Progress in Pathology Current Status of Urinary Cytology in the Evaluation of Bladder Neoplasms WILLIAM M. MURPHY, MD The examination of urinary specimens for cancer cells is often a daunting experience which has gained few enthusiastic supporters while producing much apprehension. Several factors have contributed to suboptimal results. Urothelial cells require about a year to replicate so that few are available for examination in any, particular sample. Exfoliated cells enter a hostile environment of high acidity and low osmolality which may obscure essential diagnostic features. Standards for specimen collection, preservation, processing, and interpretation have not been widely. accepted, and the literature on this subject is limlted and often dated. Perhaps influenced by these factors, clinicians have submitted few specimens for interpretation, further limiting the experience and confidence of the pathologists charged with identifying neoplastic cells and determining their aggressive potential. Nevertheless, there has been a steadily increasing demand for the pathologic evaluation of urothelial cells in recent years. In part, the influx of urinary specimens reflects a general increase in requests for cytologic services engendered by requirements for cost control and risk management, More importantly, many urologists have recognized the limitations of cystoscopy and even random or selected site biopsies in monitoring patients with bladder neoplasms, and are seeking to broaden the scope of their follow-up procedures. Cytologic monitoring has become an essential procedure to identify residual or recurrent bladder neoplasms among patients treated with topical therapeutic agents and followed for prolonged periods prior to definitive surgery. The pathologist involved in caring for patients with bladder neoplasms must be prepared to evaluate urinary specimens on a regular basis. Among the available techniques, urinary cytology using the light microscope remains the most useful approach in everyday practice. Given adequate samples, almost all potentially aggressive bladder neoplasms, whether
From the Departments of Pathology and Urology, Baptist Memorial Hospital and University of Tennessee, Memphis. TN. Accepted for publication December 14, 1989. Key wor&: cytology, urine, bladder. neoplasia, urinary cytol“k!Y’ Address correspondence and reprint requests to William M. Murphy. MD, Department of Pathology, Baptist Memorial Hospital, 899 Madison Ave. Memphis, TN 38146. 0 1990 by W.B. Saunders Company. 0046-8 17719012 109-0003$5.00/0
886
cystoscopically visible or not, can be detected with this method. Urinary cvtology has been especially valuable for the dete&nation of the efficacv of treatment, the presence of residual microsco$c tumor, and/or the possible involvement of prostatic ducts, Brunn’s nests, and the upper collecting system. Studies have shown that urinary cytology is just as reliable as cystoscopy for determimng treatment failures and the likelihood of an adverse outcome.’ Accurate evaluation is not confounded by the type of therapy and can be performed on specimens obtained from intestinal conduits as well as from residual urethras. Even nonaggressive lesions such as low-grade papillary tumors and flat dysplasias can be detected using urinary cytology, although the sensitivity of the method is not as great as for high:grade malignancies.” This review is intended to -present a conceptual approach to urinary cytolo!g_y. offer guidelines for diagnostic interpretation of frequently encountered lesions, and discuss the most common sources of error. Much of this information is available from eclucational courses and references in the literature, but the pertinent facts are widely dispersed and of uneven quality. Not surprisingly, patholo@sts continue to succumb to certain well-known diagnostic pitfalls, such as the misinterpretation of reactive superficial cells as cancerous elements and the overinterpretation of papillary aggregation as an important feature of low-grade transitional cell neoplasia. NORMAL AND REACTIVE UROTHELIUM T-he human urinary bladder functions as a reservoir to store and expel urine. Its epithelial lining has little or no function other than to maintain the structural integrity of the organ, to prevent noxious substances from gaining access to the muscular wall and blood stream, and to expand and contract as required by the amount of the urine present.” This epithelium is structurally distinctive, being composed of a superficial layer of large cells which cover several layers of uniform, smaller cells like an umbrella, Prevlous anatomists have considered the epithelium to represent a structure transitional between squamous and glandular epithelium. although the currently preferred term is urothelium. The superficial cells can produce and secrete small amounts of mucin, although their primary function is to fix the underlying cells in place and expand or contract to accommodate filling and emp-
URINARY CYTOLOGY IN BLADDER NEOPLASIA [Murphy)
tying of the organ. Superficial cells have a rigid surface membrane which does not round up in a fluid environment so that these elements are readily identified in urinary samples-’ (Fig 1). Their abundant cytoplasm (primarily composed of infolded surface membranes) may appear columnar when compressed, for example, in a proliferation of Brunn’s nests. They often have two nuclei, represented in flow cytometr\; histograms as a tetraploid peak.” When reacting to noxious stimuli, superficial cells often manifest nuclear changes identical to those of high-grade neoplasia, hut the preponderance of int’ormation indicates that truly mali,gnant cells rarely if e\‘er differentiate toward superficial cells. l‘herefore. if- a cell in a urinary specimen can be identified as a superficial cell by its abundant cytoplasm and typical c.oncave and convex surfaces, it should not be interljreted as malignant regardless of the degree of anaplasia of its nucleus. One might take one further step and recommend as a practical maxim that no cell with ;I nuclear:cytoplasmic ratio of 1:2 or less should be considered malignant. The urothelial cells underlying the superficial cell layer are much smaller, uniform. and normally filled with gl!cogen which washes out during processing, leaving ;t cleared (“vacuolated”) cytoplasm surrounding an ovoid nucleus oriented with its long axis perpendicular to the basal lamina. These cells are arranged in layers, usually three to four, but probably not exceeding six. They are inactive and only replic,tte ever!. 200 to .3X) days.” In urinary specimens, normal mmsuperficial urothelial cells tend to cluster. If the specimen has been obtained by instrumentation, normal urothelial cells may be mechanically avulsed from Brunn’s nests or mucosal undulations and appear as “papillar);” aggregates. This is especiallv like]\ if the specimen is obtained from ;I ureter
or a trabeculated bladder. It must be emphasized that the cells comprising these “papillar)” aggregates are almost alwavs uniform with nuclei which lack features of neoplasii, and that a papillary configuration per se is an unreliable feature of neoplasia. Like superficial cells, other urothelial cells can react to noxious stimuli, but they rarely manifest bizarre nuclear changes. Instead, these cells tend to develop cytoplasmic vacuoles and uniformly enlarged nuclei with evenly dispersed chromatin and prominent nucleoli. Like normal elements, clustering and even papillary aggregation is common. Most neoplastic cells. whether benign or malignant, tend to differentiate toward these subsuperficial urothelial cells, and it is these elements, not the superficial/umbrel!a elements, to which comparisons of nuclear characteristics between neoplasia and normal are most appropriately made. SPECIMEN COLLECTION, PRESERVATION, AND PROCESSING Degeneration of urothelial cells begins prior to exfoliation and continues while the cells are in contact with urine prior to and after voiding, ibut the rate at which this occurs and the effect of degeneration on diagnostic evaluation have rarel) been studied.7,x Prolonged exposure of urothelial cells to urine has detrimental effects on pathologic interpretation and specimens should not be submitted from the first morning \,oiding, 24-hour collections, or drainage bags. Random-voided urines, with or without prior hydration, are most useful as a routine procedure. but catheterized specimens and bladder washings are \velcome provided that the patient is to undergo instrumentation for some other purpose. While bladder washings tend to Field more and better-preserved
FIGURE 1. Normal cells in a urinary specimen. Note the great disparity in size between the superficial cells (arrows] and the subsuperficial elements. Also note the retention of convex and concave surfaces in the superfictal cells. (Eosin azure 50 [EA 501 stain; magnification X 768.)
887
HUMAN PATHOLOGY
Volume 21, No. 9 [September 1990)
cells than voided urines, diagnostic elements may occur only in fresh-voided samples, and the degenerative changes in urothelial cells exposed to urine for a few hours are not sufficient to confound the diagnostic interpretation of neoplasia.” Therefore, no patient should be instrumented solely to obtain a urinary specimen for light-microscopic diagnosis. Urinary specimens may be preserved in alcohols or salt solutions to reduce bacterial overgrowth and to inhibit further cellular degeneration, but for most purposes the same results can be obtained by refrigeration alone. Needless to say, specimens should not be exposed to freezing temperatures. Adequate diagnostic results can be achieved with samples kept cool up to 48 hours prior to processing, but the best yields require refrigeration of under 4 hours. Urothelial cells may be extracted from urine and resuspended in more balanced solutions for transportation to reference laboratories, but the accumulated experience with this procedure is small, and it might be wiser to train a technician to prepare specimens locally so that only the glass slides would be transported. Urinary samples can be processed by several techniques including direct smears, membrane filters, and cytocentrifugation. All have certain advantages and drawbacks.tO.” In general, direct smears are easiest to prepare but suffer from high cell loss and suboptimal display of cellular detail: membrane filters produce the best cytologic detail but are difficult to prepare; and the computer-programmed cytocentrifuge offers a compromise in cytologic yield and detail which has wide appeal. Regardless of the method used, urinary specimens should be stained using the Papanicolaou technique. Romanovsky dyes on airdried material fail to highlight important cytologic details. CYTOHISTOLOGIC CORRELATION AND NOMENCLATURE The biologic factors which determine normal epithelial cell differentiation in the bladder dictate that almost all urothelial cells will be transitional in morphology, and it is not surprising that 75% to 90% of bladder neoplasms can be categorized as transitional cell. In fact, a case might be made that all nontransitional cell neoplasms represent a failure of host environmental influences on differentiation, and are therefore much more abnormal than their degree of differentiation toward squamous, glandular, or other tissues might otherwise suggest. When the cells of transitional cell neoplasms appear in urinary specimens, they cannot be reliably recognized as three to five distinctive degrees of differentiation for direct correlation with various histologic grading schemes. Several factors account for difficulties in cytohistologic correlation: 1. In the absence of a tissue pattern which includes a delicate fibrovascular stalk, many papillary transitional cell tumors are composed of cells which differ so little from normal that they can hardly be
recognized as neoplastic. Depending upon the grading system used, these cells comprise papillomas and/ or transitional cell carcinomas, grades 0, 1, or even 2. 2. Intermediate grades of transitional cell neoplasia are usually determined not by the features of their component cells but by the frequency and position of mitotic figures, the absence of a superficial cell layer overlying the tumor, the number of cell layers, the relative degree of nuclear pleomorphism within the tumor, and/or the relative density of the nuclei comprising the tumor. None of these features can be reliably identified in cytologic preparations. 3. The cells of transitional cell neoplasms tend to differentiate within the tumor, a characteristic best visualized in low-grade lesions. Cells at the surface, those most likely to appear in urinary samples, may be more mature than those at the base, and any single specimen may not be representative of the differentiation of the entire tumor. 4. In routine preparations, cells from adenocarcinemas cannot always be distinguished from those of high-grade transitional cell carcinomas or the nonkeratinizing portions of squamous cell carcinomas. 5. Nonkeratinizing squamous carcinoma cells are not significantly different cytologically from highgrade transitional carcinoma cells. nor can the cells of carcinoma in situ ordinarily be distinguished from those of invasive high-grade transitional cell carcinoma. 6. Degeneration is common in cytologic samples, and may alter the appearance of neoplastic cells so that they seem to be of a lower or higher grade than is actually present. Recognizing these factors, we have reported the cytologic findings in urinary specimens using the nomenclature listed in Table 1. In general, only 30% to 60% of histologic lesions graded as papilloma (our classification) or transitional cell carcinoma, grade 1 (World Health Organization [WHO] classification) will exfoliate cells with sufficient cytologic abnormalities to be recognized as neoplastic, and even these cases will manifest subtle cellular details requiring a good deal of expertise for accurate identification.‘.‘2 Fortunately, these transitional cell tumors act like they look and are rarely, if ever, aggressive. Depending upon the histologic definitions applied, most transitional cell tumors of intermediate and high histoTABLE 1. Diagnostic Terminology for Urinary Cytology Terminology Negative cytology (includes reactive cells) Atypical cells, significance uncertain Dysplastic cells Abnormal cells, suspicious for malignancy Neoplastic cells present: Low-grade neoplasm High-grade neoplasm Squamous carcinoma Adenocarcinoma Undifferentiated neoplasm Non-epithelial neoplasm
URINARY CYTOLOGY IN BLADDER NEOPLASIA [Murphy]
logic grade (transitional cell carcinoma low-grade and high-grade in our system and transitional cell carcinoma, grades :’ and 3 in the WHO system) will comprise cells with anaplastic features of sufficient degree to be readily recognized as malignant by pathologists with even moderate experience. It is these lesions which are potentially lethal and are most easily evaluated c~tologica]]v,“.“‘.‘~,‘:i
CELLUlAR FEATURES OF UROTHELIAL NEOPIASMS ‘The cellular features of urothelial neoplasms are listed in ‘Fable 2 and illustrated in Figs 2 through 5.rr Our experience indicates that the most important features distinguishing high-grade neoplasms for practical diagnostic interpretation are: (1) eccentric nuclei; (2) pleomorphism; (3) clumped, irregularly distributed chromatin; and (4) large nucleoli in some but not all of the cells in any given area or aggregate. High-grade tumor cells may have cytoplasmic vacuoles. Specimens from untreated carcmoma in situ often have thousands of high-grade malignant cells. In contrast, the most important diagnostic features for low-gradie neoplasia are: (1) enlarged nuclei with extreme eccentricity; (2) irregular nuclear borders which appear as shallow depressions, notches, or creases depending upon the orientation of the cells; and (3) evenly distributed chromatin which is more granular than that of normal nonsuperficial urotheha1 cells. Neither large nucleoli nor cytoplasmic vacuoles are features of low-grade neoplasms. The presence of mumerous cells in a freshly ,voided random urine sample is unusual (normal urme contains approximate& 10 cells/ml) and should raise the suspicion of‘ a low.-grade lesion. Papillary aggregation occ m-s in cytologic samples from low-grade tumors, but this configuration is unreliable as the sole criterion of neoplasia. Squamous carcinoma can often be accurately identified from the cells in a urine sample. Despite the relatively poor prognosis of squamous cell tumors of the bladder, they are usually well-differentiated TABLE
2.
Cellular
Features
Pap~llan dnd loose ~lutrrs Inc~~cdsed. unitor-m
SI76 Kumbe1 (.vtoplaam h uclcar:~~to~~l,lslrli~ G3,LO \ ucki ?o\itlun SW hlorpholc~g~ IhNftT\ Chrum;mn ~uctcol1 ?,lodified
and exfoliate fairly mature cells with1 an elongated, spindled, or fusiform configuration. Cytoplasmic keratin can often be distinguished. Nuclear degeneration is common, but this is sometimes helpful in that it accentuates chromatin clumping and the irregularities of nuclear borders. Urine may occasionally contain unsuspected malignant squamous cells from the vagina or uterine cervix. Primary, adenocarcinomas are rare in the bladder, but do not differ cytologically from those in the rectum. Occasional tumors are so well-differentiated that their cells cannot be identified as rteoplastic, but the appearance of a high-grade neoplasm is the rule. When recognized as such, adenocarcinoma cells tend to cluster and have poorly stained. lucent, vacuolated II:
of Urothelial Neoplasia
I.,,%(;I-adc c ells .Arrangrmcllt
FIGURE 2. Low-grade neoplastic cells in a urinary specimen. The patient had a transitional cell papilloma (grade I carcinoma in the WHO system), (EA 50 stain; magnification * 768.1
(~)trcn II”mcroIIs HOWgf3IKNIS Increased
Eccvn,r1c E:11l;irgtd Vanablc within aggregate7 II-rrgular (notrhea. Cl eascsl Finr. cvetl Smalliat~sent
Higt1-GKXtC2
Isolated and loose clusters Increazed. pleomoi-phic V;r1able \‘ariable Inrreascd
E:~CeINl-lL V,&ble V‘llT3bl~ Ir1-ryul‘ir
FIGURE 3. High-grade neoplastic cells. The patient had a highgrade transitional cell carcinoma [grade III in the WHO system). Note the nuclear pleomorphism and prominence of nucleoli compared to the low-grade neoplasm in Fig 2. (EA 50 stain; magnification x 768.)
(:oarse. UlWYCll Variable
w itll ~~cn~~i~sion ’
889
HUMAN PATHOLOGY
Volume 21, No. 9 (September 1990)
FIGURE 4. Cells from a squamous carcinoma. The cells are very well-differentiated and lack many anaplastic features. (EA 50 stain; magnification x 768.)
c
e *
than the urothelial cells themselves which identifies such a lesion as a neoplasm. The cellular features of these low-grade papillary tumors are very similar to those of flat urothelial dysplasia (atypia, atypical hyperplasia) and it would not be inappropriate to adopt a cautious approach to diagnosis of these cells by labeling them dysplastic rather than neoplastic. One might argue that while the bladder tumor may be a neoplasm in the sense of a new growth. the cells of which it is composed have cytologic features more akin to dysplasia.
cytoplasm. Chromatin tends to aggregate along smooth nuclear borders and manifests so-called “clear” areas. Nucleoli are prominent. Using these criteria, almost all true transitional cell carcinomas (WHO grades 2 and 3), carcinoma in situ, squamous carcinomas, and adenocarcinomas of the urinary bladder can be detected using urinary cytology. Many urologists have recognized the value of routine monitoring of the urine for high-grade tumors and recommend less frequent follow-up with cystoscopy, especially for patients with nonaggressive lesions such as papilloma/transitional cell carcinoma, of lowgrade 1. I4 Given the cytologic composition grade tumors, it is not surprising that urinary cytology is less sensitive for their detection. In fact, in many cases it is the delicate fibrovascular stalk rather
PITFALLS AND CONFOUNDING FACTORS The most commonly encountered the diagnostic interpretation of urinary
FIGURE 5. Cells from an adenocarcinoma. Although it is easy to identify these cells as malignant, correct interpretation of histologic type is more difficult. In this case, note the scanty vacuolated cytoplasm and the nuclear “clearing”. Large nucleoli are not readily discerned in this photograph. (EA 50 stain; magnification x 768.)
890
difficulties in cytology are:
URINARY CYTOLOGY IN BLADDER NEOPLASIA (Murphy)
FIGURE 6. Abnormal cell not diagnostic of carcinoma. Although the nuclei have features of anaplasia, the low nucleaccytoplasmic ratio and the rigidity of the cell border indicate that this is a reactive superficial cell. Cells such as this are unreliable for the diagnosis of bladder cancer. (EA 50 stain; magnificcrtlon x 768.)
( 1) overjnterpreting
cells with low nuclear:cytoplascancer, (2) mistaking papillary aggregates as a reliable sign of low-grade neoplasia, (3) confusing the reactive/regenerative/reparative cells associated with urinary stones with neoplastic elements, (4) overinterpreting samples from the upper collecting system, (3) misinterpreting cells in ileal conduits, and (6) misunderstanding the cytologic effects of drugs and x-ray therapy. Most cells with nuclear:cytoplasmic ratios of 1:2 or less are superficial and should not be interpreted as malignant regardless of their nuclear anaplasia (Fig 6). While truly malignant cells may be large with relatively abundant cytoplasm, these elements cannot be reliabl) identified in urinary samples using light As a practical matter, if a specimen conmicroscopy, tains only these cells, it is almost always benign. While low-grade transitional cell neoplasms commonly shed aggregates of cells into urinary samples, the cells comprising the aggregates have the atypical nuclear features previously described. Papillary aggregates composed of normal-appearing transitional cells miglht come from a neoplasm but most likely emanate from the crypts of a trabeculated bladder or result from the mechanical avulsion caused by a catheter (Fig 7). Evaluation of cells in loose clusters rather than in tightlv adherent papillations is most likely to yield accurate results. Patients with urinary stones are often young and suffer the sudden onset of characteristically excruciating pain which leaves little doubt about the diagnosis. In some cases, however, urinary stones may form in such a way as to cause no symptoms other than intermittent hematuria. When this occurs, cytologic specimen:; may contain reactive/regenerative/reparative cells which are often arranged in papillary aggrelnic
ratios
gates. Diagnostic errors can usually be avoided if the pathologist places little reliance on the configuration of the aggregate and realizes that neither cytoplasmic vacuolization nor prominent nucleoli are features of low-grade transitional cell neoplasms (Fig 8). Reactive/regenerative/reparative elements can be differentiated from high-grade transitional cell carcinomas (which may be vacuolated and have large nucleoli) by their relatively uniform nuclear size, lack of significant irregularities of the nuclear borders. ten-
as
0
FIGURE 7. Papillary aggregate from a catheterized specimen. Note the uniformity of nuclear size and the lack of anaplastic features. Aggregates such as this are common after instrumentation and should not be overinterpreted. (EA 50 stain; magnification x 667.)
891
HUMAN PATHOLOGY
Volume 21, No. 9 (September 1990)
FIGURE 8. Papillary aggregate with reactive/regenerative/ reparative features from a patient with a urinary stone. Note the cytoplasmic vacuolization, uniformity of nuclear size, and the presence of prominent nucleoli in almost every nucleus. (IA 50 stain; magnification x 731.)
dency for peripheral concentration of chromatin, and uniform distribution of chromatin. While urinary cytology has been successful in identifying neoplasms of the bladder, it has been less reliable for recognition of renal pelvic and ureteral tumors. Random-voided urine from patients with these lesions ordinarily contains few diagnostic cells. Even when the urinary specimen represents a washing from the renal pelvis or ureter, diagnostic confusion may occur. Specimens from the upper collecting system have usually been obtained by catheterization of a bleeding urothelium. The urothelial cells of the upper collecting system are normally larger than
FIGURE 9. Clusters of relatively normal urothelial cells from a renal pelvic washing. In contrast to the cells of a lowgrade neoplasm, these elements are more uniform and lack significant nuclear atypicalities. (EA 50 stain; magnification x 768.)
892
those of the bladder and normally tend to aggregate (Fig 9). Nucleoli may be prominent in ceils of the upper collecting system, secondary to reactive changes associated with previous procedures as well as the underlying process. Clinical information is often misleading in that filling defects on intravenous pyelography are unreliable as an indication of neoplasia. Cytologic interpretation of ileal conduits (a form of “bladder” if you will) is most often confounded by failure of-the clinician to properly identify the These samples are often labeled “urine”, specimen.‘” a statement which is true enough but misleading in that almost all of the cellular elements represent degenerated intestinal mucosa (Fig 10). Ileal conduit specimens are all too frequently obtained from the storage bag and contain badly degenerated cells. Even properly collected specimens from patients with ureteral or renal pelvic neoplasms often contain only a few poorly preserved cancer cells (Fig 10). Intestinal conduit specimens are relatively rich in mucus, which may clog membrane filters so that direct smears and centrifuged specimens are usually essential for adequate evaluation. The important literature on the cytologic changes associated with therapeutic agents, whether administered systemically or topically. can be summarized very briefly: (1) therapeutic agents do not cause changes in urothelial cells which are specific for the drugs themselves, and (2) therapeutic agents rarely if ever cause changes in urothelial cells which are indistinguishable from cancer to experienced pathologists. Among the myriad drugs to which the urinary bladder may be exposed, only a few have been a source of diagnostic confusion in urinary samples, and most of these are alkylating agents such as cyclophosphamide (Cytoxan, Mead Johnson 8c Co, Evansville, IN), mitomycin C, and thio-TEPA.16 Cyclophosphamide is given systemically and metabolized to its active form.
URINARY CYTOLOGY IN BLADDER NEOPLASIA (Murphy)
FIGURE 10. cop) Clusters of degenerated intestinal cells from an ileal conduit. Since many ileal conduit specimens are labeled “urine”, the unwary might suspect low-grade neoplasia because of the papillary aggregation. Compare neoplastic urothelial cells (bottom] from the same ileal conduit specimen. (EA 50 stain; magnification x 480.)
form for the prevention or treatment o-f bladder neoplasms. The drugs remain in contact with urothelium for limited periods prior to voiding. Their cytologic manifestations are confined almost exclusively to superficial cells, where they cause enlargement of both cytoplasm and nuclei resulting in a characteristic but bizarre cell (Fig 12). These topical agents cause significant urothelial denudation, thus reducing the population of both normal and neoplastic cells in urinary samples. As a practical matter, no adjustments in diagnostic criteria for the recognition of neoplasia are necessary when examining specimens after topical chemotherapy. Cellular changes associated with bacillus Calmette-Gukrin vaccine. an increasingly popular therapeutic agent for bladder carcinoma, have not been documented, and treatment with this agent
The activ’e agent is concentrated in the urine and remains in contact with urothelial cells for relatively prolonged periods until voided. In poorly hydrated patients and certain particularly sensitive individuals, characteristic cells with increased nuclear:cytoplasmic ratios and degenerated nuclei with a “smudged” appearance have been observed (Fig 11). The nature of these elements is obscure but they are apparently not malignanl. They are, however, an important signal of a disturbed urothelium, and their presence should be recorded so that appropriate measures can be taken. At least 42 well-documented cases of bladder cancer have occurred after treatment with cyclophosphamide.17 Mitornycin C and thio-TEPA are topically applied alkylated agents administered in their active 893
HUMAN PATHOLOGY
Volume 21, No. 9 (September 1990)
FIGURE II. Cells characteristic of cyclophosphamide effect. Although not diagnostic, the presence of cells with relatively low nuclearcytoplasmic ratios and degenerated “smudged” chromatin is characteristic of treatment with cyclophosphamide. (EA 50 stain; magnification x 1,109.)
ties, but the changes are nonspecific and should not confuse the diagnosis of cancer. Since degenerative changes are common after x-ray therapy, the cytopathologist is often asked to vouch for the “viability” of any tumor cells identified. This is, of course, a fruitless exercise, if not an inappropriate question when applied to isolated cells in a urinary specimen. Most likely, x-rays affect urothelial neoplasms primarily by altering their blood supply through degenerative changes produced in endothelial cells.
should not confound the diagnosis of neoplasia in urinary specimens. Radiation has long been held responsible for certain cytologic changes, especially multinucleation, cytoplasmic eosinophilia and vacuolization, and nuclear pyknosis, but the effects of x-ray therapy on urothelial cells have not been rigorously tested in vivo and reliable examples of “radiation effect” are very difficult to ftnd.‘“~1”i~18 Like alkylating agents, it is probable that x-rays can cause bizarre nuclear abnormali-
FIGURE 12. Superficial cell with bizarre nucleus characteristic of treatment with either thio-TEPA or mitomycin C. Compare the size of this cell with the adjacent binucleated, relativety normal superficial cell. (EA 50 stain; magnification x 480.)
894
URINARY CYTOLOGY IN BLADDER NEOPLASIA (Murphy]
ADJUNCTNE APPROACHES TO URINARY SPECIMENS
nary cytology can detect ahnost all high-grade urotheliaf neoplasms. 3. The cytologic interpretation of low-grade transit.ional cell neopfasia requires expertise. These cells lack many of’ the features of‘ malignancy, a source of’ conf‘usion for the diagnostician but a positive factor for the patient since neoplasms composed of’ these cells are almost never aggressive. 5. The most usef‘ul type of urinarv specimen for routine diagnostic interpretation is :creshh voided, randomly collected urine. Catheterized sj,ecimens and bladder washings may yield more and better preserved cells, but no patient shoultf be catheterized solely to obtain diagnostic material. 6. Preservation of urinary specimlcns in alcohols is not necessary unless prolonged storage is contemplated. Refrigeration to prevent bacterial growth and inhibit further cellular degeneration is required, howe\,er. 7. Cytologic details are best displayed with membrane filtration but other types of processing are adequate. ‘l‘he computer-programmed ( ytocentrifuge is currently most popular. 8. Optimal recognition of’ c.).tologic details requires some fi)rm of’ P apanicolaou staining; Romanovsky dyes are fess desirable. 9. Urothelial cells with nucfear:c) toplasmic ratios of’ 1:fl or fess should not be interpreted as malignant regardless of the degree of anapfasia of their nuclei. IO. Papillary aggregation is not a I,eliabfe feature of’ low-grade neoplasia in urinary sam.ples. 1 1. Using appropriate criteria, the differential diagnosis of urothelial neopfasia versus the reactive/ regenerative/reparative changes seconcfary to urinal.) stones can almost always be accomplished. 12. Alkvlatinq aients such as Cvtoxan, thio‘I‘EPA. and mitomycin C produce cha;.acteristic bitt nonspecific changes in urotheliaf cells. These changes rarely mimic those of carcinoma. ‘l‘he diagnosis of’ urothelial neoplasia need not be conf’ounded by pre\,ious treatment. 13. Flolz cytometry and digitized image analysis are current]! used for diagnostic interpretations of urinary specnnens in selected centers. ‘I‘heir routine use must await further refinements in instrumentation and the formulation of more se.arching questions.
The most promising adjunctive approaches to the evaluation of urinary specimens are flow cytometry and digitized image analysis.5~1!‘.S” Ar present, both techniques are being used for diagnostic purposes in reference laboratories as well as at selected institutions, but neither has been widely accepted in routine Ipractice. Both techniques are capable of multiple functions. some relating to direct measurements and others to computations. Both ha1.e been useful in practical situations in inverse proportion to the diagnostic yield of’ fight microscopic analysis at any particular irlstitlition.~‘.~ Although both flow cytometr) and image analysis can be perf’ormed on specimens also exarninecf fly light microscopy, most studies have declined to use this approach, so that full comparisons am’ong methods have not been possible. The most common task to which these complex instruments have been dedicated has been the quantitation of‘ nucle,!tr DNA. The flow cytometer can examine nuclei faster than the image analyzer, but the pathofogist cannot actually see the cells being evaluated and must rell; on computer-generated histograms which fle has n& procfuced himself. The image analyzer is interactive. ie, the cells being analyzed can also fle seen and selected b\. the pathoioglst. fmt the preselection itself ma): p&entially introduce interpretive IGas and I:he procedure is, at least to the modern corn-puter operator. painfully slow. In our,judgment. neither method is capable of eff’ectively screening a “normal” population for flladder cancer, despite adI ertisements to the contrary. Until these acljunctive methods can be f‘urther refined, eg, capable of measuring tumor’ cell kinetics, of identifying significant abnormalities beyond the limits oflight microscopy, OI of evaluating the characteristics of tumor cells which have not been isolated from one another, their routine diagnostic use for urinary specimens will remain limited.
SUMMARY Pathologic examination of urinary specimens is increasingly recognized as an essential component of detection and monitoring for patients with bladder neoplasms. Among the available techniques, urinal-.! cvtology i:< the most usef’ul. The current status of WInarv cvtolog’. can he summarized as follows: ,
REFERENCES
1. The demand for urinary cytology is steadily, ilrcreasing as clinicians have realized the limitations of c!zstoscopy and even biopsy for monitoring bladder cancer patients, especially those havina carcinoma in situ or receiving topical therapy. 2. Urinary cvtology is currently an essential procedure for monitoring all patients with urothelial neoplasms and, if‘ consistently used, can actually decrease the frequency with which patients need to be subjected to cvstoscopy. 3. Even i’n moderately experienred hands, uri-
I. (:nnr
,JD, Murphy
M’M,
‘Llh: t’rt)gnostlc signlfi;iiiel- intravesical mitom)cin C for superficial bladder canter.
From the Departments of Pathology and Urology, Baptist Memorial Hospital and University of Tennessee, Memphis. TN. Accepted for publication December 14, 1989. Key wor&: cytology, urine, bladder. neoplasia, urinary cytol“k!Y’ Address correspondence and reprint requests to William M. Murphy. MD, Department of Pathology, Baptist Memorial Hospital, 899 Madison Ave. Memphis, TN 38146. 0 1990 by W.B. Saunders Company. 0046-8 17719012 109-0003$5.00/0
886
cystoscopically visible or not, can be detected with this method. Urinary cvtology has been especially valuable for the dete&nation of the efficacv of treatment, the presence of residual microsco$c tumor, and/or the possible involvement of prostatic ducts, Brunn’s nests, and the upper collecting system. Studies have shown that urinary cytology is just as reliable as cystoscopy for determimng treatment failures and the likelihood of an adverse outcome.’ Accurate evaluation is not confounded by the type of therapy and can be performed on specimens obtained from intestinal conduits as well as from residual urethras. Even nonaggressive lesions such as low-grade papillary tumors and flat dysplasias can be detected using urinary cytology, although the sensitivity of the method is not as great as for high:grade malignancies.” This review is intended to -present a conceptual approach to urinary cytolo!g_y. offer guidelines for diagnostic interpretation of frequently encountered lesions, and discuss the most common sources of error. Much of this information is available from eclucational courses and references in the literature, but the pertinent facts are widely dispersed and of uneven quality. Not surprisingly, patholo@sts continue to succumb to certain well-known diagnostic pitfalls, such as the misinterpretation of reactive superficial cells as cancerous elements and the overinterpretation of papillary aggregation as an important feature of low-grade transitional cell neoplasia. NORMAL AND REACTIVE UROTHELIUM T-he human urinary bladder functions as a reservoir to store and expel urine. Its epithelial lining has little or no function other than to maintain the structural integrity of the organ, to prevent noxious substances from gaining access to the muscular wall and blood stream, and to expand and contract as required by the amount of the urine present.” This epithelium is structurally distinctive, being composed of a superficial layer of large cells which cover several layers of uniform, smaller cells like an umbrella, Prevlous anatomists have considered the epithelium to represent a structure transitional between squamous and glandular epithelium. although the currently preferred term is urothelium. The superficial cells can produce and secrete small amounts of mucin, although their primary function is to fix the underlying cells in place and expand or contract to accommodate filling and emp-
URINARY CYTOLOGY IN BLADDER NEOPLASIA [Murphy)
tying of the organ. Superficial cells have a rigid surface membrane which does not round up in a fluid environment so that these elements are readily identified in urinary samples-’ (Fig 1). Their abundant cytoplasm (primarily composed of infolded surface membranes) may appear columnar when compressed, for example, in a proliferation of Brunn’s nests. They often have two nuclei, represented in flow cytometr\; histograms as a tetraploid peak.” When reacting to noxious stimuli, superficial cells often manifest nuclear changes identical to those of high-grade neoplasia, hut the preponderance of int’ormation indicates that truly mali,gnant cells rarely if e\‘er differentiate toward superficial cells. l‘herefore. if- a cell in a urinary specimen can be identified as a superficial cell by its abundant cytoplasm and typical c.oncave and convex surfaces, it should not be interljreted as malignant regardless of the degree of anaplasia of its nucleus. One might take one further step and recommend as a practical maxim that no cell with ;I nuclear:cytoplasmic ratio of 1:2 or less should be considered malignant. The urothelial cells underlying the superficial cell layer are much smaller, uniform. and normally filled with gl!cogen which washes out during processing, leaving ;t cleared (“vacuolated”) cytoplasm surrounding an ovoid nucleus oriented with its long axis perpendicular to the basal lamina. These cells are arranged in layers, usually three to four, but probably not exceeding six. They are inactive and only replic,tte ever!. 200 to .3X) days.” In urinary specimens, normal mmsuperficial urothelial cells tend to cluster. If the specimen has been obtained by instrumentation, normal urothelial cells may be mechanically avulsed from Brunn’s nests or mucosal undulations and appear as “papillar);” aggregates. This is especiallv like]\ if the specimen is obtained from ;I ureter
or a trabeculated bladder. It must be emphasized that the cells comprising these “papillar)” aggregates are almost alwavs uniform with nuclei which lack features of neoplasii, and that a papillary configuration per se is an unreliable feature of neoplasia. Like superficial cells, other urothelial cells can react to noxious stimuli, but they rarely manifest bizarre nuclear changes. Instead, these cells tend to develop cytoplasmic vacuoles and uniformly enlarged nuclei with evenly dispersed chromatin and prominent nucleoli. Like normal elements, clustering and even papillary aggregation is common. Most neoplastic cells. whether benign or malignant, tend to differentiate toward these subsuperficial urothelial cells, and it is these elements, not the superficial/umbrel!a elements, to which comparisons of nuclear characteristics between neoplasia and normal are most appropriately made. SPECIMEN COLLECTION, PRESERVATION, AND PROCESSING Degeneration of urothelial cells begins prior to exfoliation and continues while the cells are in contact with urine prior to and after voiding, ibut the rate at which this occurs and the effect of degeneration on diagnostic evaluation have rarel) been studied.7,x Prolonged exposure of urothelial cells to urine has detrimental effects on pathologic interpretation and specimens should not be submitted from the first morning \,oiding, 24-hour collections, or drainage bags. Random-voided urines, with or without prior hydration, are most useful as a routine procedure. but catheterized specimens and bladder washings are \velcome provided that the patient is to undergo instrumentation for some other purpose. While bladder washings tend to Field more and better-preserved
FIGURE 1. Normal cells in a urinary specimen. Note the great disparity in size between the superficial cells (arrows] and the subsuperficial elements. Also note the retention of convex and concave surfaces in the superfictal cells. (Eosin azure 50 [EA 501 stain; magnification X 768.)
887
HUMAN PATHOLOGY
Volume 21, No. 9 [September 1990)
cells than voided urines, diagnostic elements may occur only in fresh-voided samples, and the degenerative changes in urothelial cells exposed to urine for a few hours are not sufficient to confound the diagnostic interpretation of neoplasia.” Therefore, no patient should be instrumented solely to obtain a urinary specimen for light-microscopic diagnosis. Urinary specimens may be preserved in alcohols or salt solutions to reduce bacterial overgrowth and to inhibit further cellular degeneration, but for most purposes the same results can be obtained by refrigeration alone. Needless to say, specimens should not be exposed to freezing temperatures. Adequate diagnostic results can be achieved with samples kept cool up to 48 hours prior to processing, but the best yields require refrigeration of under 4 hours. Urothelial cells may be extracted from urine and resuspended in more balanced solutions for transportation to reference laboratories, but the accumulated experience with this procedure is small, and it might be wiser to train a technician to prepare specimens locally so that only the glass slides would be transported. Urinary samples can be processed by several techniques including direct smears, membrane filters, and cytocentrifugation. All have certain advantages and drawbacks.tO.” In general, direct smears are easiest to prepare but suffer from high cell loss and suboptimal display of cellular detail: membrane filters produce the best cytologic detail but are difficult to prepare; and the computer-programmed cytocentrifuge offers a compromise in cytologic yield and detail which has wide appeal. Regardless of the method used, urinary specimens should be stained using the Papanicolaou technique. Romanovsky dyes on airdried material fail to highlight important cytologic details. CYTOHISTOLOGIC CORRELATION AND NOMENCLATURE The biologic factors which determine normal epithelial cell differentiation in the bladder dictate that almost all urothelial cells will be transitional in morphology, and it is not surprising that 75% to 90% of bladder neoplasms can be categorized as transitional cell. In fact, a case might be made that all nontransitional cell neoplasms represent a failure of host environmental influences on differentiation, and are therefore much more abnormal than their degree of differentiation toward squamous, glandular, or other tissues might otherwise suggest. When the cells of transitional cell neoplasms appear in urinary specimens, they cannot be reliably recognized as three to five distinctive degrees of differentiation for direct correlation with various histologic grading schemes. Several factors account for difficulties in cytohistologic correlation: 1. In the absence of a tissue pattern which includes a delicate fibrovascular stalk, many papillary transitional cell tumors are composed of cells which differ so little from normal that they can hardly be
recognized as neoplastic. Depending upon the grading system used, these cells comprise papillomas and/ or transitional cell carcinomas, grades 0, 1, or even 2. 2. Intermediate grades of transitional cell neoplasia are usually determined not by the features of their component cells but by the frequency and position of mitotic figures, the absence of a superficial cell layer overlying the tumor, the number of cell layers, the relative degree of nuclear pleomorphism within the tumor, and/or the relative density of the nuclei comprising the tumor. None of these features can be reliably identified in cytologic preparations. 3. The cells of transitional cell neoplasms tend to differentiate within the tumor, a characteristic best visualized in low-grade lesions. Cells at the surface, those most likely to appear in urinary samples, may be more mature than those at the base, and any single specimen may not be representative of the differentiation of the entire tumor. 4. In routine preparations, cells from adenocarcinemas cannot always be distinguished from those of high-grade transitional cell carcinomas or the nonkeratinizing portions of squamous cell carcinomas. 5. Nonkeratinizing squamous carcinoma cells are not significantly different cytologically from highgrade transitional carcinoma cells. nor can the cells of carcinoma in situ ordinarily be distinguished from those of invasive high-grade transitional cell carcinoma. 6. Degeneration is common in cytologic samples, and may alter the appearance of neoplastic cells so that they seem to be of a lower or higher grade than is actually present. Recognizing these factors, we have reported the cytologic findings in urinary specimens using the nomenclature listed in Table 1. In general, only 30% to 60% of histologic lesions graded as papilloma (our classification) or transitional cell carcinoma, grade 1 (World Health Organization [WHO] classification) will exfoliate cells with sufficient cytologic abnormalities to be recognized as neoplastic, and even these cases will manifest subtle cellular details requiring a good deal of expertise for accurate identification.‘.‘2 Fortunately, these transitional cell tumors act like they look and are rarely, if ever, aggressive. Depending upon the histologic definitions applied, most transitional cell tumors of intermediate and high histoTABLE 1. Diagnostic Terminology for Urinary Cytology Terminology Negative cytology (includes reactive cells) Atypical cells, significance uncertain Dysplastic cells Abnormal cells, suspicious for malignancy Neoplastic cells present: Low-grade neoplasm High-grade neoplasm Squamous carcinoma Adenocarcinoma Undifferentiated neoplasm Non-epithelial neoplasm
URINARY CYTOLOGY IN BLADDER NEOPLASIA [Murphy]
logic grade (transitional cell carcinoma low-grade and high-grade in our system and transitional cell carcinoma, grades :’ and 3 in the WHO system) will comprise cells with anaplastic features of sufficient degree to be readily recognized as malignant by pathologists with even moderate experience. It is these lesions which are potentially lethal and are most easily evaluated c~tologica]]v,“.“‘.‘~,‘:i
CELLUlAR FEATURES OF UROTHELIAL NEOPIASMS ‘The cellular features of urothelial neoplasms are listed in ‘Fable 2 and illustrated in Figs 2 through 5.rr Our experience indicates that the most important features distinguishing high-grade neoplasms for practical diagnostic interpretation are: (1) eccentric nuclei; (2) pleomorphism; (3) clumped, irregularly distributed chromatin; and (4) large nucleoli in some but not all of the cells in any given area or aggregate. High-grade tumor cells may have cytoplasmic vacuoles. Specimens from untreated carcmoma in situ often have thousands of high-grade malignant cells. In contrast, the most important diagnostic features for low-gradie neoplasia are: (1) enlarged nuclei with extreme eccentricity; (2) irregular nuclear borders which appear as shallow depressions, notches, or creases depending upon the orientation of the cells; and (3) evenly distributed chromatin which is more granular than that of normal nonsuperficial urotheha1 cells. Neither large nucleoli nor cytoplasmic vacuoles are features of low-grade neoplasms. The presence of mumerous cells in a freshly ,voided random urine sample is unusual (normal urme contains approximate& 10 cells/ml) and should raise the suspicion of‘ a low.-grade lesion. Papillary aggregation occ m-s in cytologic samples from low-grade tumors, but this configuration is unreliable as the sole criterion of neoplasia. Squamous carcinoma can often be accurately identified from the cells in a urine sample. Despite the relatively poor prognosis of squamous cell tumors of the bladder, they are usually well-differentiated TABLE
2.
Cellular
Features
Pap~llan dnd loose ~lutrrs Inc~~cdsed. unitor-m
SI76 Kumbe1 (.vtoplaam h uclcar:~~to~~l,lslrli~ G3,LO \ ucki ?o\itlun SW hlorpholc~g~ IhNftT\ Chrum;mn ~uctcol1 ?,lodified
and exfoliate fairly mature cells with1 an elongated, spindled, or fusiform configuration. Cytoplasmic keratin can often be distinguished. Nuclear degeneration is common, but this is sometimes helpful in that it accentuates chromatin clumping and the irregularities of nuclear borders. Urine may occasionally contain unsuspected malignant squamous cells from the vagina or uterine cervix. Primary, adenocarcinomas are rare in the bladder, but do not differ cytologically from those in the rectum. Occasional tumors are so well-differentiated that their cells cannot be identified as rteoplastic, but the appearance of a high-grade neoplasm is the rule. When recognized as such, adenocarcinoma cells tend to cluster and have poorly stained. lucent, vacuolated II:
of Urothelial Neoplasia
I.,,%(;I-adc c ells .Arrangrmcllt
FIGURE 2. Low-grade neoplastic cells in a urinary specimen. The patient had a transitional cell papilloma (grade I carcinoma in the WHO system), (EA 50 stain; magnification * 768.1
(~)trcn II”mcroIIs HOWgf3IKNIS Increased
Eccvn,r1c E:11l;irgtd Vanablc within aggregate7 II-rrgular (notrhea. Cl eascsl Finr. cvetl Smalliat~sent
Higt1-GKXtC2
Isolated and loose clusters Increazed. pleomoi-phic V;r1able \‘ariable Inrreascd
E:~CeINl-lL V,&ble V‘llT3bl~ Ir1-ryul‘ir
FIGURE 3. High-grade neoplastic cells. The patient had a highgrade transitional cell carcinoma [grade III in the WHO system). Note the nuclear pleomorphism and prominence of nucleoli compared to the low-grade neoplasm in Fig 2. (EA 50 stain; magnification x 768.)
(:oarse. UlWYCll Variable
w itll ~~cn~~i~sion ’
889
HUMAN PATHOLOGY
Volume 21, No. 9 (September 1990)
FIGURE 4. Cells from a squamous carcinoma. The cells are very well-differentiated and lack many anaplastic features. (EA 50 stain; magnification x 768.)
c
e *
than the urothelial cells themselves which identifies such a lesion as a neoplasm. The cellular features of these low-grade papillary tumors are very similar to those of flat urothelial dysplasia (atypia, atypical hyperplasia) and it would not be inappropriate to adopt a cautious approach to diagnosis of these cells by labeling them dysplastic rather than neoplastic. One might argue that while the bladder tumor may be a neoplasm in the sense of a new growth. the cells of which it is composed have cytologic features more akin to dysplasia.
cytoplasm. Chromatin tends to aggregate along smooth nuclear borders and manifests so-called “clear” areas. Nucleoli are prominent. Using these criteria, almost all true transitional cell carcinomas (WHO grades 2 and 3), carcinoma in situ, squamous carcinomas, and adenocarcinomas of the urinary bladder can be detected using urinary cytology. Many urologists have recognized the value of routine monitoring of the urine for high-grade tumors and recommend less frequent follow-up with cystoscopy, especially for patients with nonaggressive lesions such as papilloma/transitional cell carcinoma, of lowgrade 1. I4 Given the cytologic composition grade tumors, it is not surprising that urinary cytology is less sensitive for their detection. In fact, in many cases it is the delicate fibrovascular stalk rather
PITFALLS AND CONFOUNDING FACTORS The most commonly encountered the diagnostic interpretation of urinary
FIGURE 5. Cells from an adenocarcinoma. Although it is easy to identify these cells as malignant, correct interpretation of histologic type is more difficult. In this case, note the scanty vacuolated cytoplasm and the nuclear “clearing”. Large nucleoli are not readily discerned in this photograph. (EA 50 stain; magnification x 768.)
890
difficulties in cytology are:
URINARY CYTOLOGY IN BLADDER NEOPLASIA (Murphy)
FIGURE 6. Abnormal cell not diagnostic of carcinoma. Although the nuclei have features of anaplasia, the low nucleaccytoplasmic ratio and the rigidity of the cell border indicate that this is a reactive superficial cell. Cells such as this are unreliable for the diagnosis of bladder cancer. (EA 50 stain; magnificcrtlon x 768.)
( 1) overjnterpreting
cells with low nuclear:cytoplascancer, (2) mistaking papillary aggregates as a reliable sign of low-grade neoplasia, (3) confusing the reactive/regenerative/reparative cells associated with urinary stones with neoplastic elements, (4) overinterpreting samples from the upper collecting system, (3) misinterpreting cells in ileal conduits, and (6) misunderstanding the cytologic effects of drugs and x-ray therapy. Most cells with nuclear:cytoplasmic ratios of 1:2 or less are superficial and should not be interpreted as malignant regardless of their nuclear anaplasia (Fig 6). While truly malignant cells may be large with relatively abundant cytoplasm, these elements cannot be reliabl) identified in urinary samples using light As a practical matter, if a specimen conmicroscopy, tains only these cells, it is almost always benign. While low-grade transitional cell neoplasms commonly shed aggregates of cells into urinary samples, the cells comprising the aggregates have the atypical nuclear features previously described. Papillary aggregates composed of normal-appearing transitional cells miglht come from a neoplasm but most likely emanate from the crypts of a trabeculated bladder or result from the mechanical avulsion caused by a catheter (Fig 7). Evaluation of cells in loose clusters rather than in tightlv adherent papillations is most likely to yield accurate results. Patients with urinary stones are often young and suffer the sudden onset of characteristically excruciating pain which leaves little doubt about the diagnosis. In some cases, however, urinary stones may form in such a way as to cause no symptoms other than intermittent hematuria. When this occurs, cytologic specimen:; may contain reactive/regenerative/reparative cells which are often arranged in papillary aggrelnic
ratios
gates. Diagnostic errors can usually be avoided if the pathologist places little reliance on the configuration of the aggregate and realizes that neither cytoplasmic vacuolization nor prominent nucleoli are features of low-grade transitional cell neoplasms (Fig 8). Reactive/regenerative/reparative elements can be differentiated from high-grade transitional cell carcinomas (which may be vacuolated and have large nucleoli) by their relatively uniform nuclear size, lack of significant irregularities of the nuclear borders. ten-
as
0
FIGURE 7. Papillary aggregate from a catheterized specimen. Note the uniformity of nuclear size and the lack of anaplastic features. Aggregates such as this are common after instrumentation and should not be overinterpreted. (EA 50 stain; magnification x 667.)
891
HUMAN PATHOLOGY
Volume 21, No. 9 (September 1990)
FIGURE 8. Papillary aggregate with reactive/regenerative/ reparative features from a patient with a urinary stone. Note the cytoplasmic vacuolization, uniformity of nuclear size, and the presence of prominent nucleoli in almost every nucleus. (IA 50 stain; magnification x 731.)
dency for peripheral concentration of chromatin, and uniform distribution of chromatin. While urinary cytology has been successful in identifying neoplasms of the bladder, it has been less reliable for recognition of renal pelvic and ureteral tumors. Random-voided urine from patients with these lesions ordinarily contains few diagnostic cells. Even when the urinary specimen represents a washing from the renal pelvis or ureter, diagnostic confusion may occur. Specimens from the upper collecting system have usually been obtained by catheterization of a bleeding urothelium. The urothelial cells of the upper collecting system are normally larger than
FIGURE 9. Clusters of relatively normal urothelial cells from a renal pelvic washing. In contrast to the cells of a lowgrade neoplasm, these elements are more uniform and lack significant nuclear atypicalities. (EA 50 stain; magnification x 768.)
892
those of the bladder and normally tend to aggregate (Fig 9). Nucleoli may be prominent in ceils of the upper collecting system, secondary to reactive changes associated with previous procedures as well as the underlying process. Clinical information is often misleading in that filling defects on intravenous pyelography are unreliable as an indication of neoplasia. Cytologic interpretation of ileal conduits (a form of “bladder” if you will) is most often confounded by failure of-the clinician to properly identify the These samples are often labeled “urine”, specimen.‘” a statement which is true enough but misleading in that almost all of the cellular elements represent degenerated intestinal mucosa (Fig 10). Ileal conduit specimens are all too frequently obtained from the storage bag and contain badly degenerated cells. Even properly collected specimens from patients with ureteral or renal pelvic neoplasms often contain only a few poorly preserved cancer cells (Fig 10). Intestinal conduit specimens are relatively rich in mucus, which may clog membrane filters so that direct smears and centrifuged specimens are usually essential for adequate evaluation. The important literature on the cytologic changes associated with therapeutic agents, whether administered systemically or topically. can be summarized very briefly: (1) therapeutic agents do not cause changes in urothelial cells which are specific for the drugs themselves, and (2) therapeutic agents rarely if ever cause changes in urothelial cells which are indistinguishable from cancer to experienced pathologists. Among the myriad drugs to which the urinary bladder may be exposed, only a few have been a source of diagnostic confusion in urinary samples, and most of these are alkylating agents such as cyclophosphamide (Cytoxan, Mead Johnson 8c Co, Evansville, IN), mitomycin C, and thio-TEPA.16 Cyclophosphamide is given systemically and metabolized to its active form.
URINARY CYTOLOGY IN BLADDER NEOPLASIA (Murphy)
FIGURE 10. cop) Clusters of degenerated intestinal cells from an ileal conduit. Since many ileal conduit specimens are labeled “urine”, the unwary might suspect low-grade neoplasia because of the papillary aggregation. Compare neoplastic urothelial cells (bottom] from the same ileal conduit specimen. (EA 50 stain; magnification x 480.)
form for the prevention or treatment o-f bladder neoplasms. The drugs remain in contact with urothelium for limited periods prior to voiding. Their cytologic manifestations are confined almost exclusively to superficial cells, where they cause enlargement of both cytoplasm and nuclei resulting in a characteristic but bizarre cell (Fig 12). These topical agents cause significant urothelial denudation, thus reducing the population of both normal and neoplastic cells in urinary samples. As a practical matter, no adjustments in diagnostic criteria for the recognition of neoplasia are necessary when examining specimens after topical chemotherapy. Cellular changes associated with bacillus Calmette-Gukrin vaccine. an increasingly popular therapeutic agent for bladder carcinoma, have not been documented, and treatment with this agent
The activ’e agent is concentrated in the urine and remains in contact with urothelial cells for relatively prolonged periods until voided. In poorly hydrated patients and certain particularly sensitive individuals, characteristic cells with increased nuclear:cytoplasmic ratios and degenerated nuclei with a “smudged” appearance have been observed (Fig 11). The nature of these elements is obscure but they are apparently not malignanl. They are, however, an important signal of a disturbed urothelium, and their presence should be recorded so that appropriate measures can be taken. At least 42 well-documented cases of bladder cancer have occurred after treatment with cyclophosphamide.17 Mitornycin C and thio-TEPA are topically applied alkylated agents administered in their active 893
HUMAN PATHOLOGY
Volume 21, No. 9 (September 1990)
FIGURE II. Cells characteristic of cyclophosphamide effect. Although not diagnostic, the presence of cells with relatively low nuclearcytoplasmic ratios and degenerated “smudged” chromatin is characteristic of treatment with cyclophosphamide. (EA 50 stain; magnification x 1,109.)
ties, but the changes are nonspecific and should not confuse the diagnosis of cancer. Since degenerative changes are common after x-ray therapy, the cytopathologist is often asked to vouch for the “viability” of any tumor cells identified. This is, of course, a fruitless exercise, if not an inappropriate question when applied to isolated cells in a urinary specimen. Most likely, x-rays affect urothelial neoplasms primarily by altering their blood supply through degenerative changes produced in endothelial cells.
should not confound the diagnosis of neoplasia in urinary specimens. Radiation has long been held responsible for certain cytologic changes, especially multinucleation, cytoplasmic eosinophilia and vacuolization, and nuclear pyknosis, but the effects of x-ray therapy on urothelial cells have not been rigorously tested in vivo and reliable examples of “radiation effect” are very difficult to ftnd.‘“~1”i~18 Like alkylating agents, it is probable that x-rays can cause bizarre nuclear abnormali-
FIGURE 12. Superficial cell with bizarre nucleus characteristic of treatment with either thio-TEPA or mitomycin C. Compare the size of this cell with the adjacent binucleated, relativety normal superficial cell. (EA 50 stain; magnification x 480.)
894
URINARY CYTOLOGY IN BLADDER NEOPLASIA (Murphy]
ADJUNCTNE APPROACHES TO URINARY SPECIMENS
nary cytology can detect ahnost all high-grade urotheliaf neoplasms. 3. The cytologic interpretation of low-grade transit.ional cell neopfasia requires expertise. These cells lack many of’ the features of‘ malignancy, a source of’ conf‘usion for the diagnostician but a positive factor for the patient since neoplasms composed of’ these cells are almost never aggressive. 5. The most usef‘ul type of urinarv specimen for routine diagnostic interpretation is :creshh voided, randomly collected urine. Catheterized sj,ecimens and bladder washings may yield more and better preserved cells, but no patient shoultf be catheterized solely to obtain diagnostic material. 6. Preservation of urinary specimlcns in alcohols is not necessary unless prolonged storage is contemplated. Refrigeration to prevent bacterial growth and inhibit further cellular degeneration is required, howe\,er. 7. Cytologic details are best displayed with membrane filtration but other types of processing are adequate. ‘l‘he computer-programmed ( ytocentrifuge is currently most popular. 8. Optimal recognition of’ c.).tologic details requires some fi)rm of’ P apanicolaou staining; Romanovsky dyes are fess desirable. 9. Urothelial cells with nucfear:c) toplasmic ratios of’ 1:fl or fess should not be interpreted as malignant regardless of the degree of anapfasia of their nuclei. IO. Papillary aggregation is not a I,eliabfe feature of’ low-grade neoplasia in urinary sam.ples. 1 1. Using appropriate criteria, the differential diagnosis of urothelial neopfasia versus the reactive/ regenerative/reparative changes seconcfary to urinal.) stones can almost always be accomplished. 12. Alkvlatinq aients such as Cvtoxan, thio‘I‘EPA. and mitomycin C produce cha;.acteristic bitt nonspecific changes in urotheliaf cells. These changes rarely mimic those of carcinoma. ‘l‘he diagnosis of’ urothelial neoplasia need not be conf’ounded by pre\,ious treatment. 13. Flolz cytometry and digitized image analysis are current]! used for diagnostic interpretations of urinary specnnens in selected centers. ‘I‘heir routine use must await further refinements in instrumentation and the formulation of more se.arching questions.
The most promising adjunctive approaches to the evaluation of urinary specimens are flow cytometry and digitized image analysis.5~1!‘.S” Ar present, both techniques are being used for diagnostic purposes in reference laboratories as well as at selected institutions, but neither has been widely accepted in routine Ipractice. Both techniques are capable of multiple functions. some relating to direct measurements and others to computations. Both ha1.e been useful in practical situations in inverse proportion to the diagnostic yield of’ fight microscopic analysis at any particular irlstitlition.~‘.~ Although both flow cytometr) and image analysis can be perf’ormed on specimens also exarninecf fly light microscopy, most studies have declined to use this approach, so that full comparisons am’ong methods have not been possible. The most common task to which these complex instruments have been dedicated has been the quantitation of‘ nucle,!tr DNA. The flow cytometer can examine nuclei faster than the image analyzer, but the pathofogist cannot actually see the cells being evaluated and must rell; on computer-generated histograms which fle has n& procfuced himself. The image analyzer is interactive. ie, the cells being analyzed can also fle seen and selected b\. the pathoioglst. fmt the preselection itself ma): p&entially introduce interpretive IGas and I:he procedure is, at least to the modern corn-puter operator. painfully slow. In our,judgment. neither method is capable of eff’ectively screening a “normal” population for flladder cancer, despite adI ertisements to the contrary. Until these acljunctive methods can be f‘urther refined, eg, capable of measuring tumor’ cell kinetics, of identifying significant abnormalities beyond the limits oflight microscopy, OI of evaluating the characteristics of tumor cells which have not been isolated from one another, their routine diagnostic use for urinary specimens will remain limited.
SUMMARY Pathologic examination of urinary specimens is increasingly recognized as an essential component of detection and monitoring for patients with bladder neoplasms. Among the available techniques, urinal-.! cvtology i:< the most usef’ul. The current status of WInarv cvtolog’. can he summarized as follows: ,
REFERENCES
1. The demand for urinary cytology is steadily, ilrcreasing as clinicians have realized the limitations of c!zstoscopy and even biopsy for monitoring bladder cancer patients, especially those havina carcinoma in situ or receiving topical therapy. 2. Urinary cvtology is currently an essential procedure for monitoring all patients with urothelial neoplasms and, if‘ consistently used, can actually decrease the frequency with which patients need to be subjected to cvstoscopy. 3. Even i’n moderately experienred hands, uri-
I. (:nnr
,JD, Murphy
M’M,
‘Llh: t’rt)gnostlc signlfi;iiiel- intravesical mitom)cin C for superficial bladder canter.
