More than 50,000 new cases of urothelial carcinoma, which represents 90% of bladder cancer cases are diagnosed annually in Europe and in North America 1. About 70% of bladder urothelial carcinomas are superficial (TNM stage pTa-1) and may be viewed, diagnosed and treated by cystoscopy aided by biopsies and transurethral resection 2.

Despite it is recognized as the biological standard for the diagnosis and follow up of bladder tumors, urinary cytology has a mean sensitivity of about 50% and it is hampered by a large amount of non-diagnostic samples 3. Although urinary cytology detects about 80% of aggressive, high grade (G3) urothelial tumors, some results remain falsely negative, particularly in patients having had TUR or bacillus Calmette-Guérin immunotherapy. In urology practice, cystoscopy is commonly combined with urinary cytology, particularly in the search for high grade wherever its location in the urinary tract.

Liquid-based cytology (LBC) has been developed as a replacement to cytocentrifugation and/or smearing, owing to cell recovery capabilities and better cell preservation. Some LBC methods use a filtration process and a computer-assisted thin-layer deposition of cells (Cytyc Thinprep^® supplied by Cytyc Corp., Boxborough, MA), whereas others are based on a sedimentation process (AutoCyte^® PREP supplied by TRiPath Imaging, Burlington, NC). In the urine, the use of Cytyc Thinprep 2000 results in increased cellularity and marked reduction of debris, red blood cells (RBC) and crystals 4567.

However, optimization of cell capture and fixation as well as thin-layer deposition of cells can be achieved by other methods than LBC, particularly while using modern cytocentrifugation methods 7. In our experience based on 2500 specimens/year for 15 years, and provided specific requirements are followed, direct smears and cytocentrifugation with the Shandon Cytospin^® 4 (Thermo Electron Corp., Waltham, MA) produce highly satisfying cytological specimens.

Accordingly, the aim of our study was 1) to objectively analyze the quality of urine samples processed by a body of conventional thin-layer methods as compared with Cytyc Thinprep LBC and 2) to verify if differences noted have an impact on diagnostic accuracy.

The study population was composed of 224 urine samples taken in patients with symptoms suggesting bladder cancer (gross hematuria, micturition disorders, chronic urinary infection) in 89 cases (39.7%), or followed after transurethral resection for bladder urothelial carcinoma in 135 cases (60.3%).

Urinary samples were taken after cystoscopy in 157 cases (63.8%), and after simple micturition in other cases. All samples were immediately fixed with 50% ethanol (V/V) or with a 20% Polyethyleneglycol 1500 (Merck, Darmstadt, Germany) solution in 50% ethanol (1/3 fixative and 2/3 urine).

Urine samples were sent to the laboratory and separated into two aliquots after homogeneization. One of the aliquots was processed according to the Thinprep LBC recommendations, and the other was processed according to a smear method, by cytocentrifugation or by filtration.

Using the scoring system as described in the Materials and Methods section, and considering a 0–3 scale, mean and standard deviations as well as the statistical significance of each parameter are shown in Figure 1,2, 3, 4, 5.

Differences noted concern global quality (cellularity and fixation combined) on the one hand, number of RBC and leukocytes on the other hand. Surprisingly, we found that smears allowed obtaining a global quality superimposable to that of Cytyc Thinprep slides (Figure 6). More precisely, the cellularity scores obtained by smears and LBC were 1.97 ± 0.86 versus 1.96 ± 0.68, respectively (p = ns), whereas values for fixation were 2.58 ± 0.48 versus 2.50 ± 0.59 (p = ns).

Cytocentrifugation with 3 years' old reusable sample chambers resulted in significant decrease in both cellularity and fixation quality, whereas cytocentrifugation with disposable sample chambers (whatever the type of chamber used) allowed obtaining the better results (Figure 7).

Millipore filtration resulted in impaired cell preservation even after blotting of cells on coated slides and careful fixation.

The concentration of RBC was significantly decreased after LBC treatment of samples in all circumstances except in Millipore* filtration using 5 μm porosity membranes. Similar comments may be done about leukocytes.

Whatever the technique studied, the search for cell groups and atypias gave results identical than those of LBC except for smears which showed a slightly higher percentage (p = 0.01). However, the values obtained were not strikingly different.

As far back as the late 'seventies, authors have attempted to compare cytocentrifugation with other methods such as filtration 1112. In those preliminary studies, Millipore filtration was found to give better cell recovery and better morphologic details than cytocentrifugation. However the methods used (reusable sample chambers) was suboptimal: a significant cell loss can be attributed to the roughness of sample chamber walls secondary to repeated cleaning 13.

Waiting the 'nineties was necessary for obtaining comparisons between the Cytyc Thinprep LBC and other methods, with some contradictory results. Many of the studies, published as abstracts of the 40^th and 41^st Annual Scientific Meetings of the International Academy of Cytology, were not transformed into full length articles 451415.

Except for one study which showed processing time and cost several times greater for Cytyc Thinprep LBC than for polycarbonate membrane filtration 15, most series recognize advantages in using LBC. In a recent study comparing cytocentrifugation to Cytyc Thinprep, Cytospin preparations were found superior to LBC in terms of cytomorphologic details and preservation of architectural patterns 16. However the advantage of LBC concerning cleaner background was noted.

Cytocentrifugation and LBC are not the only available methods for improving diagnostic accuracy: potentially interesting results were previously shown by Albright and Frost 17. Using a simple density gradient to separate atypical cells from normal cells after fixation with the Saccomanno method, the authors were able to enrich up to 20-fold the atypical and cancer cell fraction. To our knowledge however, these results have not been resumed at a later date.

A more recent study assessed the quality and cost of AutoCyte PREP versus cytocentrifugation of urine specimens in a general laboratory setting 18. It was shown that the Cytospin method, despite longer preparation time, had 1) shorter screening time, 2) higher number of diagnostic cells, 3) better fixation and staining quality than the AutoCyte PREP. Additionally, the Cytospin method was found 7 times less expensive than the AutoCyte* PREP method.

Concerning conventional methods, the values obtained in our series show that despite differences in quality, the techniques studied have no impact on the diagnostic accuracy as evaluated by the rate of abnormalities (nuclear features and cell groups). About each technique studied, the following comments may be done:

1. Smearing allows obtaining good overall results for the lowest cost. However the longer screening time renders the method suboptimal. Additionally the glycerine/albumin coating used renders slides useless for immunocytochemistry or other molecular studies,

2. Cytocentrifugation with reusable chambers should be avoided if annual renewal cannot be guaranteed,

3. Millipore filtration followed by blotting of cells on coated slides should be avoided, owing to poor global quality and high cost,

4. Cytocentrifugation using disposable chambers (Cytofunnels or Megafunnel chambers) gives excellent results equalling or surpassing LBC if one considers cellularity, fixation and the comfort for screening.

Concerning cost-efficacy comparisons, it has been shown that the monthly cost of the two most efficient methods (Cytocentrifugation with disposable chambers and Cytyc Thinprep LBC) is strikingly different: there is a 92.8% to 154.5% increased cost for LBC versus cytocentrifugation with disposable Megafunnels and Cytofunnels, respectively 19.

However in our opinion, one must consider not only the diagnostic performance and cost, but also the ultimate goal of technical improvements provided by LBC. LBC aims primarily to provide reproducible and well preserved material for additional techniques such as immunocytochemistry, fluorescence in situ hybridization (FISH) and other types of molecular analyses. It has been shown that Thinprep-processed samples allowed efficient recovery of the DNA, RNA and proteins related to the p53 tumor suppressor gene 20.

We conclude that Cytyc Thinprep LBC, despite its cost, may still be considered as a technical progress for cytology-based molecular studies. To an economical point of view and taking into account the value of a meticulous technique, cytocentrifugation with disposable chambers remains the technical standard for current treatment of urinary samples.

LBC: liquid-based cytology technique

RBC: red blood cells

RT: room temperature

TUR: transurethral resection

The author(s) declare that they have no competing interest.

EP planned the study and prepared the manuscript.

JF, KH and MCR performed the liquid-based techniques (cytocentrifugations and Thinprep processing) as well as smears and Millipore filtrations.

EP and MC performed the cytopathologic evaluations and the statistical analysis All authors read and approved the final manuscript.