eosin, anti-prostate-specific antigen (PSA) (ERPR8, 1:100; PASE/4LJ, 1:100, Dako,Glostrup, Denmark), anti-Ki-67 (clone MIB1, 1:100, Immunotech SA, France), anti-high molecular weight cytokeratin (HMWCK) (34␤E12,1:50, Dako), anti-androgen receptor (AR) (NCL-ARp(AR70), 1:200, Novocastra Laboratories, UK), MUC-1 (MUSE 11, undiluted ), and MUC-2 (CCP58, 1:1,000 ). Heat antigen retrieval was used in the immunostaining for HMWCK, Ki-67, and AR. Intraductal-acinar cancer areas were identified by previously published criteria . Distinction from high-grade dysplasia (prostatic intraepithelial neoplasia) required the presence of a “central cell compartment” of epithelium completely spanning the gland lumen. The intraductal-acinar pattern almost always comprised one or two well-demarcated areas within which most contiguous glandular lumens were filled with cancer of relatively uniform histology. The immunohistochemical staining was graded according to the intensity of the stain as well as the Intraductal Carcinoma of the Prostate Fig. 1. Association of basal cells (arrows) with the epithelial stromal boundary of cribriform (pattern B) carcinoma growing in ducts (HMWCK immunostain, ×40). number of positive cells. Intense positivity in almost all tumour cells was designated “+++”; 30–50% of stained tumor cells intensely positive or greater than 50% but with less intense staining was designated “++”; intense staining of more than 5% but less than 30% of cells or less intense staining of more than 10% but less than 50% of cells was designated “+”. Lesser degrees of staining were designated “±”, and complete absence of staining was noted as “−”. McNemar chi-square analysis was used to compare the different staining patterns. For the invasive component of all cancers, the Gleason grade  and Gleason score (sum) were determined. Total cancer volume (invasive and IDC-P) was determined for all 26 prostatectomy cancers [6,9], as previously reported. 13 Fig. 2. Pattern A intraductal carcinoma (IDC-P) with a “trabecular” architecture, showing two cytologic compartments. The perimeter compartment associated with pleomorphic cells containing macronucleoli can be distinguished from the bland central cells with small nuclei (×240). RESULTS with, the periglandular stroma had prominent malignant features which were indistinguishable from those of the glandular lining cells of severe dysplasia. This population was identical in all 36 cases. This area of histologic equivalence to dysplasia was designated the perimeter compartment. We designated as the central compartment all the remaining luminal epithelial cells, which comprised the transluminal mass and which lay beyond the thickness of the typical dysplastic glandular lining. It was this compartment alone which showed great variation in histologic appearance between cases, and these differences were the basis for defining different histologic patterns of IDC-P. In almost all cases, the central compartment was histologically different from the perimeter compartment, and a sharp line of demarcation was usually visible, facilitating their distinction. Histologic Characterization of Intraductal Carcinoma (IDC-P) Histologic Patterns of IDC-P Using previous criteria , 26 peripheral zone carcinomas at 4–10 cm3 volume had foci of contiguous IDC-P ducts and acini which comprised more than 10% of the estimated total cancer volume. In IDC-P foci of all cases, there were frequent areas mimicking normal branching duct-acinar prostatic architecture. Among the 26 cases, 20 (79%) demonstrated a complete basal-cell layer surrounding the intraductal tumor masses (Fig. 1), while in 6 cases the basal-cell layer was less complete, with spaces and gaps between immunostained areas. Detailed cytologic examination of epithelial cells of all glandular IDC-P foci showed that the ring of secretory epithelial cells which was close to, or in contact The histologic features of the central compartment in all cases showed one of three architectural patterns. In pattern A, empty luminal space was the larger component centrally (Fig. 2). The epithelium exhibited a trabecular pattern; narrow strands, often only two cell layers thick, spanned the lumen without stromal support, intersecting randomly to form a lacework of very orderly structure. Sometimes coalescence of strands in the very center of the lumen created a small cell mass. The central compartment nuclei were small and uniform, enhancing the orderly appearance, and their cytoplasm was often pale; thus, demarcation between compartments was clear. In pattern B, the classic cribriform pattern, multiple 14 Cohen et al. Fig. 3. Pattern B intraductal carcinoma (IDC-P). The tumor expands existing ducts and acini, and has a cribriform architectural arrangement (×48). small, round to elongated lumens perforated a central compartment cell mass which typically comprised more than half of the luminal space (Fig. 3). The component microlumens varied considerably between cases in their size and frequency. Central-compartment nuclei were usually somewhat smaller and more bland than perimeter-compartment nuclei, with frequent suggestion of a line of demarcation. A small focus of coagulation necrosis was found at the exact center of an occasional gland. Pattern C was a solid pattern throughout the central compartment (Fig. 4), without clear differentiation between this central area and the perimeter cells. In a subgroup of this pattern, prominent coagulation necrosis (Fig. 5) obliterated the cells of most of or the entire central compartment (comedonecrosis), and both compartments tended to show an even greater degree of nuclear enlargement and pleomorphism. Most cases demonstrated a single pattern throughout. Occasional IDC-P showed areas of blending between two patterns, a predominant pattern (>90%) and a minor pattern (<10%), which were always adjacent categories of the above listing, suggesting evolution along a continuum. Fig. 4. Pattern C intraductal carcinoma (IDC-P) with a solid architectural arrangement. The biphasic nature is less obvious with the loss of the central compartment (×48). Fig. 5. “Comedo”-like necrosis, identified most frequently in solid pattern C IDC-P (×60). Cytoplasmic Differentiation of IDC-P Patterns by Immunostaining PSA staining of “+” level or higher in pattern A was limited to the central compartment in 4 cases (Fig. 6) and occupied both compartments in the remaining 2 cases, one of which was the only “+++” case among 26 cases (Fig. 7). In pattern B, 4 of 15 cases had no PSA staining, but in the remaining 8 cases with “+” level or greater, staining was limited to the central compart- Fig. 6. Immunostaining of IDC-P with anti-prostate-specific antigen. Luminal staining is noted, while the pleomorphic perimetercell compartment demonstrates little immunoreactivity (×480). Intraductal Carcinoma of the Prostate 15 Fig. 7. PSA and androgen receptors detected in the two compartments (central and perimeter) in the three patterns (A–C) of intraductal carcinoma. PSA is mainly confined to the central compartment in all types, while androgen receptors are mostly identified in the perimeter except in a proportion of less differentiated tumors, where their expression is more widespread. Fig. 8. MUC-2 immunostaining (arrows) demonstrates positive staining focally in the central cell compartment of IDC-P (×60). ment. Likewise, in pattern C, 2 of 5 cases showed no staining, but the remaining 3 were centrally stained. Overall, the difference between central and perimeter staining was highly significant (P < 0.001). MUC-2, which stains intestinal goblet-cell mucus, was “+” or higher in 83% of pattern A cases, 80% of pattern B, and none of pattern C. Positive staining was selectively localized to the central compartment, with little or no perimeter stain (Fig. 8). Positively stained cells were often adjacent to pools of extracellular hematoxiphilic mucin. MUC-1 was negative in all cases. Androgen receptor was regarded as a differentiation antigen because of its exclusive presence in nonproliferative secretory cells in benign epithelium. In pattern A IDC-P, it was localized to the perimeter compartment in 100% of 6 cases (Fig. 7). This compartment is the analogue of the entire secretory lining of dysplastic (PIN) or normal epithelium. Hence, there was no shift in location (Fig. 9). In pattern B, 5 of 15 cancers had no staining, 4 stained only peripherally, and the remaining 6 were nearly equally expressed in both compartments. In pattern C, only 1 of 5 cases showed “+” staining in the central compartment, and it was equally stained in the perimeter. One case showed no stain and one was “±”. The remaining 2 were exclusively identified in the perimeter compartment. Strong staining was found in 50% of pattern A and 20% of patterns B and C. Overall perimeter staining for the androgen receptor was significantly different from the central staining (P < 0.01). 16 Cohen et al. Fig. 9. Androgen receptor immunostaining confirms nuclear positivity in the perimeter cell compartment of IDC-P (×120). Fig. 10. MIB-1 immunostaining identifies many nuclei of cells in cycle, which are restricted to the perimeter compartment in pattern A IDC-P (×120). Proliferation in IDC-P Patterns by Immunostaining Immunostaining for MIB-1 was positive in many nuclei of patterns A–C among the 26 prostatectomies. In nearly all cases, frequency and intensity of staining were comparable to the nuclei of adjacent invasive carcinoma and much higher than in benign glands. Distinctively, in patterns A and B alike, proliferationpositive nuclei were almost completely confined to the secretory cells of the perimeter compartment (Fig. 10). By contrast, in pattern C prostatectomies, MIB-1positive nuclei were scattered randomly throughout the entire luminal masses without regard for the central vs. perimeter compartments. Correlations Between IDC-P Pattern and Invasive Cancer or Cancer Stage The frequency of IDC-P pattern C among patients selected for prostatectomy (metastases not demonstrable clinically) was 20%, while 70% of patients with known metastases had IDC-P pattern C. Although the numbers are small and selection bias cannot be ruled out, a relationship between late stage and high IDC-P pattern is suggested. Prostatectomy cases were split into two sets according to whether Gleason sum in the invasive element was 7 or less as compared to those with a sum of 8 or more (Table I). The number of cases within these sets and their coexistent IDC-P patterns (A–C) were analyzed in a chi-square table which demonstrated a significant association (P < 0.01). Follow-up data were obtained on 17 of the Stanford prostatectomy cases and all of the Auckland cases (total of 24 of 26 cases). None of the 4 pattern A Stanford patients and only 1 of the 2 pattern A Auckland cases had progressed (biochemically) by 3 years . Of the 13 Stanford pat- tern B and C cases, 54% had relapsed, while all Auckland pattern B and C (3 and 2 cases, respectively) had clinical or biochemical evidence of tumor progression. DISCUSSION In a series of 26 prostatectomy samples with adenocarcinoma, all showing prominent sectors of intraductal, duct-spanning epithelial proliferation, we distinguished three duct-acinar histologic subpatterns. All three met the recently presented criteria for intraductal carcinoma (IDC-P), which has been proposed as one pathway of aggressive progression of invasive cancer . The IDC-P central cell compartment (lumen spanning cell plug) was the histological basis for distinguishing between subpatterns: it showed progressive loss of differentiation from pattern A through B to C. Cells in this compartment manifested a progressive loss of both number and aggregate area of microlumens within a duct or acinus. In close parallel, the centralcompartment cells developed larger, more malignantappearing nuclei. In pattern C, central cells reached a level of anaplasia equivalent to the perimeter cells, and there were no longer two distinguishable compartments. The perimeter cell compartment (cells in close proximity to the stroma) showed the same anaplastic features in all three subpatterns: these anaplastic features were indistinguishable from those of high-grade dysplasia (PIN), from which it has been proposed that IDC-P evolves . Morphological evolution of IDC-P from dysplasia is further supported by the similar patterns of immunostaining for the androgen receptor , which in both Intraductal Carcinoma of the Prostate 17 TABLE I. Tumor Parameters of 26 Prostate Cancers With IDC-P Gleason grade Case (n = 26) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 Major Minor IDC-P subtype % IDC-P Tumor volume (cm3)a Serum PSA (g/l) 3 4 3 4 3 3 4 3 3 4 5 3 3 3 3 3 4 3 4 4 5 4 3 3 4 4 3 3 3 3 4 3 4 3 4 5 4 3 4 3 4 4 4 4 4 4 4 5 4 4 5 5 A A A A A A B B B B B B B B B B B B B B B C C C C C 11–25 11–25 50–75 11–25 50–75 11–25 50–75 75–100 50–75 50–75 50–75 50–75 50–75 50–75 50–75 25–50 50–75 11–25 50–75 50–75 50–75 50–75 11–25 50–75 75–100 50–75 7.0 4.0 22.0 5.0 2.0 9.0 6.0 5.0 10.0 7.0 4.5 6.0 9.0 5.0 6.0 6.0 5.0 6.0 8.0 8.0 3.5 7.0 5.0 6.0 8.5 35.0 14.3 22.5 11.4 3.2 5.5 15.0 8.3 11.6 13.2 8.9 36.3 19.8 13.2 27.9 15.0 14.3 14.7 9.4 7.5 17.0 9.0 25.9 7.8 6.0 12.6 8.0 a Tumor volumes were rounded to the nearest 0.5 cm3. lesions identifies a population of proliferating secretory cells. Conversely, in the normal prostate gland, androgen receptors are identified only in secretory cells that are nonproliferating, which contrasts with dysplasia . Similarly in IDC-P, androgen-receptor staining is restricted to the perimeter (proliferative) compartment in pattern A, while in pattern B and more particularly pattern C, loss of distinction between the central and perimeter compartments parallels a loss of segregation of androgen-receptor staining, which we interpret as a loss of differentiation. Immunohistochemical staining for differentiation and proliferation markers also supports the concept that patterns A–C represent a continuum of increasing malignant potential which is assessed by loss of differentiation. The central compartment in pattern A displayed the cytoplasmic differentiation markers PSA and MUC-2, while proliferation was absent, a phenotype consistent with the benign appearance of these cells. All proliferation was confined to the perimeter compartment, where it was largely displaced from the basal cells to the secretory cells, as is described in dysplasia . In the central compartment, proliferation increased with advance in pattern reciprocal to decreasing differentiation, and in pattern C, proliferation in the central compartment was identical with the perimeter. We propose that the differences in histologic pattern and immunostaining in the central compartment are coordinated markers of the biologic capacity of the perimeter compartment to generate differentiated progeny. By contrast, in dysplasia, progeny do not survive in the lumen, presumably because they have not yet acquired the capacity to survive without stromal contact. Thus, in IDC-P the acquisition of the ability to survive among epithelial cells which are bound only to each other without stromal contact reflects an abrupt increase in malignant potential, drawing similar parallels to the distinction between invasive grade 4 cancer and the common type of grade 3 carcinoma. Our findings are of practical clinical significance: patient prognosis worsens in parallel with advancing pattern of IDC-P. This finding confirms that of a recent larger study of 108 cases , patterns corresponding to our pattern A plus B showed a 35% progression rate, while tumors corresponding to our pattern C pro- 18 Cohen et al. TABLE II. Features of IDC-P-Containing Cancers Which Differ From Those of Carcinomas Without IDC-P* Rubin et al. series  Cancer volume >4 cm3 Invasive carcinoma, Gleason score ⱖ7 Invasive carcinoma, Gleason score ⱖ8 Frequency of progression Wilcox et al. series  IDC-P+ (%) IDC-P− (%) IDC-P+ (%) IDC-P− (%) 60 80 17 41 54 30 61 14 29 49 6 12 *IDC-P+, IDC-P-containing cancers; IDC-P−, carcinomas without IDC-P. gressed in more than 80% of cases. This follows observations that IDC-P unsorted for pattern has an overall independent effect on prognosis , and progression frequency is fourfold higher [7,8] when compared to cancers without IDC-P (Table II). With IDC-P, the frequency of Gleason grade 4/5 cancer is reportedly twice that of other prostate cancers [7,8]. This partly explains the adverse influence of IDC-P overall (Table II). However, we have also found that the frequency of Gleason grades 4 and 5 also correlated with advancing pattern (C > B > A). A third adverse factor is cancer volume; cancers with IDC-P were four times more likely to exceed 4 cc than cancers without IDC-P elements [6–8]. Thus there is good evidence that the relationship between cancer volume, grade, and IDC-P play an important role in aggressive malignant behavior. By contrast, dysplasia (PIN) without the cribriform pattern stands apart from IDC-P as a lesion that is unable to influence volume, grade, or progression of the invasive carcinoma. The recent references [7,8] to intraductal carcinoma as “cribriform PIN” contradict the caution given by both sets of authors that this lesion must be considered separate from PIN, since it influences the clinical course of disease. We propose that this lesion should be classified as a distinct biological and histological entity and termed intraductal carcinoma of the prostate (IDC-P). In the Gleason grading system , patterns A and B of IDC-P were called grade 3 (invasive) cancer, cribriform subpattern. Pattern C was called grade 4 if it was composed of solid cell masses, and it was called grade 5 comedocarcinoma if it showed prominent central comedonecrosis. These entities should be recognized and called by the names that fit their true biological and clinical significance. For diagnostic purposes, only in Gleason grade 5 carcinoma does it seem not to matter whether the lesion is intraductal or invasive . CONCLUSIONS Confirming recent studies, we found evidence for the decreasing cure rate and increasing stage along the histologic pattern continuum of IDC-P. The pattern also appeared to correlate with histologic grade of the invasive component. The close kinship between suggested subpatterns of IDC-P and grades of invasive cancer supports the concept that IDC-P evolves synchronously with invasive high-grade elements. Consequently, an attempt should be made to eliminate disparities in terminology because of its adverse affects on accurate diagnosis. The identification of a single or several isolated cribriform ducts on needle biopsy has significant clinical implications and should stimulate multiple rebiopsies, with the conviction that invasive and probably high-grade/large-volume cancer will be found. ACKNOWLEDGMENTS We thank Dr. Amanda Segal (Consultant Pathologist, PathCentre, Perth, Western Australia) for her contribution, and Mr. Mario D’Antuono for biostatistical advice. We also acknowledge the Richard M. Lucas Foundation’s financial support to J.E.M, and Abbott Australasia Pty Ltd. for their financial support. REFERENCES 1. Dube V, Farrow G, Green L. Prostatic carcinoma of ductal origin. Cancer 1973;32:402–409. 2. Kovi J, Jackson M, Heshmat M. Ductal spread in prostatic carcinoma. Cancer 1985;56:1566–1573. 3. McNeal J, Reese J, Redwine E, Freiha F, Stamey T. Cribriform adenocarcinoma of the prostate. Cancer 1986;58:1714–1719. 4. Gleason DF. Histologic grading and clinical staging of prostatic carcinoma (the Veterans Administration Cooperative Urological Research Group). In: Tannenbaum M, editor. Urologic pathology: the prostate. Philadelphia: Lea and Febiger; 1977. p 171– 174. 5. Bostwick D, Amin M, Dundore P, Marsh W, Schultz D. Architectural patterns of high-grade prostatic intraepithelial neoplasia. Hum Pathol 1993;24:298–310. 6. McNeal JE, Yemoto CEM. Spread of adenocarcinoma within ducts and acini: morphologic and clinical correlations. Am J Surg Pathol 1996;20:802–814. 7. Rubin M, de La Taille A, Bagiella E, Olsson C, O’Toole K. Cribriform carcinoma of the prostate and cribriform intraepithelial Intraductal Carcinoma of the Prostate neoplasia: incidence and clinical implications. Am J Surg Pathol 1998;22:840–848. 8. Wilcox G, Soh S, Chakraborty S, Scardino P, Wheeler T. Patterns of high-grade prostatic intraepithelial neoplasia associated with clinically aggressive prostate cancer. Hum Pathol 1998;29:1119– 1123. 9. Cohen R, Chan W, Edgar S, Robinson E, Dodd N, Hoscek S, Mundy I. Prediction of pathologic stage and clinical outcome in prostate cancer; an improved pre-operative sequential model incorporating biopsy determined intraductal carcinoma. Br J Urol 1998;81:413–418. 10. Ohc Y, Hinoda Y, Irimura T, Imai K, Yechi A. Expression of sulphated carbohydrate chains detected by monoclonal anti- 19 body 91,9H in human gastric cancer tissues. Jpn J Cancer Res 1994;85:400–408. 11. Blank M, Klussmann E, Kruger-Krasagakes S, Schmitt-Graff A, Stolte M, Bornhoeft G, Stein H, Xing P, McKenzie I, Verstijnen C, Riecken EO, Hanski C. Expression of MUC-2 mucin in colorectal adenomas and carcinomas of different histological types. Int J Cancer 1994;59:301–306. 12. Leav I, McNeal JE, Kwan PW-L, Komminoth P, Merk FB. Androgen receptor expression in prostatic dysplasia (prostatic intraepithelial neoplasia) in the human prostate. An immunohistochemical and in situ hybridization study. Prostate 1996;29: 137–145.