Histochemistry
of the Prostate
WILLIAM C. ALLSBROOK,
JR, MD, AND WESLEY
Histochemistry, including immunohistochemistry, is helpful to the practicing pathologist in the diagnosis of prostatic carcinoma. Of equal importance, histochemistry is being increasingly used to study the pathobiology of the prostate. This article reviews these histochemical techniques and their applications. HUM PATHOL 23:297305. Copyright (c) 1992 by W.B. Saunders Company
Serum PAP has been used primarily to assist staging and monitoring of patients with prostatic carcinoma. Acid phosphatase was first demonstrated histochemically in normal and neoplastic prostate by Comori”’ using an enzymatic technique. In 1964, Parkin et al” using another enzymatic technique, studied acid phosphatase in prostatic carcinoma and noted decreased staining in carcinoma relative to normal I-issue and also decreased staining of carcinoma with decreasing diflerentiation. Jobsis et al’” subsequently reported a large group of cases showing positive indirect immunoperoxidase staining for PAP in 29 of 30 primary prostatic carcinomas, in ‘LOmetastatic prostatic carcinomas, and in none of 55 nonprostatic carcinomas. Prostate-specific antigen (PSA) was first reported in prostatic tissue by Wang et al.“’ Previously, it had been independently demonstrated in semen under the following names: ~30, gamma-seminoprotein, and semen E, antigen. The cDNA for PSA has been cloned and sequenced. “‘.I’ Prostate-specific antigen is a 33-kd glycoprotein that has been identified as a kallikrein-like serine protease.‘” It is produced by both benign and malignant prostatic epithelial cells. Prostate-specific antigen is presently replacing PAP as the preferred serum tnarker for prostatic carcinoma. Serum levels correlate with volume” of both prostatic carcinoma and benign prostatic hyperplasia, but the latter produces roughly only one tenth as much PSA as the former. Serum PSA is an excellent marker for monitoring response to therapy, including residual and recurrent disease, and may predict tumor stage.” The first report of immunoperoxidase staining for PSA occurred in 1981, when Papsidero et all9 detnonstrated positive staining in benign and tnalignant prostatic epithelium, but negative staining in a number of nonprostatic tissues. Nadji et al”” subsequently demonstrated PSA staining in normal and hyperplastic prostatic epithelium and in all 110 primary and metastatic prostatic carcinomas, but no staining in 78 nonprostatic malignancies. Staining of normal and hyperplastic prostatic epithelium for PAP and PSA was, with few exceptions, uniConversely, staining of carcinomas form and strong.“,“’ was less than that seen in benign prostatic tissue.‘“.” Decreased PAP in carcinomas, as compared with benign prostatic tissue, also has been demonstrated by enzymatic assays and molecular biological techniques.“5 In contrast to the benign tissue, most reports of prostatic carcinoma demonstrated heterogeneity in cellto-cell and/or field-to-field staining for both PAP and PSA. In these studies, an apparent correlation between staining-variability and increasing tumor grades was noted.2”,?7 A study using biochemical assays of PAP and PSA in fine needle aspirate tumor cytosols demonstrated decreasing PAP and PSA with increasing cytologic grade, tumor stage, and tumor ploidy.” A few studies of either
Pathologists are required to diagnose carcinoma of the prostate on a daily basis. At times, unequivocal diagnosis of malignancy is exceedingly difficult or even impossible. In the future, difficult cases will become even more common because of newer diagnostic modalities, automated biopsy techniques, and increasing numbers of surgical procedures for benign prostatic enlargement in the aging population. All of these factors are leading to the discovery of increased numbers of hitherto clinically inapparent carcinomas, many of which are well dilfercntiated and, consequently, potential diagnostic and therapeutic problems.‘.” Following the diagnosis of prostatic carcinoma, a urologist is faced with the dilemma of selecting the appropriate treatment, including the point when treatment should be instituted or whether treatment of a particular tumor is even necessary.” Clinical decisions following a histologic diagnosis of “atypical hyperplasia” or “prostatic intraepithelial neoplasia (PIN)“’ must be made. These diagnostic and clinical problems reflect fundamental deficits in our knowledge of the pathobiology of the prostate and emphasize the urgency for a clearer understanding of prostatic disease. Histochemistry, including imnlunohistochemistry, offers assistance in resolving these problems. PROSTATIC ACID PHOSPHATASE PROSTATE-SPECIFIC ANTIGEN
W. SIMMS, MD
AND
Prostatic acid phosphatase (PAP) was reported in the prostate, urine, and seminal fluid by Demuth” in 1925 (cited1 by Pontes”) and by Kutscher and Wolbergs7 in 1935. II. was subsequently shown to be present in skeletal metastases of prostatic carcinoma and in the serum of patients with metastatic prostatic carcinoma.” Prostatic acid phosphatase is a sialoglycoprotein with molecular weight of approximately 100 kd. Its production by benign and malignant prostatic epithelial cells is regulated by androgens. The cDNA encoding human PAP has been cloned and sequenced.” Serum levels can be measured by enzymatic or immunologic techniques, the latter being generally more sensitive and specific. F‘r~mthe I)cp;~rtn)mt of Pathology, School of Medicine, Medical College of- Georgia. Augusta. (iA. Adds-ess corrrspondencr and repr-int requests to William C. Allsbrook. Jr, Department of Pathology, School of Medicine, Medical (:ollege of Georgia, 1120 15th St, Augusta, GA 30912-360.5. (Zupyright 9 1992 by W.B. Saunders Companv 00-H-x1 :‘7/9”;/2303-00~ 1$f;.oo/n 297
HUMAN PATHOLOGY
Volume 23, No. 3 (March 1992)
PAP or PSA staining alone showed occasional negative staining of prostatic carcinoma. A large percentage of these cases were high-grade tumors or tumors with only small amounts of tumor available for study. Presently, there is no consistently “best” stain. While the great majority of prostatic carcinomas stain with both PAP and PSA, a few stain with only one or the other while, occasionally, neither stain is positive. Most “false-negatives” have occurred in high-grade tumors. A few reports have specifically evaluated PAP and PSA staining in primary and metastatic high-grade carcinomas. In some of these series, PAP staining was positive in a greater percentage of the casesz4*” and generally, but not always, PAP staining was stronger. Two recent studies compared monoclonal antibodies for PAP and PSA with polyclonal antisera.“.24 Abrahamsson et al*’ showed a decreased number of high-grade cases staining with PAP monoclonal antibody as compared with polyclonal PAP, polyclonal PSA, and monoclonal PSA. The likely explanation for this disparity is that a polyclonal antiserum identifies multiple antigenic sites while only a single epitope is recognized by a monoclonal antibody. Gallee et al 24 showed a decreased number of cases with positive monoclonal or polyclonal PSA staining relative to polyclonal PAP staining. Variable monoclonal and polyclonal PSA staining has not been a major problem. Variable or decreased tumor staining with both PAP and PSA has also been noted in some,” although not all, patients undergoing treatment for prostatic carcinoma. False-positive stainin for PAP has been reported ’ With the exce tion of posin a variety of tissues. “so_gY itive staining of some carcinoid tumors, g occasional primary and metastatic breast cancers, a renal cell carcinoma, occasional islet cell tumors,31 and a few primary bladder adenocarcinomas,32 it has not been a diagnostic problem. False-positive staining for PSA has been an even lesser problem. Cho and Epstein” reported faint PSA staining in a metastatic lung cancer in a mediastinal lymph node. In this case, PAP was negative. Epstein et al”’ demonstrated rare single cells positive for PSA in some bladder adenocarcinomas. Immunoperoxidase staining for PAP and PSA has been demonstrated in Skene’s glandss3 (the female homologue of the prostate) and in a carcinoma arising from Skene’s glands.34 Immunoperoxidase staining for PAP and PSA has been particularly useful to the diagnostic surgical pathologist in the identification of metastatic prostatic cancer and of morphologic variants of prostate cancer, in differentiating prostatic carcinoma secondarily involving the bladder from primary urothelial carcinoma and in differentiating urothelial or other carcinomas secondarily involving the prostate from prostatic carcinoma.“’ Decreased PAP and PSA staining has been reported in bone metastases decalcified with nitric acid or formic acid and sodium citrate, or in those fixed in Zenker’s fixative.35.36 With the exception of rare positive cells in two cases of primary bladder adenocarcinomas reported by Epstein et al, PSA was negative.“’ In the differential diagnosis of prostate cancer, both PAP and PSA stains should be performed. It is mandatory that the stains be interpreted in the appropriate 298
histologic and clinicopathologic setting with due consideration of false-positive and false-negative staining. With these provisos, a positive PAP or PSA stain rules in the diagnosis of prostatic carcinoma while negative PAP and PSA stains do not completely exclude a diagnosis of prostatic carcinoma, particularly one that is poorly differentiated. When both PAP and PSA stains are negative and the suspicion of prostatic carcinoma is still high, consideration should be given to increasing the antibody concentrations or incubation times. Hammond et alz3 recently correlated strong PAP staining of stage As-D1 prostatic carcinoma with survival. Prostatic acid phosphatase staining was independent of tumor grade although there was a trend associating PAP staining with hormonal responsiveness. Since PAP production is androgen dependent, variable staining might represent clones of cells with varying hormonal responsiveness. Further, cells with decreased staining appear less differentiated histochemically. Epstein and Eggleston have correlated progression in sta e A2 prostatic carcinoma with decreased PSA staining. 5 ’ McNeal et al recently demonstrated decreased PAP and PSA staining in PIN, noting that the cells of PIN are histochemically less differentiated.37 MUCINS Mucins are hexosamine-containing complex polysaccharides bound to variable amounts of protein by covalent bonds. They are produced by many different epithelial and connective tissues. Although mucins are heterogeneous compounds, they have been broadly categorized according to their reactions with a variety of histochemical stains. Many atypical and neoplastic lesions, including those of the prostate, show variations in mucin-staining patterns from the norm. In 1963, Foster and Levine3* used Mayer’s mucicarmine stain to demonstrate mucin in 63% of 125 surgical cases of prostatic carcinoma and in 36% of 22 autopsy cases of prostate carcinoma with metastases. The presence of mucin correlated with the degree of differentiation. Twelve poorly differentiated surgical tumors were uniformly negative. The decreased percentage of cases staining in the autopsy group was believed to reflect less differentiated tumors found at autopsy. In a follow-up study, the same researchers”” found that 61% of well-differentiated tumors and 23% of less-differentiated tumors stained with mucicarmine stain. They also found small amounts of mucicarmine positivity in acini of 50% of 25 prostate glands that did not contain carcinoma. In 1964, Franks et a14’ used a battery of stains to study mucin in 155 autopsy prostates, some normal and others containing latent or clinical carcinoma. Forty biopsy specimens of clinical prostatic carcinoma from living patients were also studied. Franks et a14’ demonstrated the presence of neutral mucins in normal prostates and in prostatic carcinoma. In addition, alcian blue-PAS staining techniques demonstrated that both latent and clinical prostatic carcinomas contained acid mucins with sulfated sialic acid residues. This staining
HISTOCHEMISTRY
OF THE PROSTATE
was most often seen in well and moderately differentiated carcinomas and in colloid carcinomas. It was seen only rarelv in benign prostates. A non-sulphated acid mucin was also found in colloid carcinomas and in carcinomas with extensive mucin production. These findings were subsequently largely confirmed by Hukill and Vidone,“’ who found a variety of staining reactions for mucin in prostatic carcinomas. Thirty-nine (78%) of their cases produced some type of mucin and 33 of these were acidic. Acid mucin was only rarely found in nonmalignant prostatic tissue, with the exception of occasional positive staining in corpora amylacea, thus also suggesting that demonstration of acid mucin is helpful in differentiating atypical acinar proliferations from welldifferentia.ted prostatic carcinoma. Later., ‘faylor,32 using Kreyberg and alcian blue-PAS methods, demonstrated acid mucin production in 17 of 27 (63%) prostatic carcinomas and re-stated that acid mucin stains were he1 ful in the diagnosis of prostatic ? carcinoma, Dollberg4. emphasized the critical importance of interpreting acid mucin staining in the appropriate morphologic context. True mutinous adenocarcinoma of the prostate is rare.4”‘.4” Epstein and I.ieberman recently defined mutinous adenocarcinoma of the prostate as a primary tumor containing more than 25% mucin lakes. Using this definition, they concluded that mutinous adenocarcinomas are aggressive tumors that do not respond, or respond only briefly, to hor1110nal nianiplllation.“”
(Allsbrook
& Simms)
finding might be useful in distinguishing certain benign lesions from well-differentiated prostatic carcinoma. These findings have been confirmed, with decreasing numbers of “basal cell-specific” cytokeratin positive cells in an increasingly discontinuous pattern from hyperplasia to atypical hyperglasia to carcinoma, the latter having no basal cells.5’,“. Chastonay et al used different antibodies and demonstrated basal cells in all their cases of hyperplasia but in none of their cases of atypical hyperplasia or carcinoma.“” Chou et al also mentioned positive staining in two of 26 cases of prostatic carcinoma, although it is not clear whether the staining is of basal cells or malignant epithelium5’ The presence or absence of immunoperoxidasestained basal cells is not an absolute histologic criterion in the differential diagnosis of carcinoma. These stains must be interpreted in the appropriate histologic context, the staining method must be reliable, and, further, allowance must be made for the reported attenuation of the basal cell layer and patchiness of staining due to formalin fixation. In general, the presence of basal cells is strong evidence that a difficult lesion is benign, whereas absence of basal cells does not invariably indicate malignancy. Brawer et al recently suggested that similar evaluation of basal cells is helpful in the diagnosis of carcinoma in previously irradiated prostates.“’ Bostwick and Brawer demonstrated decreased numbers of basal cells in an increasingly discontinuous pattern with increasing grade of PIN.“” Further, occasional buds of cells arising from the high-grade PIN did not have a basal cell layer, nor did surrounding infiltrating carcinoma. These investigators suggested that the buds were early invasive carcinoma arising from PIN. Hedrick and Epsteir?’ also reported decreased staining of basal cells in cases of high-gtade PIN, but the changes were not as striking as those reported by Bostwick and Brawer.‘” Finally, basal cell-specific cytokeratin staining has been demonstrated in the cells of basal cell hyper-
CYTOKERATINS Cvtokeratins are a complex group of polypeptides that form cytoskeletal intermediate filaments specific for epithelial cells. Different epithelial cell types may be delineated 1’;; cytokeratin composition.“” Expression of cytokeratins by epithelial cells is conserved in both benign and malignant neoplasms of all epithelia. including prostate. ‘,’ Cytokeratin investigations have centered around differential cytokeratin expression in benign and malignant prostatic epithelium”x and the use of “basal cell-specific” cytokeratin antibodies in the differential diagnosis Iof well-differentiated prostatic carcinoma. Well-differentiated prostatic carcinoma must be distin@shed f’i-om a number of benign lesions, including atypical adenomatous hyperplasia.’ (The articles by Scardmo et al. Brawer, and Mostofi et al in this sympc)sium [this issue] discuss in detail different aspects of the problems involved in distinguishing atypically benign and early malignant prostatic epithelium.) A helpful criterion for the diagnosis of carcinoma in this setting is I he absent-e of basal cells.“!’ Immunohistochemical stains for “basal cell-specific” (high molecular weight) cytokrratin make identification of basal cells much easier than with routine hematoxylin-eosin stains.“’ Brawer et al used immunoperoxidase stains for specific cytokeratins to dlemonstrate basal cells in normal and hyperplastic prostates.“’ Luminal secretory epithelial cells did not stain. Of particular note was the finding that no phenotypical basal cells were identified in their cases of ;~clenocarc.int,Ina. Brawer et al suggested that the latter
plasia,“,“7
BLOOD
GROUP
ANTIGENS
Blood group antigens are glycocortjugates (both glycoprotein and glycolipid) in which ant igenicity is cletermined by terminal oligosaccharides. hlodulation of their expression varies with different stages of tissue morphogenesis as well as with oncogenesis. Earlv studies of ABO blood group antigens in the prostate used specific red blood cell adherence techniques. Antigens were demonstrated in 36% to 100% of cases of‘ benign prostatic hyperplasia (BPH), but in no case of carcinoma. Immunoperoxidase studies of formalin-fixed, paraffin-embedded tissue have used a variety of mono&ma1 antibodies to blood groups A, B, and H(0)“8.“!’ as well as the lectin Ulex europaeus-I (UEA). which binds H(0) determinants. Reported staining of normal glands varies from study to study, ranging from all to most’!’ to less than 10% (increased to 50% to 90% with frozen tissue)Go,W to none to only central zone staining.“’ Nodules of benign prostatic hyperplasia exhibit staining for blood group antigens that is similar to or sli,ghtly less than
299
HUMAN PATHOLOGY
Volume 23, No. 3 (March 1992)
normal. In general, A and B antigen expression is lost in prostatic carcinoma, but UEA staining, which is greater and more diffuse than that seen in normal glands, is usually maintained. Abel et al”’ showed striking increases in UEA staining in cases of prostatic carcinoma, particularly in formalin-fixed, paraffin-embedded tissue but also in frozen tissue, the latter having more staining of normal tissue. There was no correlation with blood group or tumor stage. In the study of Perlman and Epstein, low-grade carcinomas had less UEA staining than either high-grade carcinoma or normal epithelium.“Y This study demonstrated enhanced UEA staining in PIN with a tendency for increased staining with higher grade of PIN. Interestingly, four of these cases contained hypercellular acini that did not meet the criteria for PIN but that expressed strong diffuse UEA staining, suggesting that they might be intermediate stages between normal and PIN. McNeal et al”’ recently demonstrated UEA staining in normal central zone glands but not in normal peripheral zone glands. They also demonstrated virtual absence of UEA staining in PIN. In cases of PIN involving the peripheral zone, adjacent normal epithelium expressed positive UEA staining characteristic of the normal central zone. In contrast, Perlman and Epstein found no differences in UEA staining of central and peripheral zones and positive staining in PIN.“” Two immunoperoxidase studies have examined Lewis antigen expression in prostatic epithelium.5g,“3 Both used a variety of monoclonal antibodies to Lewis a, Lewis b, and X antigens. A few cases with Lewis a-,b+ erythrocyte phenotype revealed positive staining of normal prostatic epithelium with Lewis a antibodies, a finding reported in other types of epithelium.“3 Lewis antigen staining of prostatic carcinoma was absent in one study5” and markedly decreased or absent in the other.‘j3 Bischof and Aumuller studied expression of Tantigen expression by binding of peanut agglutinin (PNA) in formalin-fixed, paraffin-embedded human prostates from various age groupsfi4 Peanut agglutinin binding to secretory cells strongly increased with puberty, increased to maximum levels in middle age, and rapidly decreased after the age of 50 years. Neuraminidase pretreatment uniformly increased staining, thus indicating the identified determinant to be “cryptic” due to masking by sialic acid. These investigators also suggested that PNA staining probably correlated with androgen levels. Ghazizadeh et al65 studied T antigen and cryptic T antigen expression in 39 cases of benign prostatic hyperplasia and 25 cases of prostatic carcinoma. Whereas most cases of BPH were T antigen negative and all demonstrated cryptic T expression following desialylation, 14 of 25 cases of prostatic carcinoma were T positive, four were both T negative and cryptic T negative, and seven were T negative but cryptic T positive. Most of the latter were grade 1 tumors. Conversely, Abel et al”’ identified strong (3+) PNA staining in normal prostate glands and weak (l+) staining in stroma in 39 patients aged 64 to 89 years, but noted no staining correlation with age. Soderstrom”” used fluorescent techniques to study PNA binding in prostates of patients 50 to 70 300
years of age. There was (I+) PNA binding in normal and hyperplastic prostate and (2+) PNA staining in carcinoma.
LECTINS Cellular glycoconjugates play important roles in a number of cell functions, including cell-cell and cellmatrix interactions, cell surface receptors, and cellular regulation. They comprise specific oligosaccharide sequences attached to particular sites on protein or lipid. Their synthesis is highly regulated. Glycoconjugates are variably expressed during development and oncogenesis and, in the latter, become important determinants of tumor behavior. The mechanisms of action of glycoconjugates in cell functions are complex and remain poorly understood. Lectins are naturally occurring proteins and glycoproteins of nonimmunological origin that are derived from both plants and animals. Lectins recognize and bind specific oligosaccharide sequences. To date, lectin binding has been the major histochemical tool for studying cellular glycoconjugates although monoclonal antibodies are being increasingly used. Lectin binding studies in the prostate have given both variable and contradictory results. Abel et al”’ noted “false-negative” lectin binding to be due either to masking of binding sites by sialic acid and other carbohydrates or to extraction of lipid-based binding sites during routine tissue processing. In general, studies have shown little diff‘erence in lectin binding between epithelium of normal and hyperplastic prostates.“‘.“’ Variations in lectin binding between carcinoma and hyperplasia or normal have been reported.“” Soderstrom”” also suggested that soybean agglutinin might be helpful in the diagnosis of carcinoma, although Loy et al”’ demonstrated soybean agglutinin binding in most cases of hyperplasia, atypical hyperplasia, and carcinoma. Recently, Abel et a16H,69studied binding of a panel of lectins in the normal adult rat and subsequently demonstrated variation from normal binding patterns in the Dunning rat prostate cancer model, including different binding results between hormone sensitive, nonmetastasizing (3327H) and hormone insensitive, metastasizing (MAT-LyLu) variants. These investigators suggested that comparable changes in lectin binding in human prostatic carcinoma might provide useful diagnostic or prognostic information.“* McNeal et al”’ used a panel of lectins to demonstrate different binding patterns in central and peripheral zones of the prostate as well as in PIN. Succinyl-wheat germ agglutinin, PNA, and UEA selectively stained the central zone. Reduced binding of all lectins occurred in cases of PIN. In three of four cases of PIN involving the peripheral zone, staining was identical to central zone staining while adjacent normal peripheral zone tissue also exhibited staining characteristic of central zone. (Specific aspects of lectin binding and oligosaccharide expression are discussed in greater detail in the article by Foster et al in this symposium [HUMAN PATHOLOGY,April 19921.)
HISTOCHEMISTRY
CELL PROLIFERATION
OF THE PROSTATE
STUDIES
:2ssessmmt of cellular proliferation hi, histochemic al techniques has been used as an adjunct to thymidine labeling and both static and flow cytometric DNA analy5es in attempts 10 identify associations between proliferativc indices and prognosis in prostate cancer. Of the c urrently ,:t\;ailable reagents, monoclonal antibody Ki67 is the best documented. Gerdes et a17”.” described the production of this antibody and found it to recognize an undefined proliferation-associated nuclear antigen expressed in (; 1. S, G’L, and M phases of the cell cycle, but IIO~ in GO and early Gl cells. Since then, immunoperoxidasr staining of Ki67, with establishment of Ki67 proliferation indices, has been successfully used to estimate proliferative cell fractions of various malignant tumors. A drawback of this technique is that staining rc-quires firozell tissue. Statistically significant differences between the mean Iii67 indices of benign prostatic hypertrophy and prostatic. carcinoma have been uniformly reported.7”m74 Raymond et al” noted significant correlations between Iii67 positi\ it?, including intensity, and increasing Gleason grade. A definite trend toward correlation with the .Jcwett stage was also noted. However, Oomens et a17’ found no significant correlation between Ki67 indices and grade Callee et a17’ found the highest Ki67 growth fractions tl:) be associated with either cribriform or solid undifferentiated growth patterns. This group also used Ki67 to assess the proliferative fraction of the transplanrable. hormone-dependent, human prostatic carcinoma, PC232, in response to androgen withdrawal and reinstitution.” I‘hev found that the number of Ki67positive tunior cells’in androgen-supplemented, tumorbearing HALB/c mice dropped significantly within 10 days of androgen deprivation but rose agiin within 3 days of androgen reinstitution and, hence, suggested that the Ki67 proliferation index could be used to monitor response IO hormone therapy. These results were (.onfirmed in a group of patients with prostatic carcinoma treal ed hy androgen ab1ation.74 A statistically significant decrease in the Ki67 index occurred wi&in I month after starting therapy, with further decreases at P and 3 months. The tumor of one patient who did not respond tc~ therapy exhibited an increase in the proliferativc cell frac.tion. NUCLEOLAR
ORGANIZING
REGIONS
Nucleolar organizing regions (NORs) are loops of chromosomal DNA encoding ribosomal RNA (rRNA). Ttley can be detected by their association with acidic, nonhistone, arhyrophilic protein (AgNOR) and are usually located in, or adjacent to, nucleoli. The numbers of AgNORs within nuclei c~orretate with growth fractions as shown h,!, S-phase determinations, including Ki67.7” (:urrently, there are only two studies of AgNOR counts in prostatic hyperplasia, dysplasia, and carcinoma. These have given conflicting results. Deschenes and Weidner,‘” using a manual counting procedure. fo~md AgNOR counts to increase progressively from 301
(Allsbrook
& Simms)
BPH through atypical adenomatous hyperplasia to carcinoma. Prostatic intraepithelial neoplasia and carcinoma AgNOR counts were significantlv different from those in BPH. No overlap was observed between mean AgNOR counts in lesions of carcinoma, BPH, or atypical adenonlatous hyperplasia. Nucleolar diameters were measured in this study and were found to increase J~Pgressively with increasing AgNOR counts. Conversely, Cheville et al, using image analysis, found no differences between AgNOR scores for hyperplasia, PIN, and aderlocarcinoma.‘x No difference was found between the three grades of PIN. However, a difference was found in AgNOR counts between high- and low-grade tumors. but not in intermediateor high-grade tumors. Some of the disparity in these studies may reflect counting techniques. The automated image analysis approach is ob,jective and reproducible and can c&mt large numbers of cells. Since the automated system only counts particles in one focal plane, many AgNORs may not be counted, thus resulting in a falsely tow, count. The manual method” allows careful focusing and more precise counting.
STEROID
HORMONE
RECEPTORS
Since Huggins and Hodges treated a k~oup of stage D prostatic carcinoma patients by castration or by estrogen injection, hormonal therapy aimed at lowering androgen levels has been the primary treatment for highstage prostatic carcinonla.79 Although bilateral orchiectomy continues to be the standard against which other forms of hormonal therapy are compared, luteinizing hormone releasing agonists and anti-androgens are currently used in clinical trials with good results. Hormonal manipulation has been associated with initial improvement in 60% to 80% of patients, but it is not curative. Trachtenberg and Walsh found the overall length of survival for men with metastatic prostate cancer varies with the length of time those patients respond to hormonal therapy.“” Steroid hormones regulate biochemical reactions in target cell nuclei by modulating production of RNA, apparently by enhancing the template function of target cell chromatirl.x’ The initial model for steroid hormone action involved binding of the steroid hormone to a cytosolic receptor with subsequent translocation to the nucleus.“’ Recent immunocytochemical data indicate that steroid hormone receptors reside in nuclei, both in the presence and absence of steroid. and that the receptor recovered in the cytosolic fraction of a tissue homogenate may represent receptor that is loosely associated with the nucleus and is hence ;m extraction artefact.x’ Many studies assessing steroid I-eceptors histochemically have used direct a plication (of fluoresceinconjugated steroid hormones. Ei In others, hormone is first incubated with the tissue and, subsequently, antibodies are directly or indirectly attached and marked with appropriate chromagens. A 70% to 80% correlation between histochemical and biochemical assays has been reported.x4
HUMAN PATHOLOGY
ANDROGEN
Volume 23, No. 3 (March 1992)
RECEPTORS
Androgen receptors (ARs) are intracellular ligandbinding proteins that influence transcription via direct interaction with specific DNA-responsive elements.85 The structure of the AR has been determined following cloning and sequencing studies.*” Both cytoplasmic and nuclear ARs have been described. Androgen receptors have been studied in BPH. Masai et al, using monoclonal antibody AN 1- 15, found nuclear staining in both glandular and stromal cells of BPH.s7 The concentration of AR in BPH compared with adenocarcinoma has been the object of a number of studies, with the majority showing no differences.s*-“’ Barrack et al found no difference in cytosolic AR, but did find an increased ratio of nuclear salt extractable to salt-resistant receptors in carcinoma compared with BPH and normal tissue.“’ The relationship of AR status to tumor grading is controversial. Studies using biochemical methodssass and immunoperoxidase methods show no correlation between nuclear AR concentration and histologic grade.“’ These results concur with a quantitative autoradiographic assay by Hulka et al.
of the Prostate
WILLIAM C. ALLSBROOK,
JR, MD, AND WESLEY
Histochemistry, including immunohistochemistry, is helpful to the practicing pathologist in the diagnosis of prostatic carcinoma. Of equal importance, histochemistry is being increasingly used to study the pathobiology of the prostate. This article reviews these histochemical techniques and their applications. HUM PATHOL 23:297305. Copyright (c) 1992 by W.B. Saunders Company
Serum PAP has been used primarily to assist staging and monitoring of patients with prostatic carcinoma. Acid phosphatase was first demonstrated histochemically in normal and neoplastic prostate by Comori”’ using an enzymatic technique. In 1964, Parkin et al” using another enzymatic technique, studied acid phosphatase in prostatic carcinoma and noted decreased staining in carcinoma relative to normal I-issue and also decreased staining of carcinoma with decreasing diflerentiation. Jobsis et al’” subsequently reported a large group of cases showing positive indirect immunoperoxidase staining for PAP in 29 of 30 primary prostatic carcinomas, in ‘LOmetastatic prostatic carcinomas, and in none of 55 nonprostatic carcinomas. Prostate-specific antigen (PSA) was first reported in prostatic tissue by Wang et al.“’ Previously, it had been independently demonstrated in semen under the following names: ~30, gamma-seminoprotein, and semen E, antigen. The cDNA for PSA has been cloned and sequenced. “‘.I’ Prostate-specific antigen is a 33-kd glycoprotein that has been identified as a kallikrein-like serine protease.‘” It is produced by both benign and malignant prostatic epithelial cells. Prostate-specific antigen is presently replacing PAP as the preferred serum tnarker for prostatic carcinoma. Serum levels correlate with volume” of both prostatic carcinoma and benign prostatic hyperplasia, but the latter produces roughly only one tenth as much PSA as the former. Serum PSA is an excellent marker for monitoring response to therapy, including residual and recurrent disease, and may predict tumor stage.” The first report of immunoperoxidase staining for PSA occurred in 1981, when Papsidero et all9 detnonstrated positive staining in benign and tnalignant prostatic epithelium, but negative staining in a number of nonprostatic tissues. Nadji et al”” subsequently demonstrated PSA staining in normal and hyperplastic prostatic epithelium and in all 110 primary and metastatic prostatic carcinomas, but no staining in 78 nonprostatic malignancies. Staining of normal and hyperplastic prostatic epithelium for PAP and PSA was, with few exceptions, uniConversely, staining of carcinomas form and strong.“,“’ was less than that seen in benign prostatic tissue.‘“.” Decreased PAP in carcinomas, as compared with benign prostatic tissue, also has been demonstrated by enzymatic assays and molecular biological techniques.“5 In contrast to the benign tissue, most reports of prostatic carcinoma demonstrated heterogeneity in cellto-cell and/or field-to-field staining for both PAP and PSA. In these studies, an apparent correlation between staining-variability and increasing tumor grades was noted.2”,?7 A study using biochemical assays of PAP and PSA in fine needle aspirate tumor cytosols demonstrated decreasing PAP and PSA with increasing cytologic grade, tumor stage, and tumor ploidy.” A few studies of either
Pathologists are required to diagnose carcinoma of the prostate on a daily basis. At times, unequivocal diagnosis of malignancy is exceedingly difficult or even impossible. In the future, difficult cases will become even more common because of newer diagnostic modalities, automated biopsy techniques, and increasing numbers of surgical procedures for benign prostatic enlargement in the aging population. All of these factors are leading to the discovery of increased numbers of hitherto clinically inapparent carcinomas, many of which are well dilfercntiated and, consequently, potential diagnostic and therapeutic problems.‘.” Following the diagnosis of prostatic carcinoma, a urologist is faced with the dilemma of selecting the appropriate treatment, including the point when treatment should be instituted or whether treatment of a particular tumor is even necessary.” Clinical decisions following a histologic diagnosis of “atypical hyperplasia” or “prostatic intraepithelial neoplasia (PIN)“’ must be made. These diagnostic and clinical problems reflect fundamental deficits in our knowledge of the pathobiology of the prostate and emphasize the urgency for a clearer understanding of prostatic disease. Histochemistry, including imnlunohistochemistry, offers assistance in resolving these problems. PROSTATIC ACID PHOSPHATASE PROSTATE-SPECIFIC ANTIGEN
W. SIMMS, MD
AND
Prostatic acid phosphatase (PAP) was reported in the prostate, urine, and seminal fluid by Demuth” in 1925 (cited1 by Pontes”) and by Kutscher and Wolbergs7 in 1935. II. was subsequently shown to be present in skeletal metastases of prostatic carcinoma and in the serum of patients with metastatic prostatic carcinoma.” Prostatic acid phosphatase is a sialoglycoprotein with molecular weight of approximately 100 kd. Its production by benign and malignant prostatic epithelial cells is regulated by androgens. The cDNA encoding human PAP has been cloned and sequenced.” Serum levels can be measured by enzymatic or immunologic techniques, the latter being generally more sensitive and specific. F‘r~mthe I)cp;~rtn)mt of Pathology, School of Medicine, Medical College of- Georgia. Augusta. (iA. Adds-ess corrrspondencr and repr-int requests to William C. Allsbrook. Jr, Department of Pathology, School of Medicine, Medical (:ollege of Georgia, 1120 15th St, Augusta, GA 30912-360.5. (Zupyright 9 1992 by W.B. Saunders Companv 00-H-x1 :‘7/9”;/2303-00~ 1$f;.oo/n 297
HUMAN PATHOLOGY
Volume 23, No. 3 (March 1992)
PAP or PSA staining alone showed occasional negative staining of prostatic carcinoma. A large percentage of these cases were high-grade tumors or tumors with only small amounts of tumor available for study. Presently, there is no consistently “best” stain. While the great majority of prostatic carcinomas stain with both PAP and PSA, a few stain with only one or the other while, occasionally, neither stain is positive. Most “false-negatives” have occurred in high-grade tumors. A few reports have specifically evaluated PAP and PSA staining in primary and metastatic high-grade carcinomas. In some of these series, PAP staining was positive in a greater percentage of the casesz4*” and generally, but not always, PAP staining was stronger. Two recent studies compared monoclonal antibodies for PAP and PSA with polyclonal antisera.“.24 Abrahamsson et al*’ showed a decreased number of high-grade cases staining with PAP monoclonal antibody as compared with polyclonal PAP, polyclonal PSA, and monoclonal PSA. The likely explanation for this disparity is that a polyclonal antiserum identifies multiple antigenic sites while only a single epitope is recognized by a monoclonal antibody. Gallee et al 24 showed a decreased number of cases with positive monoclonal or polyclonal PSA staining relative to polyclonal PAP staining. Variable monoclonal and polyclonal PSA staining has not been a major problem. Variable or decreased tumor staining with both PAP and PSA has also been noted in some,” although not all, patients undergoing treatment for prostatic carcinoma. False-positive stainin for PAP has been reported ’ With the exce tion of posin a variety of tissues. “so_gY itive staining of some carcinoid tumors, g occasional primary and metastatic breast cancers, a renal cell carcinoma, occasional islet cell tumors,31 and a few primary bladder adenocarcinomas,32 it has not been a diagnostic problem. False-positive staining for PSA has been an even lesser problem. Cho and Epstein” reported faint PSA staining in a metastatic lung cancer in a mediastinal lymph node. In this case, PAP was negative. Epstein et al”’ demonstrated rare single cells positive for PSA in some bladder adenocarcinomas. Immunoperoxidase staining for PAP and PSA has been demonstrated in Skene’s glandss3 (the female homologue of the prostate) and in a carcinoma arising from Skene’s glands.34 Immunoperoxidase staining for PAP and PSA has been particularly useful to the diagnostic surgical pathologist in the identification of metastatic prostatic cancer and of morphologic variants of prostate cancer, in differentiating prostatic carcinoma secondarily involving the bladder from primary urothelial carcinoma and in differentiating urothelial or other carcinomas secondarily involving the prostate from prostatic carcinoma.“’ Decreased PAP and PSA staining has been reported in bone metastases decalcified with nitric acid or formic acid and sodium citrate, or in those fixed in Zenker’s fixative.35.36 With the exception of rare positive cells in two cases of primary bladder adenocarcinomas reported by Epstein et al, PSA was negative.“’ In the differential diagnosis of prostate cancer, both PAP and PSA stains should be performed. It is mandatory that the stains be interpreted in the appropriate 298
histologic and clinicopathologic setting with due consideration of false-positive and false-negative staining. With these provisos, a positive PAP or PSA stain rules in the diagnosis of prostatic carcinoma while negative PAP and PSA stains do not completely exclude a diagnosis of prostatic carcinoma, particularly one that is poorly differentiated. When both PAP and PSA stains are negative and the suspicion of prostatic carcinoma is still high, consideration should be given to increasing the antibody concentrations or incubation times. Hammond et alz3 recently correlated strong PAP staining of stage As-D1 prostatic carcinoma with survival. Prostatic acid phosphatase staining was independent of tumor grade although there was a trend associating PAP staining with hormonal responsiveness. Since PAP production is androgen dependent, variable staining might represent clones of cells with varying hormonal responsiveness. Further, cells with decreased staining appear less differentiated histochemically. Epstein and Eggleston have correlated progression in sta e A2 prostatic carcinoma with decreased PSA staining. 5 ’ McNeal et al recently demonstrated decreased PAP and PSA staining in PIN, noting that the cells of PIN are histochemically less differentiated.37 MUCINS Mucins are hexosamine-containing complex polysaccharides bound to variable amounts of protein by covalent bonds. They are produced by many different epithelial and connective tissues. Although mucins are heterogeneous compounds, they have been broadly categorized according to their reactions with a variety of histochemical stains. Many atypical and neoplastic lesions, including those of the prostate, show variations in mucin-staining patterns from the norm. In 1963, Foster and Levine3* used Mayer’s mucicarmine stain to demonstrate mucin in 63% of 125 surgical cases of prostatic carcinoma and in 36% of 22 autopsy cases of prostate carcinoma with metastases. The presence of mucin correlated with the degree of differentiation. Twelve poorly differentiated surgical tumors were uniformly negative. The decreased percentage of cases staining in the autopsy group was believed to reflect less differentiated tumors found at autopsy. In a follow-up study, the same researchers”” found that 61% of well-differentiated tumors and 23% of less-differentiated tumors stained with mucicarmine stain. They also found small amounts of mucicarmine positivity in acini of 50% of 25 prostate glands that did not contain carcinoma. In 1964, Franks et a14’ used a battery of stains to study mucin in 155 autopsy prostates, some normal and others containing latent or clinical carcinoma. Forty biopsy specimens of clinical prostatic carcinoma from living patients were also studied. Franks et a14’ demonstrated the presence of neutral mucins in normal prostates and in prostatic carcinoma. In addition, alcian blue-PAS staining techniques demonstrated that both latent and clinical prostatic carcinomas contained acid mucins with sulfated sialic acid residues. This staining
HISTOCHEMISTRY
OF THE PROSTATE
was most often seen in well and moderately differentiated carcinomas and in colloid carcinomas. It was seen only rarelv in benign prostates. A non-sulphated acid mucin was also found in colloid carcinomas and in carcinomas with extensive mucin production. These findings were subsequently largely confirmed by Hukill and Vidone,“’ who found a variety of staining reactions for mucin in prostatic carcinomas. Thirty-nine (78%) of their cases produced some type of mucin and 33 of these were acidic. Acid mucin was only rarely found in nonmalignant prostatic tissue, with the exception of occasional positive staining in corpora amylacea, thus also suggesting that demonstration of acid mucin is helpful in differentiating atypical acinar proliferations from welldifferentia.ted prostatic carcinoma. Later., ‘faylor,32 using Kreyberg and alcian blue-PAS methods, demonstrated acid mucin production in 17 of 27 (63%) prostatic carcinomas and re-stated that acid mucin stains were he1 ful in the diagnosis of prostatic ? carcinoma, Dollberg4. emphasized the critical importance of interpreting acid mucin staining in the appropriate morphologic context. True mutinous adenocarcinoma of the prostate is rare.4”‘.4” Epstein and I.ieberman recently defined mutinous adenocarcinoma of the prostate as a primary tumor containing more than 25% mucin lakes. Using this definition, they concluded that mutinous adenocarcinomas are aggressive tumors that do not respond, or respond only briefly, to hor1110nal nianiplllation.“”
(Allsbrook
& Simms)
finding might be useful in distinguishing certain benign lesions from well-differentiated prostatic carcinoma. These findings have been confirmed, with decreasing numbers of “basal cell-specific” cytokeratin positive cells in an increasingly discontinuous pattern from hyperplasia to atypical hyperglasia to carcinoma, the latter having no basal cells.5’,“. Chastonay et al used different antibodies and demonstrated basal cells in all their cases of hyperplasia but in none of their cases of atypical hyperplasia or carcinoma.“” Chou et al also mentioned positive staining in two of 26 cases of prostatic carcinoma, although it is not clear whether the staining is of basal cells or malignant epithelium5’ The presence or absence of immunoperoxidasestained basal cells is not an absolute histologic criterion in the differential diagnosis of carcinoma. These stains must be interpreted in the appropriate histologic context, the staining method must be reliable, and, further, allowance must be made for the reported attenuation of the basal cell layer and patchiness of staining due to formalin fixation. In general, the presence of basal cells is strong evidence that a difficult lesion is benign, whereas absence of basal cells does not invariably indicate malignancy. Brawer et al recently suggested that similar evaluation of basal cells is helpful in the diagnosis of carcinoma in previously irradiated prostates.“’ Bostwick and Brawer demonstrated decreased numbers of basal cells in an increasingly discontinuous pattern with increasing grade of PIN.“” Further, occasional buds of cells arising from the high-grade PIN did not have a basal cell layer, nor did surrounding infiltrating carcinoma. These investigators suggested that the buds were early invasive carcinoma arising from PIN. Hedrick and Epsteir?’ also reported decreased staining of basal cells in cases of high-gtade PIN, but the changes were not as striking as those reported by Bostwick and Brawer.‘” Finally, basal cell-specific cytokeratin staining has been demonstrated in the cells of basal cell hyper-
CYTOKERATINS Cvtokeratins are a complex group of polypeptides that form cytoskeletal intermediate filaments specific for epithelial cells. Different epithelial cell types may be delineated 1’;; cytokeratin composition.“” Expression of cytokeratins by epithelial cells is conserved in both benign and malignant neoplasms of all epithelia. including prostate. ‘,’ Cytokeratin investigations have centered around differential cytokeratin expression in benign and malignant prostatic epithelium”x and the use of “basal cell-specific” cytokeratin antibodies in the differential diagnosis Iof well-differentiated prostatic carcinoma. Well-differentiated prostatic carcinoma must be distin@shed f’i-om a number of benign lesions, including atypical adenomatous hyperplasia.’ (The articles by Scardmo et al. Brawer, and Mostofi et al in this sympc)sium [this issue] discuss in detail different aspects of the problems involved in distinguishing atypically benign and early malignant prostatic epithelium.) A helpful criterion for the diagnosis of carcinoma in this setting is I he absent-e of basal cells.“!’ Immunohistochemical stains for “basal cell-specific” (high molecular weight) cytokrratin make identification of basal cells much easier than with routine hematoxylin-eosin stains.“’ Brawer et al used immunoperoxidase stains for specific cytokeratins to dlemonstrate basal cells in normal and hyperplastic prostates.“’ Luminal secretory epithelial cells did not stain. Of particular note was the finding that no phenotypical basal cells were identified in their cases of ;~clenocarc.int,Ina. Brawer et al suggested that the latter
plasia,“,“7
BLOOD
GROUP
ANTIGENS
Blood group antigens are glycocortjugates (both glycoprotein and glycolipid) in which ant igenicity is cletermined by terminal oligosaccharides. hlodulation of their expression varies with different stages of tissue morphogenesis as well as with oncogenesis. Earlv studies of ABO blood group antigens in the prostate used specific red blood cell adherence techniques. Antigens were demonstrated in 36% to 100% of cases of‘ benign prostatic hyperplasia (BPH), but in no case of carcinoma. Immunoperoxidase studies of formalin-fixed, paraffin-embedded tissue have used a variety of mono&ma1 antibodies to blood groups A, B, and H(0)“8.“!’ as well as the lectin Ulex europaeus-I (UEA). which binds H(0) determinants. Reported staining of normal glands varies from study to study, ranging from all to most’!’ to less than 10% (increased to 50% to 90% with frozen tissue)Go,W to none to only central zone staining.“’ Nodules of benign prostatic hyperplasia exhibit staining for blood group antigens that is similar to or sli,ghtly less than
299
HUMAN PATHOLOGY
Volume 23, No. 3 (March 1992)
normal. In general, A and B antigen expression is lost in prostatic carcinoma, but UEA staining, which is greater and more diffuse than that seen in normal glands, is usually maintained. Abel et al”’ showed striking increases in UEA staining in cases of prostatic carcinoma, particularly in formalin-fixed, paraffin-embedded tissue but also in frozen tissue, the latter having more staining of normal tissue. There was no correlation with blood group or tumor stage. In the study of Perlman and Epstein, low-grade carcinomas had less UEA staining than either high-grade carcinoma or normal epithelium.“Y This study demonstrated enhanced UEA staining in PIN with a tendency for increased staining with higher grade of PIN. Interestingly, four of these cases contained hypercellular acini that did not meet the criteria for PIN but that expressed strong diffuse UEA staining, suggesting that they might be intermediate stages between normal and PIN. McNeal et al”’ recently demonstrated UEA staining in normal central zone glands but not in normal peripheral zone glands. They also demonstrated virtual absence of UEA staining in PIN. In cases of PIN involving the peripheral zone, adjacent normal epithelium expressed positive UEA staining characteristic of the normal central zone. In contrast, Perlman and Epstein found no differences in UEA staining of central and peripheral zones and positive staining in PIN.“” Two immunoperoxidase studies have examined Lewis antigen expression in prostatic epithelium.5g,“3 Both used a variety of monoclonal antibodies to Lewis a, Lewis b, and X antigens. A few cases with Lewis a-,b+ erythrocyte phenotype revealed positive staining of normal prostatic epithelium with Lewis a antibodies, a finding reported in other types of epithelium.“3 Lewis antigen staining of prostatic carcinoma was absent in one study5” and markedly decreased or absent in the other.‘j3 Bischof and Aumuller studied expression of Tantigen expression by binding of peanut agglutinin (PNA) in formalin-fixed, paraffin-embedded human prostates from various age groupsfi4 Peanut agglutinin binding to secretory cells strongly increased with puberty, increased to maximum levels in middle age, and rapidly decreased after the age of 50 years. Neuraminidase pretreatment uniformly increased staining, thus indicating the identified determinant to be “cryptic” due to masking by sialic acid. These investigators also suggested that PNA staining probably correlated with androgen levels. Ghazizadeh et al65 studied T antigen and cryptic T antigen expression in 39 cases of benign prostatic hyperplasia and 25 cases of prostatic carcinoma. Whereas most cases of BPH were T antigen negative and all demonstrated cryptic T expression following desialylation, 14 of 25 cases of prostatic carcinoma were T positive, four were both T negative and cryptic T negative, and seven were T negative but cryptic T positive. Most of the latter were grade 1 tumors. Conversely, Abel et al”’ identified strong (3+) PNA staining in normal prostate glands and weak (l+) staining in stroma in 39 patients aged 64 to 89 years, but noted no staining correlation with age. Soderstrom”” used fluorescent techniques to study PNA binding in prostates of patients 50 to 70 300
years of age. There was (I+) PNA binding in normal and hyperplastic prostate and (2+) PNA staining in carcinoma.
LECTINS Cellular glycoconjugates play important roles in a number of cell functions, including cell-cell and cellmatrix interactions, cell surface receptors, and cellular regulation. They comprise specific oligosaccharide sequences attached to particular sites on protein or lipid. Their synthesis is highly regulated. Glycoconjugates are variably expressed during development and oncogenesis and, in the latter, become important determinants of tumor behavior. The mechanisms of action of glycoconjugates in cell functions are complex and remain poorly understood. Lectins are naturally occurring proteins and glycoproteins of nonimmunological origin that are derived from both plants and animals. Lectins recognize and bind specific oligosaccharide sequences. To date, lectin binding has been the major histochemical tool for studying cellular glycoconjugates although monoclonal antibodies are being increasingly used. Lectin binding studies in the prostate have given both variable and contradictory results. Abel et al”’ noted “false-negative” lectin binding to be due either to masking of binding sites by sialic acid and other carbohydrates or to extraction of lipid-based binding sites during routine tissue processing. In general, studies have shown little diff‘erence in lectin binding between epithelium of normal and hyperplastic prostates.“‘.“’ Variations in lectin binding between carcinoma and hyperplasia or normal have been reported.“” Soderstrom”” also suggested that soybean agglutinin might be helpful in the diagnosis of carcinoma, although Loy et al”’ demonstrated soybean agglutinin binding in most cases of hyperplasia, atypical hyperplasia, and carcinoma. Recently, Abel et a16H,69studied binding of a panel of lectins in the normal adult rat and subsequently demonstrated variation from normal binding patterns in the Dunning rat prostate cancer model, including different binding results between hormone sensitive, nonmetastasizing (3327H) and hormone insensitive, metastasizing (MAT-LyLu) variants. These investigators suggested that comparable changes in lectin binding in human prostatic carcinoma might provide useful diagnostic or prognostic information.“* McNeal et al”’ used a panel of lectins to demonstrate different binding patterns in central and peripheral zones of the prostate as well as in PIN. Succinyl-wheat germ agglutinin, PNA, and UEA selectively stained the central zone. Reduced binding of all lectins occurred in cases of PIN. In three of four cases of PIN involving the peripheral zone, staining was identical to central zone staining while adjacent normal peripheral zone tissue also exhibited staining characteristic of central zone. (Specific aspects of lectin binding and oligosaccharide expression are discussed in greater detail in the article by Foster et al in this symposium [HUMAN PATHOLOGY,April 19921.)
HISTOCHEMISTRY
CELL PROLIFERATION
OF THE PROSTATE
STUDIES
:2ssessmmt of cellular proliferation hi, histochemic al techniques has been used as an adjunct to thymidine labeling and both static and flow cytometric DNA analy5es in attempts 10 identify associations between proliferativc indices and prognosis in prostate cancer. Of the c urrently ,:t\;ailable reagents, monoclonal antibody Ki67 is the best documented. Gerdes et a17”.” described the production of this antibody and found it to recognize an undefined proliferation-associated nuclear antigen expressed in (; 1. S, G’L, and M phases of the cell cycle, but IIO~ in GO and early Gl cells. Since then, immunoperoxidasr staining of Ki67, with establishment of Ki67 proliferation indices, has been successfully used to estimate proliferative cell fractions of various malignant tumors. A drawback of this technique is that staining rc-quires firozell tissue. Statistically significant differences between the mean Iii67 indices of benign prostatic hypertrophy and prostatic. carcinoma have been uniformly reported.7”m74 Raymond et al” noted significant correlations between Iii67 positi\ it?, including intensity, and increasing Gleason grade. A definite trend toward correlation with the .Jcwett stage was also noted. However, Oomens et a17’ found no significant correlation between Ki67 indices and grade Callee et a17’ found the highest Ki67 growth fractions tl:) be associated with either cribriform or solid undifferentiated growth patterns. This group also used Ki67 to assess the proliferative fraction of the transplanrable. hormone-dependent, human prostatic carcinoma, PC232, in response to androgen withdrawal and reinstitution.” I‘hev found that the number of Ki67positive tunior cells’in androgen-supplemented, tumorbearing HALB/c mice dropped significantly within 10 days of androgen deprivation but rose agiin within 3 days of androgen reinstitution and, hence, suggested that the Ki67 proliferation index could be used to monitor response IO hormone therapy. These results were (.onfirmed in a group of patients with prostatic carcinoma treal ed hy androgen ab1ation.74 A statistically significant decrease in the Ki67 index occurred wi&in I month after starting therapy, with further decreases at P and 3 months. The tumor of one patient who did not respond tc~ therapy exhibited an increase in the proliferativc cell frac.tion. NUCLEOLAR
ORGANIZING
REGIONS
Nucleolar organizing regions (NORs) are loops of chromosomal DNA encoding ribosomal RNA (rRNA). Ttley can be detected by their association with acidic, nonhistone, arhyrophilic protein (AgNOR) and are usually located in, or adjacent to, nucleoli. The numbers of AgNORs within nuclei c~orretate with growth fractions as shown h,!, S-phase determinations, including Ki67.7” (:urrently, there are only two studies of AgNOR counts in prostatic hyperplasia, dysplasia, and carcinoma. These have given conflicting results. Deschenes and Weidner,‘” using a manual counting procedure. fo~md AgNOR counts to increase progressively from 301
(Allsbrook
& Simms)
BPH through atypical adenomatous hyperplasia to carcinoma. Prostatic intraepithelial neoplasia and carcinoma AgNOR counts were significantlv different from those in BPH. No overlap was observed between mean AgNOR counts in lesions of carcinoma, BPH, or atypical adenonlatous hyperplasia. Nucleolar diameters were measured in this study and were found to increase J~Pgressively with increasing AgNOR counts. Conversely, Cheville et al, using image analysis, found no differences between AgNOR scores for hyperplasia, PIN, and aderlocarcinoma.‘x No difference was found between the three grades of PIN. However, a difference was found in AgNOR counts between high- and low-grade tumors. but not in intermediateor high-grade tumors. Some of the disparity in these studies may reflect counting techniques. The automated image analysis approach is ob,jective and reproducible and can c&mt large numbers of cells. Since the automated system only counts particles in one focal plane, many AgNORs may not be counted, thus resulting in a falsely tow, count. The manual method” allows careful focusing and more precise counting.
STEROID
HORMONE
RECEPTORS
Since Huggins and Hodges treated a k~oup of stage D prostatic carcinoma patients by castration or by estrogen injection, hormonal therapy aimed at lowering androgen levels has been the primary treatment for highstage prostatic carcinonla.79 Although bilateral orchiectomy continues to be the standard against which other forms of hormonal therapy are compared, luteinizing hormone releasing agonists and anti-androgens are currently used in clinical trials with good results. Hormonal manipulation has been associated with initial improvement in 60% to 80% of patients, but it is not curative. Trachtenberg and Walsh found the overall length of survival for men with metastatic prostate cancer varies with the length of time those patients respond to hormonal therapy.“” Steroid hormones regulate biochemical reactions in target cell nuclei by modulating production of RNA, apparently by enhancing the template function of target cell chromatirl.x’ The initial model for steroid hormone action involved binding of the steroid hormone to a cytosolic receptor with subsequent translocation to the nucleus.“’ Recent immunocytochemical data indicate that steroid hormone receptors reside in nuclei, both in the presence and absence of steroid. and that the receptor recovered in the cytosolic fraction of a tissue homogenate may represent receptor that is loosely associated with the nucleus and is hence ;m extraction artefact.x’ Many studies assessing steroid I-eceptors histochemically have used direct a plication (of fluoresceinconjugated steroid hormones. Ei In others, hormone is first incubated with the tissue and, subsequently, antibodies are directly or indirectly attached and marked with appropriate chromagens. A 70% to 80% correlation between histochemical and biochemical assays has been reported.x4
HUMAN PATHOLOGY
ANDROGEN
Volume 23, No. 3 (March 1992)
RECEPTORS
Androgen receptors (ARs) are intracellular ligandbinding proteins that influence transcription via direct interaction with specific DNA-responsive elements.85 The structure of the AR has been determined following cloning and sequencing studies.*” Both cytoplasmic and nuclear ARs have been described. Androgen receptors have been studied in BPH. Masai et al, using monoclonal antibody AN 1- 15, found nuclear staining in both glandular and stromal cells of BPH.s7 The concentration of AR in BPH compared with adenocarcinoma has been the object of a number of studies, with the majority showing no differences.s*-“’ Barrack et al found no difference in cytosolic AR, but did find an increased ratio of nuclear salt extractable to salt-resistant receptors in carcinoma compared with BPH and normal tissue.“’ The relationship of AR status to tumor grading is controversial. Studies using biochemical methodssass and immunoperoxidase methods show no correlation between nuclear AR concentration and histologic grade.“’ These results concur with a quantitative autoradiographic assay by Hulka et al.
