Urine cytology was popularized by George Papanicolaou in the 1940s as a way to detect and follow patients with bladder cancer.¹ By the 1960s, the cytologic, histologic, and clinical features of high-grade urothelial carcinoma (UC) were well established.² Cytology remains an inexpensive, quick, and reliable way to diagnose high-grade UC³ as well as a variety of less common tumors. Its reproducibility is variable among laboratories,⁴ however, and this is a significant weakness of the test. A number of ancillary tests have been introduced (e.g., UroVysion) in an attempt to improve accuracy, but their value and precise role are still debated.

Some have argued that there is no significant cost benefit to urine cytology in specific settings ⁵ like hematuria.⁶ Furthermore, the U.S. Preventive Services Task Force has found scant high-quality evidence in support of urine cytology and related ancillary studies.⁷ Perhaps it is not surprising that a test adopted more than 50 years ago does not meet contemporary standards for high-quality evidence. It is unlikely that such standards will be met any time in the near future, but alternatives to a randomized prospective trial have been proposed.⁸

Clinical indications for urine cytology

• hematuria

• follow-up of patients treated for urothelial carcinoma

• high risk of bladder cancer

The most common indication for urinary cytology is hematuria.9,10 The yield, however, is low: Hematuria is caused by a malignancy in only 5% to 10% of patients.^11–13 Another indication is surveillance for recurrent UC, because patients with a previously diagnosed and treated urothelial cancer are at risk for recurrence or a de novo primary elsewhere in the urinary tract.² Urine cytology is not used for screening asymptomatic individuals, because the benefits are outweighed by the considerable cost.^14–16 It is used, however, when an individual has risk factors for bladder cancer, such an occupational exposure to aniline dyes,¹⁷ the aromatic amines used in the petrochemical industry,¹⁸ or cyclophosphamide treatment for diseases like multiple sclerosis.¹⁹

Specimen Collection

There are six types of urinary specimens, each with its relative advantages (Table 3.1).

TABLE 3.1

RELATIVE ADVANTAGES AND DISADVANTAGES OF URINE SPECIMEN TYPES

Specimen Type	Advantages	Disadvantages
Voided urine	Noninvasive No instrumentation artifact	Low cellularity Vaginal contamination Poor preservation
Catheterized	High cellularity	Invasive Instrumentation artifact Poor preservation
Bladder washing	High cellularity Good cell preservation	Invasive Instrumentation artifact
Upper tract washing	High cellularity Good preservation Selective sampling	Invasive Instrumentation artifact
Brush cytology	Selective sampling	Invasive Air drying possible (if direct smear)
Ileal loop	Permits screening for recurrent bladder cancer	Low cellularity Poor preservation

Voided Urine

Voided urine should be obtained 3 to 4 hours after the patient has last urinated.20–22 First morning voided urine specimens should be avoided, because cells in a stagnant, low pH, and hypertonic environment undergo degenerative changes, making cytologic assessment difficult. The minimum amount of urine necessary to ensure adequate cellularity is unknown, but it may be as high as 25 to 100 mL.²³

In women, voided urine can be contaminated by vaginal cells, but in most instances this does not compromise a diagnosis. Still, to help ensure the adequacy of the sample, a midstream (“clean catch”) specimen is recommended.

The most common upper tract specimen is the directed washing with or without an upper tract biopsy.²⁵ Directed washing specimens are particularly challenging for urologists and cytologists.^25–32 Urologists often cannot visualize lesions in the upper tract as well as those in the bladder; hence they rely on cytology here even more than for lesions in the bladder. (Although they may try to obtain a biopsy, often these specimens are small and crushed.) The stakes are high, because the operation of choice for a tumor in the upper tract is removal of the kidney and/or ureter. In most cases, the significant imaging finding is a filling defect, and the differential diagnosis is a tumor or a stone. Unfortunately, benign cytologic atypia produced by some stones can mimic the cytologic features of urothelial neoplasms. Finally, normal specimens from the upper tract often show diffuse nuclear enlargement, an elevated nuclear-to-cytoplasmic ratio, and very high cellularity. These entirely benign changes can suggest a tumor and result in a false-positive diagnosis.²⁹ For these reasons, it is impossible to accurately diagnose low-grade lesions in upper tract specimens. For high-grade tumors, the sensitivity rates of ureteral washing cytology and ureteral biopsy for the detection of malignancy are similar and approach 70% to 80%.²⁵ With bilateral specimens, one can compare subtle changes between a lesion (on one side) and a presumably normal specimen (on the other side). The preparation of a “cell block” (a formalin-fixed, paraffin-embedded sediment of the urine sample) can be particularly useful, because small pieces of tumor are often easier to evaluate with this preparation method.^25,33

Ileal Conduit Samples

At the time of cystectomy for bladder cancer, a segment of ileum is isolated and reanastomosed to the ureters (or to one ureter if a nephrectomy is also performed) to provide a conduit for urine. Urine samples from these conduits contain a large number of degenerated intestinal epithelial cells. It is important that these specimens be screened for malignant cells, because patients with a history of bladder cancer have an increased risk for developing tumors of the ureters and kidneys.34,35

Processing

Fresh specimens, between 1 and 12 hours old, do not need fixation. If it will take a specimen 12 to 24 hours to reach the laboratory, refrigeration is recommended, and if more than 24 hours, preservation with an equal volume of 50% to 70% ethanol ± 2% carbowax is advised to avoid degeneration.²²

Slides can be prepared using a variety of concentration techniques, depending on the resources and preferences of the laboratory. These include sedimentation and smearing, membrane filtration, cytocentrifugation, and thinlayer methods. Slides prepared using one of these methods are fixed in ethyl alcohol and stained with the Papanicolaou stain. Urine samples and bladder washings can also be prepared using the cell block technique;³³ the centrifuged sediment is fixed in formalin, and the slides are stained using hematoxylin and eosin.

Reporting Terminology and Adequacy Criteria

Clinically meaningful adequacy criteria have not been defined. Nevertheless, there are four situations where the question of adequacy arises.

Patterns of nondiagnostic (unsatisfactory) specimens

• vaginal cells only

• obscuring inflammation or lubricant

• blood only

• marked degenerative changes

With the exception of ileal pouch specimens, a urine sample must contain at least some urothelial cells to ensure that it is indeed urine. For example, voided urine samples from women commonly consist almost entirely of squamous cells from the vagina. In some cases, the urothelial cells are obscured by abundant acute inflammation, presumably due to an infection, or by lubricant jelly. Some specimens consist only of blood. In all three cases, it is reasonable to require the identification of at least one urothelial cell in order to be sure the specimen is adequate and represents urine. Alternatively, one can report the case as benign, but append a note stating that urothelial cells are poorly represented, and clinical correlation concerning the adequacy of the specimen is advised.

Finally, degenerated cells are very common. If the specimen is so degenerated that an interpretation cannot be made (an uncommon circumstance), the specimen is inadequate. If only some cells are degenerated, and particularly if the degenerated cells appear atypical, it is advisable to interpret the specimen as abnormal rather than nondiagnostic (see “Diagnosing Difficult or Borderline Specimens: Common Patterns”).

In most laboratories, the results of urine cytology and bladder washings are reported using simple diagnostic categories such as “negative” (“no malignant cells identified,” or “benign”), “atypical” (“atypia of undetermined significance,” “mildly atypical urothelial cells, favor reactive;” “rare atypical cells, too few to further characterize”), “suspicious” (i.e., abnormal-appearing urothelial cells that are suspicious for carcinoma), and “positive” (i.e., conclusive for malignancy). Because urine cytology is an insensitive test for low-grade urothelial neoplasms, some laboratories prefer the phrase “negative for a high-grade malignancy/carcinoma” rather than “negative for malignant cells.” For reasons given later (see “Diagnosing Difficult or Borderline Specimens: Common Patterns”), it is recommended that the “atypical” category be used as sparingly as possible.

Accuracy

Urine

Urine cytology is, at best, only moderately sensitive in detecting bladder cancer. A representative summary of published studies on the sensitivity of cytology is provided in Table 3.2. These studies likely overestimate the sensitivity of cytology, because virtually all are affected by selection bias: Patients with a malignant biopsy are more likely to have had a suspicious or positive cytology. “Positive” results are obtained in 25% to 72% of patients with bladder cancer when all grades and stages of tumors are included in the analysis. Several variables affect the sensitivity of urine cytology. First, sensitivity is higher (37% to 89%) when suspicious diagnoses are included with positive diagnoses.^36,37 Second, the sensitivity increases when more than one specimen is examined: Tumor cells may be absent from one urine specimen but present in subsequent specimens.^36,38,39 For this reason, it has been recommended that at least three specimens per patient be examined.⁴⁰ Third, the sensitivity of urine cytology is highly dependent on the grade of the bladder tumor. Low-grade UCs are detected less reliably or not at all by cytology, as compared with high-grade carcinomas.^{36,37,41–43} Finally, the sensitivity of urine cytology may be reduced in patients who have been treated with radiation or chemotherapy.³⁷ Molecular cytogenetic analysis using the fluorescence in situ hybridization (FISH) method may be more sensitive than cytology but less specific (see “Ancillary Techniques”).

The moderate sensitivity of cytology is complemented by its very high specificity (range of most studies: 95% to 100%).⁴² False-positive results, in other words, are uncommon. Some investigators report finding no false-positive results in their studies^36,44; others describe rates ranging from 1.3% to 15%.^40,45,46 False-positive results occur in patients with bladder stones,⁴⁷ human polyomavirus infection,⁴⁸ and chemotherapy.⁴¹ A positive cytologic result in the face of a negative biopsy result does not necessarily mean that the cytologic diagnosis is false. In many cases, a carcinoma that escaped histologic detection is discovered on a subsequent cystoscopic examination. Such a scenario (termed “anticipatory positives”) occurs for ancillary tests as well. Other sites besides the bladder can be the source of the malignant cells; a primary tumor of the ureters, kidneys, prostate, and other contiguous organs must be considered.^21,49

Urine cytology is complemented by cystoscopy. Low-grade tumors missed by cytology are papillary lesions readily visualized with the help of a cystoscope and thus earmarked for biopsy.50,51 Whether low-grade tumors can be reliably diagnosed on a voided urine sample, by cytology and/or an ancillary study, thus sparing the patient confirmatory cystoscopy, is not clear. Nevertheless, the chance of clinical progression of such a lesion is low. For high-grade carcinomas, however, particularly carcinoma in situ (CIS), which is more difficult to detect by cystoscopy, cytology provides a high degree of diagnostic accuracy, even in voided urine specimens.

Bladder Washings

Bladder washings have the advantage over urine samples in improved cellularity and cell preservation. The sensitivity of a positive bladder washing cytology is slightly higher than that of urine cytology, ranging from 66% to 77% when all grades and stages of bladder tumor are included (Table 3.2). The superiority of bladder washings over voided urine cytology has been well documented.^52–54 This is not to say that voided urine from patients undergoing cystoscopy can be neglected. From 7% to 13% of bladder tumors not detected by bladder washings are discovered in urine samples obtained before cystoscopic examination.^53,55

Bladder washing cytology is not without its drawbacks. Most significantly, catheterization is required to obtain the specimen. Bladder washings sample the bladder epithelium only, whereas urine contains cells exfoliated from the ureters and kidneys.

Washings of the ureter and pelvis have similar sensitivity (70% to 80% for high-grade lesions) but are particularly prone to false-positive results ²⁹ because of the marked cellularity of these specimens.

Normal Elements

Normal elements

• urothelial cells

• intermediate and superficial (“umbrella”) cells (voided urine)

• intermediate, superficial, and basal cells (catheterized urine, washings)

• squamous cells

• seminal vesicle epithelial cells (very rare)

• degenerated intestinal epithelial cells (ileal conduit specimens)

A normal voided urine specimen is sparsely cellular, but urothelial (synonym: transitional) cells are usually present (Fig. 3.1). They are dispersed as isolated individual cells; tight clusters of urothelial cells are distinctly uncommon in a normal voided urine sample. In voided specimens, most urothelial cells are intermediate in size, with a moderate amount of homogeneously granular or finely vacuolated cytoplasm, round nuclei, and small nucleoli. In some cases they are columnar or spindled; this is a normal finding, although the reason for these shapes is not known (Figs 3.2 and 3.3). When degenerated, urothelial cells resemble histiocytes, especially because they sometimes contain round, red or green hyaline cytoplasmic inclusions called Melamed-Wolinska bodies⁵⁶ (see Fig. 3.2). They are seen in almost 50% of urine specimens and are more common in voided than in catheterized samples. The pathogenesis of these bodies is obscure, and they have no diagnostic value in urine, but they are useful in suggesting a urothelial origin for malignant cells in pleural effusions.⁵⁷ Umbrella cells are large and have abundant cytoplasm and large nuclei; binucleation and multinucleation are common (see Fig. 3.3). Although large and often multinucleate, umbrella cells have a very low nuclear-to-cytoplasmic ratio, relatively fine chromatin, and thin, smooth nuclear membranes, which helps to distinguish them from malignant cells. Some squamous cells are common in voided urine samples; they exfoliate from foci of squamous metaplasia in the trigone of the bladder (a normal finding, especially in women). Squamous cells can also be picked up as urine passes through the urethral orifice and is contaminated by cells from the vagina. Significant vaginal contamination of a voided urine sample (composed almost exclusively of squamous cells and bacteria, the latter either normal flora or coccobacilli) may necessitate catheterization to obtain a more pure specimen.⁹

Figure 3.1 Normal voided urine.
Most benign voided urine samples show a mixture of urothelial cells and squamous cells. In voided urine, most of the urothelial cells are of “intermediate” type, with an oval or pyramidal shape.

Figure 3.2 Cytoplasmic inclusions (Melamed-Wolinska bodies).
Degenerating urothelial cells frequently have round, eosinophilic (or sometimes green) cytoplasmic inclusions of varying sizes. A normal columnar-shaped urothelial cell is also present.

Figure 3.3 Umbrella cells.
These are the largest urothelial cells and cover the surface of the urothelium. Normal columnar urothelial cells are also present.

In catheterized samples, including washings and brushings, clusters of urothelial cells, some quite large, are an entirely normal finding: The instrument mechanically abrades the mucosal surface, resulting in large numbers of cells in fragments. Thus, the entire spectrum of basal, intermediate, and superficial (umbrella) cells is seen. Normal catheterized specimens, particularly washings and brushings, may appear worrisome for malignancy, particularly to the novice cytologist, because of the presence of intact mucosal fragments and the marked polymorphism of the cell population. Umbrella cells alone may be worrisome because of their size: They are among the largest of human epithelial cells. Even the small basal urothelial cells (Fig. 3.4), because of their scant cytoplasm and dark nuclei, are occasionally mistaken for carcinoma cells.

Figure 3.4 Basal urothelial cells (catheterized specimen).
A, These cells are rare in voided urine but common in catheterized specimens and usually tightly clustered. B, Higher magnification reveals predominantly round, regular nuclear contours.

On rare occasions, seminal vesicle epithelial cells are seen in urine samples from male patients. They sometimes have hyperchromatic nuclei and can be mistaken for malignant cells. The clue to their benign nature is the presence of lipofuscin, a golden-brown cytoplasmic pigment (Fig. 3.5).

Figure 3.5 Seminal vesicle epithelial cells (voided urine).
These cells are recognizable as seminal vesicle cells because of their golden-brown pigment. Sometimes they are less well preserved than seen here.

The intestinal epithelial cells in ileal loop specimens are dispersed as isolated cells and show marked degenerative changes (Fig. 3.6), with eosinophilic intracytoplasmic inclusions such as those seen in degenerated urothelial cells. They are commonly mistaken for macrophages by the novice.

Figure 3.6 Ileal loop specimen.
Most cells in ileal loop specimens are degenerated intestinal cells which resemble macrophages (inset).

Crystals are commonly present. Occasional red blood cells are common, but the presence of many red blood cells is abnormal. Bacteria and Candida are commonly seen, usually without concomitant inflammation or clinical significance.

Benign Lesions

Infections

• bacteria, including malakoplakia

• fungi (especially Candida)

• herpes simplex virus

• cytomegalovirus

• Trichomonas vaginalis

• polyomavirus

• human papillomavirus

Infections of the bladder are caused by bacteria, viruses, parasites, or fungi. In the United States, bacterial infections are the most common. Cytologic preparations in cases of bacterial cystitis show a dense concentration of white blood cells (predominantly neutrophils), with plasma cells, lymphocytes, histiocytes, and numerous red blood cells. Bacteria are present, sometimes in overwhelming numbers. The inflammation can obscure the urothelial cells. The presence of bacteria in the absence of abundant neutrophils is nonspecific. This may be the result of vaginal or urethral contamination, but many patients, particularly females, have bacteria in the bladder without clinical symptoms of cystitis. Likewise, neutrophils are not necessarily indicative of a cystitis of infectious etiology.

Malakoplakia is an uncommon histiocytic inflammatory lesion of the bladder or upper respiratory tract that results from bacterial infection. Diagnosis by urine cytology is very uncommon. The cytologic hallmark is the presence of histiocytes, whose abundant granular cytoplasm is filled with bacteria and bacterial fragments. Often one finds basophilic, round, lamellated bodies, known as Michaelis-Gutmann bodies, which measure approximately 8 μm in diameter and may be intracellular within histiocytes or extracellular.

The most common fungus that infects the bladder is Candida. The organism is present as yeast forms and pseudohyphae, accompanied by a cellular, mixed inflammatory background. Urothelial cells usually show reactive changes. When Candida is present in the urine of female patients, the possibility of vaginal contamination should be considered. Vaginal contamination, rather than true infection, is likely when the background contains numerous squamous cells and bacteria and few neutrophils. As with bacteria, some patients have Candida in their bladder without symptoms of cystitis.

Viral infections of the bladder include herpes simplex virus, cytomegalovirus (CMV), polyomavirus, and human papillomavirus (HPV). Herpetic infection of the bladder is uncommon, usually seen in the immunocompromised patient. The cytopathic changes include multinucleation, a ground-glass chromatin texture, and peripheral condensation of chromatin. In some cases the cells have a large eosinophilic nuclear inclusion which can be sharply angulated. Nuclear molding is often observed. Infected cells can be greatly enlarged and bizarrely shaped, with dense, opaque cytoplasm.

CMV affects urinary epithelium, most commonly renal tubular cells, in immunocompromised patients. Affected cells are markedly enlarged and have both nuclear and cytoplasmic inclusions. The nuclear inclusion is solitary, darkly basophilic, and surrounded by a zone of chromatin clearing. The multiple cytoplasmic inclusions are more variable in appearance: smaller, basophilic, and either finely or coarsely granular.

The parasitic protozoan Trichomonas vaginalis is responsible for one of the most common sexually transmitted diseases. It is usually associated with vaginitis, but it can cause urethritis and even prostatitis. In urine from a woman, the organisms are most likely contaminants from a vaginal infection if they are accompanied by abundant squamous cells and vaginal flora.⁵⁸ Rarely, they cause urethritis in men and are identified in urine cytology from such patients.^59,60 The organism varies in size, with an average length and width of 10 and 7 μm, respectively. The nucleus is small and oval, and the cytoplasm contains fine red granules.

The human polyomaviruses (JC and BK viruses), members of the papovavirus family, commonly infect urothelial cells in both healthy and immunocompromised individuals, and characteristic viral cytopathic changes are seen in 4% of urine samples.⁶¹ The infection usually has no clinical significance, except in immunocompromised patients, particularly renal transplant recipients, but infected cells appear atypical and can be confused with malignant cells.⁴⁸ Infected urothelial cells have large, eccentrically placed nuclei with basophilic nuclear inclusions that completely replace the nucleus and appear glassy, opaque, or cloudy (Fig. 3.7). Nuclear membranes are markedly thickened. In children and immunosuppressed adults, the altered cells are numerous, whereas in immunocompetent adults, they are usually few in number.⁶² In contrast with CMV-infected cells, in which a halo surrounds the inclusion, the inclusion of polyomavirus fills the entire nucleus. Because of their increased nuclear size and hyperchromasia, polyomavirus-infected cells can be confused with malignant cells—hence their pseudonym “decoy cells.” Unlike most malignant cells, however, decoy cells have perfectly smooth and round nuclei. In contrast with tumor cells, which often cluster to form groups, polyomavirus-infected urothelial cells are found only as isolated cells.⁴⁸ Because degenerated UC cells can resemble decoy cells, however, one should not interpret specimens as negative unles the morphology of the decoy cells is indeed classic.⁶¹ Testing for polyomavirus by immunohistochenistry or polymerase chain reaction is available and can be helpful in difficult cases. The differential diagnosis of decoy cells also includes degenerated benign urothelial cells. Polyomavirus-infected nuclei appear smudged and densely basophilic, and the chromatin is more uniform in texture than that of degenerated urothelial cells.

Figure 3.7 Polyomavirus infection.
Some enlarged, round nuclei are virtually replaced by a glassy, homogeneous inclusion.

HPV can infect the urinary tract,⁶³ but when cytopathic changes characteristic of this virus are seen in a voided urine specimen from a woman, the cells most likely have originated from the vulva or vagina. Koilocytes in a catheterized specimen, however, indicate a condyloma in the urinary tract.