CHAPTER 12 Urine cytology
Introduction
For practical purposes, clinicians obtain urinary tract cytology specimens for evaluation for the presence or absence of cancer. Urine specimens provide a critical role in the evaluation of patients who have signs such as haematuria and/or symptoms such as painful urination suggestive of pathology within the urinary tract. In some regions with high industrial exposure to known bladder carcinogens, urine cytology is used as the initial test for screening for early detection of bladder cancer.1,2 Urine cytology is also currently a mainstay in the surveillance of patients who have a history of urinary tract malignancy.
Specimen types
Voided urine samples
Voided specimens are also known as clean catch specimens and are obtained in patients who have a variety of clinical histories. A random specimen (and not an early morning specimen) is recommended to limit cellular degeneration.3 As mentioned above, voided urine specimens tend to be of low cellularity, unless the patient has:
Highly cellular specimens require thorough evaluation, and the cytomorphology of these four conditions will be discussed below. Voided urines are superior to instrumented urines in detecting malignancies of the urethra.
Diagnostic categories
As in all cytology, an adequate sample is critical to making a proper interpretation.4 Although cytopathologists and cytotechnologists intuitively assess the adequacy of urine specimens, well-established criteria for specimen adequacy are lacking for different specimen types and preparation methods. The greatest difficulty arises in the adequacy assessment of voided urine specimens, as they are of lower cellularity compared to other specimen types. Optimally, the cellularity of voided samples should be described, e.g. reporting no cells, a less than optimal number of cells and numerous cells, in addition to rendering a diagnosis. Adequacy is currently assessed informally, based on a number of factors including experience. If no urothelial cells are seen, specimen re-preparation should be considered. Generally speaking, most voided urines contain at least a few urothelial and inflammatory cells or acellular components such as crystals or debris.
Clinicians use urinary tract cytology diagnostic categories (Box 12.1) to triage patients for additional follow-up and treatment.5 At present, there is neither general agreement on the use of these categories by cytopathologists nor coherent clinical recommendations on how patients with different clinical histories, signs, and/or symptoms should be managed, based on specific diagnoses. Currently used diagnostic categories confer specific risks of malignancy, and these risks depend on cytopathologist experience and other factors. In the ideal practice setting, clinicians behave in a Bayesian fashion and use their pre-urine test probability of malignancy with the specific diagnostic category risk of malignancy to calculate the post-urine test probability of malignancy.6,7 As expected, the categories of benign, atypical, suspicious, and malignant confer an increasing probability of malignancy, although the actual patient risk of malignancy depends on far more than the cytopathologist’s interpretation. Published articles on the use of diagnostic categories and clinical decision-making describe the complexity of this process.6–8
One of the most understudied areas in urine cytology is in the cytopathologist’s use and meaning of indeterminate diagnoses, especially the diagnosis of atypical. Providing cytomorphological criteria for atypia is extremely difficult, and to state the obvious, atypical cells do not have definitive features of benignity or malignancy.9,10 Compared with cervical smear test cytology in which the risk of pre-neoplasia is relatively well established for an ‘atypical squamous cells – undetermined significance’ (ASC-US) interpretation, the risk of neoplasia/malignancy associated with an atypical urine is not. Throughout this chapter, the concept of atypia is discussed.
Cytomorphological criteria of benign and neoplastic urinary tract specimens
Non-neoplastic or ‘normal’ urine specimens
Voided urine specimens
Urothelial cells. Like squamous epithelium, urothelium is divided into three cell layers: superficial, intermediate and basal/parabasal. Urothelium varies in thickness from two to three cells in the renal pelvis to five to six in the bladder (Fig. 12.1). The superficial urothelial cell layer is a single cell in thickness and superficial cells are large, sometimes multinucleated, and also known as umbrella cells because of their contours. Superficial urothelial cells may be larger than 100 μm in diameter, although their nuclear to cytoplasmic ratio is low, except in reactive conditions or in degenerate specimens. Most superficial cells contain from one to three nuclei, although it is not unusual to see more nuclei, as these cells respond to a variety of insults. Overall, a superficial urothelial cell is approximately the size of an intermediate squamous cell. The cytoplasm of superficial cells is variable, but tends to be finely granular (Fig. 12.2); cytoplasmic vacuolation is fairly common, especially in reparative processes. The cytoplasmic borders are well defined and the cells have a rounded appearance. Superficial cells are the most common urothelial cell type seen in voided urine specimens from non-diseased individuals. Seeing cells from other urothelial layers is a sign of disease in a voided urine specimen.
The intermediate cell layer is variable in thickness and the cells are cuboidal and much smaller than intact superficial cells. The basal/parabasal cell layer also is a single layer in thickness and the cells are a similar shape but slightly smaller compared to intermediate urothelial cells (Fig. 12.3).
Squamous cells. The female urethra is entirely lined by stratified squamous epithelium, and the male urethra is lined by stratified squamous epithelium at the orifice of the penile (spongy) portion. The remainder of the penile portion and the membranous portion are lined partially by stratified squamous epithelium and partially by urothelium. Squamous epithelium also may be found in the bladder, and is normally seen in the trigone in 10% of men and 50% of women.11 In voided urine specimens, superficial squamous cells from the urothelial tract may be present (Fig. 12.4). In women, the urinary tract squamous cells are often confused with benign squamous cell contaminant from the gynecological tract. The presence of bacteria associated with squames, in the absence of reactive changes and inflammation, is usually a sign that the squamous cells are not of urinary tract origin (Fig. 12.5).
Glandular cells. Glandular cells are more frequently seen in instrumented urine specimens and have a number of sources (Box 12.2).12,13 Most glandular cells arise from the urothelial lining itself, as the urothelium is capable of differentiating along several pathways. Glandular epithelium also may arise as a response to chronic irritation. Mucous-secreting glands are normally found in the spongy portion of the male urethra and in the female urethra, hence these cells may be shed in voided urine specimens. Glandular cells exhibit oval, basal nuclei with finely granular blue to green cytoplasm. Cytoplasmic vacuolation and even cilia may be seen. Renal tubular cells are observed in casts or small sheets and reflect kidney disease such as infarct or tubular necrosis.
Box 12.2 Sources of glandular cells in urine specimens
Urothelial carcinoma with adenocarcinomatous differentiation
A variety of cells other than urothelial, squamous and glandular cells and other structures may be seen in voided urine specimens. These cells include inflammatory cells, red blood cells (Fig. 12.6) and sperm. Other structures include crystals (Fig. 12.7), casts (Fig. 12.8) and corpora amylacea.
Instrumented urine specimens
Instrumented specimens are characterised by high cellularity, as the method of cellular procurement forcibly removes epithelium (Fig. 12.9). These specimens show large groups of cells, small cell clusters and single cells from all urothelial cell layers. The large groups of urothelial cells often demonstrate all cell layers, with superficial cells overlying intermediate and basal/parabasal cells (Fig. 12.10). These groups are termed ‘instrumentation artefact’ and may be confused with low-grade urothelial carcinomas.14 At high power, the smaller groups of urothelial cells lack significant atypia (Fig. 12.11).
Catheterised urines tend to show more inflammation than other instrumented urine types, as indwelling catheters cause irritation and inflammation (Fig. 12.12).
Ileal conduit specimens
Ileal conduit specimens are predominantly composed of numerous intestinal glandular cells and inflammatory cells that often have a degenerate appearance (Fig. 12.13).15,16 The high cellularity arises from the natural sloughing of the intestinal epithelium, as glandular cells lack the hardiness to withstand the toxicity of urine. The glandular cells may form flat sheets, similar to the findings in colonic brushing specimens. The degenerate glandular cells contain a pyknotic and eccentrically placed nucleus. The nuclear to cytoplasmic ratios are not increased, but the smudged hyperchromasia may be alarming. Occasional upper tract superficial urothelial cells are seen.
Malignancy and its pitfalls
Urothelial cancer
Urothelial carcinoma is the most common cancer detected by urine cytology and the most common site of origin is the bladder. Worldwide, over 390 000 bladder cancers are detected annually.17 In most parts of the world, 90% of bladder cancers are urothelial in origin, 5% squamous and 5% mixed urothelial and squamous. Primary adenocarcinoma of the bladder is rare. Bladder cancer often presents insidiously, with more than 75% of patients having haematuria, which is often painless.18 Symptoms include urinary frequency, dysuria and urgency.
Most cytological diagnostic systems classify urothelial carcinomas into the categories of high grade and low grade. This system is far less complex than the currently used histopathological diagnostic grading systems (Box 12.3).19–21 Although the low-grade/high-grade cytological grading system does not precisely map to the histopathological grading systems, there generally is correlation between the better-differentiated tumours on histopathology with cytologically low-grade carcinomas and the more poorly differentiated tumours on histopathology with cytologically high-grade carcinomas.
As Oosterhuis et al. wrote, tumour progression may occur in all patients with all tumour types, except for those with papillomas, which often are not diagnosed on urine specimens.19 The key to the cytological diagnostic system is in flagging this risk, as clinicians use triage protocols based on clinical findings and the cytological diagnosis of low or high-grade urothelial carcinoma. For example, on a voided urine specimen, the diagnosis of a high-grade urothelial carcinoma generally results in additional follow-up, such as cystoscopic examination. On a bladder wash specimen, the diagnosis of low- or high-grade urothelial carcinoma is used in conjunction with the cystoscopic impression.
Diagnostic accuracy
Koss et al. reported that the sensitivity of voided urine for the diagnosis of high-grade urothelial carcinoma was 94.2%.22 The sensitivity increased as the number of voided urine specimens increased, with 79% of cancers detected on the first specimen, 14% on the second specimen and 7% on the third specimen. Other authors have confirmed this high level of sensitivity.2,23–25 Monolayer preparation methods detect high-grade lesions at a similar percentage,26 although indeterminate rates may initially rise after implementation.27
The calculation of the specificity of voided urine cytology for high-grade urothelial carcinoma depends on how indeterminate diagnoses are assessed, although the false positive rate of an outright diagnosis of high-grade urothelial carcinoma is extremely low. In some patients, lesions are not visualised on cystoscopic examination even following a positive urine cytology, indicating the limits of cystoscopy and biopsy as a gold standard.4
In a study using histological follow-up as the gold standard, 40% (55 of 136) patients had an indeterminate voided urine diagnosis (atypical or suspicious) and follow-up of a high-grade urothelial carcinoma; only 26% of patients had a benign lesion.4 Although this study involved a number of high-risk patients who had a history of cancer, the data indicate that indeterminate diagnoses harbour a relatively high risk of cancer.
The sensitivity of bladder wash specimens for the diagnosis of high-grade urothelial carcinoma is higher than that seen in voided urine specimens.4,28,29
The sensitivity and specificity of the diagnosis of low-grade urothelial carcinoma in instrumented urine specimens is controversial. The reported sensitivity (based on making a definitively malignant diagnosis) in the literature ranges from 0% to 73%, although in experienced hands, the sensitivity is over 50%.23,30–32 Based on the same follow-up study listed above, 67% (14 of 21) patients had a diagnosis of atypical, suspicious or malignant and a low-grade urothelial carcinoma on follow-up.4
The literature is virtually unanimous that the outright diagnosis of a low-grade urothelial carcinoma is exceedingly difficult to make on a voided urine specimen. However, one study showed that 48% of biopsy proven low-grade urothelial carcinomas had a voided urine diagnosis of atypical or suspicious.4 In this study, 11% of all institutional voided urine cases had an indeterminate diagnosis, as the vast majority of patients did not have histology follow-up. These data indicate that indeterminate diagnoses raise the post-test probability of low-grade cancer, even in voided urine specimens.
High-grade urothelial carcinoma
Cytological findings: high-grade urothelial carcinoma
The diagnosis of high-grade urothelial carcinoma corresponds to either an in situ or an invasive lesion. As a rule, the cytological features are not definitive in making this separation, although the more anaplastic the tumour, the more likely it is that the tumour is invasive.24,25 Invasive high-grade urothelial carcinomas have a tendency to be clinically aggressive and most cancer deaths from papillary urothelial carcinomas are from the high-grade types.33
Voided urine specimens
The malignant cells of a high-grade urothelial carcinoma are generally observed singly, although loose clusters also may be present.24,34 In voided urines, only rare cells may be seen, indicating that careful screening, as for cervical smear tests, is necessary.
The malignant cells vary in size and shape, and in some cases, huge malignant cells may be seen (Fig. 12.14). The nuclear:cytoplasmic ratio is increased, and most malignant cells have only a thin rim of dense, homogeneous or non-vacuolated cytoplasm (Fig. 12.15).32 Cytoplasmic vacuolation is not a typical feature, but may be observed in cases with abundant acute inflammation. The cytoplasmic membranes are easily distinguished in high-grade urothelial carcinomas.
The appearance of the nucleus is the defining feature of high-grade urothelial carcinoma. The nucleus is large and exhibits marked nuclear membrane irregularities (Fig. 12.16). The nuclear chromatin is extremely coarse and some cells have a West Virginia ‘coal black’ appearance, indicating that the chromatin is uniformly dark and impenetrable in appearance due to degenerative changes (Fig. 12.17).32,35 When present, nucleoli are large and prominent and cytophagocytosis may be present (Fig. 12.18).36 Nassar et al. reported that greater smudging of cancer cells was observed in monolayer preparations, although monolayer preparations also showed more even disbursement of malignant cells.37
High-grade urothelial carcinomas occasionally exhibit squamous or glandular differentiation, squamous being the more frequent (Fig. 12.19).23 Squamous differentiation is characterised by keratinisation, sometimes manifested by parakeratosis or keratin pearls (Fig. 12.20). Malignant glandular differentiation is difficult to recognise but is discernible by cytoplasmic vacuolation. Although invasive high-grade urothelial carcinomas may exhibit sarcomatoid differentiation, the malignant cells are not typically observed in voided urine specimens but may be seen in washing specimens (Fig. 12.21).
Ileal conduit urine specimens
High-grade urothelial carcinoma cells tend to be few and are admixed with the numerous degenerate glandular cells, as described above. The malignant cells often are better preserved, are larger, and maintain crisper nuclear membranes, compared to the degenerate cells (Fig. 12.22).15,16
Instrumented urine specimens
In instrumented urine specimens, numerous malignant cells usually are seen (Fig. 12.23) and the malignant cells have the same cytological features as described for voided urine specimens. Small- to moderate-sized clusters of malignant cells may be seen, but large groups of malignant cells are unusual, as high-grade urothelial carcinomas typically show extensive cellular dissociation.
Upper tract specimens
Benign urothelial cells of the upper tract, compared with their lower tract counterparts, show higher nuclear:cytoplasmic ratios, more prominent nuclear membrane irregularities and larger nuclei (Fig. 12.24). Instrumentation produces large groups of cells. Nonetheless, the diagnosis of high-grade urothelial carcinoma tends to be straightforward, as the nuclear size, hyperchromasia and membrane irregularities exceed the changes of benign urothelium.38–40 Potts et al reported that the criteria of high nuclear:cytoplasmic ratios, anisonucleosis and nuclear overlapping were among the most important criteria in diagnosing high-grade urothelial carcinoma of the upper tract (Fig. 12.25).41