An epi-fluorescence microscope equipped with a 100-watt mercury lamp and appropriate filters is recommended to detect multicolor fluorescent signals. Specimens should be scanned at 400× magnification to locate cells with nuclear signals and then analyzed under 600–1000× magnification. Systematic analysis of signals follows and a standardized and repeatable approach is essential (e.g., starting analysis in the upper left quadrant of the target area and continue scanning from left to right and top to bottom). Signals should be counted in abnormal cells, which are defined by their nuclear features, namely enlarged nuclear size, irregular nuclear borders, and “patchy” 4,6-diamidino, 2-phenylindole dihydrochloride (DAPI) staining. Besides the single cells, cell clusters may be evaluated avoiding counting signals in overlapping nuclei. The number of signals for all four probes should be counted for each abnormal cell and recorded only when there is a gain (≥2 signals) for two or more chromosomes 3 (red), 7 (green), 17 (aqua), or there is a loss of both copies of 9p21 (gold) as recommended by the manufacturer. Total number of abnormal cells analyzed should be recorded. The test is considered positive when ≥4 of the 25 analyzed cells show gains for two or more chromosomes or ≥12 of the 25 cells have zero 9p21 signals. If these criteria for test positivity are not met after viewing 25 cells, analysis should be continued until the entire sample is analyzed. Some authors have recommended optimization of the criteria with regard to the presence of benign tetraploid cells [2, 3]. Tetraploid, or less commonly octaploid, cells showing four or eight signals for each of the four FISH probes can occur in reactive cells (e.g., umbrella cells) and may give rise to false-positive FISH results. Thus, FISH results should not be considered positive based on a tetraploid pattern unless they are numerous (e.g., ≥10). Along this line, the FDA-approved package insert emphasizes that results at or near the cut-off point should be interpreted with caution. A schematic illustration of typical U-FISH findings is shown in Fig. 9.2.

There are several conditions that may prevent accurate FISH examination, including the presence of lubricant jelly, corpora amylacea, degenerated cells, crystals, sperm, and bacteria.

In the case of performing FISH on conduit urines, interference by inflammation and cell debris may occur. Squamous contamination, especially in urines from female subjects, may obscure urothelial cells. If there are only a few atypical cells present, then a targeted FISH approach, using the same Papanicolaou-stained preparation as the cytology slide, is advisable.

U-FISH testing of voided urine has been approved by the FDA as an aid for initial diagnosis of bladder cancer in patients with hematuria and/or subsequent monitoring for tumor recurrence in patients previously diagnosed with bladder cancer. In routine practice, there are several additional situations in which U-FISH is frequently used [8]. In a meta-analysis, the pooled sensitivity and specificity of FISH were 72 % and 83 %, respectively, as compared with 42 % and 96 %, respectively for cytology [10]. Putting published data of U-FISH results into the right perspective is challenging due to great variability of critical parameters; these include the type of specimen (voided urine versus bladder washing), clinical situation (history or no history of bladder cancer), concurrent cystoscopy findings, length and type of follow-up, proportion of low grade and high grade tumors, local experience in FISH analysis, and the definition of a positive result, among others. This emphasizes the need for a well-controlled and independently funded study to clarify the performance of FISH in clearly defined situations.

Despite the evident utility of FISH in patients with AUC results, there is a possibility of false-positive results. Besides “anticipatory positive” FISH results, low specificity and low PPV in some studies might also be due to an increased number of tetraploid or dividing cells under reactive conditions. Although tetraploid cells are not specifically mentioned in the scoring guidelines of the manufacturer, it has now been recognized that rare cells with a tetraploid signal pattern (i.e., four signals of each probe) should not be taken as an unequivocally positive FISH result but interpreted with caution [3, 24–26]. In contrast, unbalanced numerical changes of one or more chromosomes (e.g., 2, 3, 5) or loss of 9p21 is virtually specific for neoplasia in bladder cytology. An exception is that pelvic irradiation (e.g., of the prostate or the uterus) often results in permanent chromosomal aberrations carrying a risk of a false-positive diagnosis by U-FISH. This is particularly the case for polysomies, whereas homozygous or heterozygous deletion of 9p21 remains highly specific for neoplasia even after irradiation. This emphasizes the need of appropriate clinical information and consideration of the typical postirradiation cytological changes. Decoy cells, i.e., polyoma-infected urothelial cells, can easily mislead by a false-positive cytological diagnosis of UC if one is not familiar with the morphological spectrum of decoy cells. Although there have been reports on rare FISH-positive decoy cells, most data suggest that polyomavirus infection does not interfere with U-FISH results [27].

Target FISH, in which there is cytological pre-screening of slides and cells in question prior to FISH, has been proposed as another way of making the FISH analysis more precise and specific by avoiding the analysis of obviously reactive bystander cells or the analysis of remaining material that might be devoid of the atypical cells and therefore not representative of the original slide [6, 24, 25]. This approach requires hybridization and scoring of the original stained slides and automated re-localization of the cells in question, ideally coupled with image-aided interpretation.

Several cytology -based tests other than DNA in situ hybridization have been developed to improve diagnostic accuracy of urinary cytology, especially its negative predictive value. To put the role of such markers into the right perspective, the performance of cytology is only low for LGUC but not for HGUC. Therefore, combining results of these tests is misleading.

Some of the tests have been approved for diagnostic use and surveillance by the FDA but none of them has been incorporated into recommended guidelines or clinical management [1, 28, 29]. Multiple studies have demonstrated that some urine markers are more sensitive than conventional urine cytology and there is also evidence that these tests may predict tumor progression and risk-stratify patients who are being treated with intravesical therapies [1, 28–32]. Data from meta-analyses suggest that cell-based assays may be more sensitive if compared to other assays especially in low grade tumors [29, 32]. This observation may be explained by the fact that current cell-based assays include markers for low grade tumors and are less confounded by urinary tract infections than liquid-based markers.

Liquid- or non-cytology-based tests can be applied to voided urine samples either without the need for further preparation or on sediments of centrifuged urine samples. The two most commonly applied liquid-based urine tests on pure voided urine samples are the BTA™ (Polymedco Inc., Cortlandt Manor, NY, USA) and the NMP22™ (Bladder Check) test (Alere Inc., Waltham, MA, USA), both approved by the FDA, both for detection of bladder cancer in symptomatic patients and for monitoring of patients with a history of bladder cancer. For both tests a qualitative point of care test (BTA stat™ and NMP22-BladderChek™) and a quantitative version (NMP22™ and BTA TRAK™) are available. The latter tests require a dedicated laboratory with specialized personnel. The advantage of a point of care test is that it gives an immediate result and therefore can be done immediately prior to cystoscopy. A positive urine test may lead to a diagnostic review bias, which means that with the knowledge of a positive urine test result, the urologist may be more likely to detect a bladder neoplasm at cystoscopy [44]. These tests have been developed for both voided urine samples (not first-morning urine) and catheter-collected urines but not for barbotage fluids.