 |
|
|
|
|
|
 |
 |
 |
 |
 |
|
Immunostaining as a Diagnostic Aid in Cytopathologic Study of Upper Urinary Tract Urothelial Carcinoma
Kien T. Mai, M.D., F.R.C.P.C., Iris Teo, M.D., Susan J. Robertson, M.D., F.R.C.P.C., E. Celia Marginean, M.D., F.R.C.P.C., Shahidul Islam, M.D., Ph.D., F.A.C.P., and Hossein M. Yazdi, M.D. Objective
To evaluate preoperative diagnosis of low-grade urothelial carcinoma (LGUC) and urothelial neoplasms of unknown malignant potential (UMP UN) of the upper urinary tract (UUT) and its role in disease management, especially in the context of nephron-sparing treatment possibilities.
Study Design
Wash and brush ureteral specimens of LGUC/UMP UN of the UUT with histopathologic correlation were retrieved at our institution for 7 years and studied along with 7 ureteral specimens from nonneoplastic ureteral lesions.
Results Of 30 specimens from 25 LGUC/UMP UN, 5 were negative for tumor cells and 3 showed cytologic atypia. The remaining 22 contained tumor cells with characteristic features of urothelial carcinoma, including hard and soft criteria. The 4 hard criteria included branching stromal cores, dyshesive cell networks, 3-dimensional papillary clusters with stromal core and atypia associated with CK20-positive cells. The 2 soft criteria were hypercellularity and atypia in CK20-negative cells. All LGUC/UMP UN of the UUT were associated with at least 1 hard criterion or both soft criteria.
Conclusion Branching stromal cores, 3-dimensional papillary clusters, dyshesive cell networks and CK20-positive atypia immunostaining appear specific for LGUC/UMP UN of the UUT but are seen in few cases. Combined soft and hard criteria will increase sensitivity to 83%. (Acta Cytol 2009;53:611–618)
Keywords: cytology, immunohistochemistry, ureter, urothelial carcinoma.
 LGUC and UMP UN of the UUT can be confidently diagnosed with wash or brush cytology because of the architecture of stromal and cell arrangements in urine cytologic specimens…. Urothelial carcinoma (UC) of the upper urinary tract (UUT UC) accounts for 5% of the total UC cases. The incidence of UUT UC in follow-up of UC of the urinary bladder ranges from 0.7% to 4%, and UUT UC is associated with urinary bladder carcinoma in 20–50% of cases.1 Carcinoma in the upper tract that occurs during the surveillance period after treatment of UC of the urinary bladder is usually identified by an imaging technique visualizing the lesion or hydronephrosis in association with urine cytology. Voided, catheterized and cytoscopic urine cytology positive for atypical or malignant cells in the absence of carcinoma in the urinary bladder is usually one of the earliest indications for further investigation for UC of the UUT. Biopsy of the lesion is rarely performed in the UUT,2-4 so ureteral wash or brush cytology plays a major role in the preoperative diagnosis of the UC of the UUT.5-11 With the development of flexible ureteroscopes, nephron-sparing treatment of low-grade and low-stage UUT is feasible.12-20 Thus the cytopathologic diagnosis of low-grade UC (LGUC) plays an important role in management of the disease. Despite this, there are only a few cytopathologic studies of LGUC of the UUT. As a result, criteria for a positive cytopathologic diagnosis of these lesions are based on those used in urinary bladder UC. Furthermore, recent advances in immunohistochemical studies have suggested that many UCs, particularly LGUC with intermediate degree of cytologic atypia, display an abnormal pattern of reactivity for cytokeratin 20 (CK20), whereas nonneoplastic urothelium and many LGUCs display the normal pattern of CK20 reactivity, mainly in the umbrella cell layer.21-32 In this study, we review cytologic specimens from cases of UC of the UUT with histopathologic correlation.
Materials and Methods
Pilot Study
Cytologic specimens of ureteral washing or brushing from 6 cases of the most recent ureteronephrectomy for UUT UC were reviewed to develop criteria for positive diagnosis of the UUT UC.
Main Study
Files at the Anatomical Pathology Laboratory of our institution were reviewed to identify remaining cases of ureterectomy and ureteronephrectomy for UUT UC. Surgical cases without cytologic specimens of ureteral washing or brushing were excluded from the study. Cases from the pilot study and main study were classified as group 1. The control group (group 2) was composed of patients with a history of pyelonephrostomy, patients with ureteral fibroepithelial polyps and patients in surveillance with history of UC of the urinary bladder.
Each surgical specimen was represented by 3–12, 4-µm-thick sections, stained with hematoxylin-phloxin-saffron. The urine cytology specimens were collected in Shandon cytospin (Shandon Lipshaw, Life Sciences International Ltd, Astmoor, Runcorn, Cheshire, England) and submitted for centrifugation at 3,000 rpm for 10 minutes. The obtained sediment specimens were then submitted for regular smear or cytospin. Urine specimens consisting only of umbrella cells or <10 nonumbrella urothelial cells per high-power field in the most cellular areas were excluded from the study. Immunostaining was performed on cytospins, cell block sections and destained slides from previous Papanicolaou-stained cytospin samples. The immunostaining was done on formalin-fixed tissue sections using the avidin-biotin-peroxidase complex (ABC) method with an ABC kit (Vector Laboratories, Burlingame, California, U.S.A.) for CK20 (dilution 1:100, Dako, Glostrup, Denmark). The slides were preheated by microwave at 25 psi, at 125°C for 5 minutes. Cytospin specimens were fixed with 75% alcohol and then processed without antigen retrieval. Papanicolaou-stained slides destained with xylol and then processed as cytospin slides. The negative controls for immunostaining were performed on cell block sections or unstained or destained cytospins from each case with omission of the CK20 antiserum. The positive controls were performed on tissue of colonic adenocarcinoma for each batch of immunostaining.
All cytologic slides were reviewed; the immunostaining was interpreted by KTM who was blinded regarding the final surgical or clinical diagnoses. The cases were characterized by 6 criteria: (a) branching stromal cores, (b) 3-dimensional (3-D) papillary cell clusters with features suggestive of branching stromal cores, (c) networks of dyshesive cells, (d) CK20-positive urothelial cells (nonumbrella), (e) hypercellularity with dyshesive cells and (f) cytologic atypia associated with CK20-negative reactivity. After correlation with the surgical specimens and/or clinical and imaging findings, the cases with LGUC or urothelial neoplasms of unknown malignant potential (UMP UN) (group 1) (World Health Organization/International Society of Urological Pathology classification of bladder cancer, introduced in 1998)33 were classified into subgroup 1a having at least 1 of 4 criteria: a, b, c and d, that are diagnostic for UC; subgroup 1b having CK20-negative atypical cells and hypercellularity, thought to be suspicious but not diagnostic of UC; and subgroup 1c that lacked significant cytologic features diagnostic of or suspicious for UC. For statistical analysis Sisa software (http: //www.quantitativeskills. com/sisa/) was used.
Results There were a total of 25 LGUC/UMP UN cases with histologic and cytologic correlation (group 1: 30 wash [12 specimens] or brush ureteral specimens [18 specimens]). The patients were 44–88 years old (mean, 57±12), and the male to female ratio was 2:1. Group 1a, 1b and 1c consisted of 22 specimens (18 patients), 3 specimens (3 patients) and 5 specimens (4 patients), respectively. The control group (group 2) included 7 specimens from 7 cases (Table I).
 The specimens from group 1a displayed hypercellularity in the cytospin in 8 cases. In addition, 6 cases consisted of monomorphic columnar-like cells with elongated or oval nuclei with minimal atypia (Figures 1–3). Hypercellularity was also present in the specimens from group 1b and in 1 and 2 cases from groups 1c and 2, respectively. In the case from group 2, this was associated with a large number of umbrella cells and a lack of abundant monomorphic columnar-like cells. Fibrous stroma forming delicate and branching tissue cores (2–4 tissue cores per case) wrapped by dyshesive cells were identified in 3 specimens (3 cases) (Figure 3). Of these 3 cases, 2 also contained multiple fragments of fibrous tissue in the cytospin. In 5 cases, there were networks of dyshesive cells with varying degrees of mucinous background. Within these networks, cell clusters with and without cell collars were often seen (Figure 2). 3-D papillary cell clusters were identified in 4 specimens (3 cases), and a few 3-D cell clusters contained structures suggestive of branching cores (Figure 4). Two cases also contained numerous cell clusters characterized by cohesive cells in the cytospin. In group 2 (controls), 3-D cell clusters were seen in 1 case, but these were fewer in number and not associated with a branching stromal core. With respect to cytologic atypia, this was barely discernible in 10 specimens from group 1 (Figure 5) and in 3 specimens from group 2.
     Immunostaining for CK20 was performed on 14 specimens, including all specimens with cytologic atypia and 1 specimen without atypia from group 1c (cell block: 5; unstained cytospin: 4; destained cytospin: 5). CK20 reactivity was negative in 11 specimens and was positive in 3 specimens (2 from the unstained cytospin and 1 from the destained cytospin) (Figure 5). Of these 3 CK20-reactive cases, 2 were not associated with criteria a, b and c.
Table II shows the statistical analysis for the sensitivity, specificity and positive and negative predictive value for criteria a, b, c, d and e in the diagnosis of LGUC/UMP UN. Criteria a, b, c and d were associated with the highest values of specificity but with low sensitivity. A combination of hard criteria a, b, c and d and soft criteria e plus f increases sensitivity and the negative predictivity.
 Discussion
UC of the urinary tract can be categorized into 2 main groups: LGUC and high-grade UC (HGUC).24 The differences in these 2 groups are based on the degree of architecture and cytopathologic atypia seen. LGUC and UMP UN are associated with mild to no cytologic atypia. UMP UN is distinguished from LGUC by the degrees of cytologic atypia. In LGUC, the mild nuclear variation is appreciable at the scanning magnification, with occasional mitotic figures seen. Architecturally, the UMP UN and LGUC form papillary structures and often display a thick layer of urothelium >7 cell layers. These tumors follow a favorable biologic evolution with low potential for stromal or lymphovascular invasion and uncommonly exhibit regional or distant metastases. Because of this minimal deviation from normal urothelial cells, LGUC and UMP UN are often underdiagnosed in urine cytology. Recently, some authors emphasized improvement in the rate of positive diagnosis of the UC of the urinary bladder using a combination of cytologic features and the results of immunostaining for CK20.21-26 Although LGUC and UMP UN of the UUT are cytologically and histopathologically similar to those in the urinary bladder, the brush and wash specimens of the ureter are different because of the yield of a larger amount of tumor tissue diluted in smaller amount of liquid. Therefore UC and UMP UN in cytologic ureteral specimens are more readily recognized because of the preserved papillary architecture and easier to use an additional immunohistochemical modality. Also, one might consider that such an improvement in diagnostic sensitivity using cytology is more critical for upper tract lesions than for urinary bladder lesions, because the latter, in general, can be readily identified and a biopsy performed with cystoscopy. In contrast, in the upper tract, LGUC/UMP UN is not readily visualized and biopsies not readily obtained, but it can be treated conservatively with ureteroscopy; improvement of the cytologic diagnosis of these lesions would be very helpful in the early management of the disease.
In this study, we compared the cytologic features of 25 cases of LGUC and UMP UN with 7 cases of brush and wash of the ureter in cases of inflammation and pyelonephrostomy.
LGUC and UMP UN could be distinguished from the group that was negative for tumor by using 6 criteria: hypercellularity, branching of stroma cores, networks of dyshesive cells, 3-D papillary cell clusters, cellular atypia with CK20-positive staining and cellular atypia with CK20-negative staining. The 4 cases of LGUC and UMP UN with false negative diagnosis were likely related to procedural problems resulting in a lack of sampling of the tumor. As determined when performing the pilot study, the presence of 1 of the criteria: (a) branching stromal cores, (b) 3-D papillary cell clusters with features suggestive of branching stromal cores, (c) networks of dyshesive cells and (d) CK20-positive urothelial cells (nonumbrella) is diagnostic for LGUC or UMP UN. It is very unlikely for this type of branching stromal core and CK20-reactive urothelial cells to be seen in nonneoplastic ureteral lesions. The mechanism of dyshesive urothelial cells forming the characteristic cell network in cytologic preparations is unknown. It may be due to the combination of cell adhesivity in the papillary formations, the abundance of mucin lining the papillae and the small amount of liquid used in the procedure. The 3-D papillary cell clusters (criterion d) were seen mainly in the neoplastic group of our study; however, occasional clusters without papillary stromal cores and fewer in number may be seen in the nonneoplastic group, probably because of the papillary hyperplasia. Hypercellularity (criterion e) is often associated with criteria a, b, c and d, and is very suggestive of a neoplastic lesion; however, it may be seen in nonneoplastic lesions. It is often very difficult to distinguished cytologic atypia (criterion f) in neoplastic lesions of low grade from reactive atypia. Regarding CK20 immunostaining, it is well known that a large number of LGUC and UMP UN are not reactive for this antiserum.24-26 Of interest, immunocytochemical staining and molecular assay for CK20 recently have been used to detect malignant urothelial cells in the urine and peripheral blood to detect the recurrence and for staging.24-32 Due to the absence of the 4 criteria (a, b, c and d) in the negative group (subgroups 1c and 2), they are considered as hard criteria for positive diagnosis for UC, whereas criterion e plus f (with negative CK20 reactivity) are considered soft criteria. Specimens associated with only 2 soft criteria (subgroup 1b) can be considered as diagnostic of UC. Due to the small number of cases in the study, statistical analysis of the 6 criteria for the differential diagnoses between neoplastic and nonneoplastic lesions was not performed. The specificity and positive predictive value of hard criteria for UC were high, ranging from 80% to 100%. The sensitivity and negative predictive value were low, ranging from 13% to 27%. A combination of the soft and hard criteria will increase the sensitivity, specificity and positive and negative predictive values to 83%, 86%, 96% and 45%, respectively.
In summary, LGUC and UMP UN of the UUT can be confidently diagnosed with wash or brush cytology because of the architecture of stromal and cell arrangements in urine cytologic specimens, but diagnosis can in some cases be enhanced by immunostaining with CK20. Study with larger series of cases is necessary to test the diagnostic value of the morphologic and immunologic features as suggested in this study.
References
From the Division of Anatomical Pathology, Department of Laboratory Medicine, The Ottawa Hospital, and Department of Pathology and Laboratory Medicine, University of Ottawa, Ottawa, Ontario, Canada; and Department of Pathology, New York Medical College, New York, U.S.A.
Dr. Mai is Associate Professor, Division of Anatomical Pathology, Department of Laboratory Medicine, The Ottawa Hospital, and Department of Pathology and Laboratory Medicine, University of Ottawa. Dr. Teo is Resident, Division of Anatomical Pathology, Department of Laboratory Medicine, The Ottawa Hospital, and Department of Pathology and Laboratory Medicine, University of Ottawa.
Drs. Robertson, Marginean and Islam are Assistant Professors, Division of Anatomical Pathology, Department of Laboratory Medicine, The Ottawa Hospital, and Department of Pathology and Laboratory Medicine, University of Ottawa. Dr. Yazdi is Professor, Department of Pathology, New York Medical College, and Senior Pathologist, CBLPath Laboratory, Rye Brook, New York.
Address correspondence to: Kien T. Mai, M.D., F.R.C.P.C., Division of Anatomical Pathology, Department of Laboratory Medicine, The Ottawa Hospital, General Campus, 501 Smyth Road, Ottawa, Ontario K1H 8L6, Canada (ktmai@ottawahospital.on.ca). Financial Disclosure: The authors have no connection to any companies or products mentioned in this article.
Received for publication February 23, 2008. Accepted for publication August 1, 2008.
 0001-5547/09/5306-0611/$21.00/0 © The International Academy of Cytology
Acta Cytologica Send website comments to Webmaster@acta-cytol.com |
|||||||||||