Journal of the American Society of Cytopathology (2014) xx, 1e6

Available online at www.sciencedirect.com

ScienceDirect

journal homepage: www.jascyto.org/

Evidence-based adequacy criteria for urinarybladder barbotage cytology

Jennifer Prather, MDa, Brent Arville, DOa, Grazina Chatt, CT (ASCP)a,Stefan E. Pambuccian, MDa, Eva M. Wojcik, MDa,b, Marcus L. Quek, MDb,Güliz A. Barkan, MDa,b,*

aDepartment of Pathology, Loyola University Medical Center, 2160 S. First Avenue, Maywood, IllinoisbDepartment of Urology, Loyola University Medical Center, 2160 S. First Avenue, Maywood, Illinois

Received 5 April 2014; received in revised form 17 September 2014; accepted 22 September 2014

KEYWORDSUrine;Adequacy;Bladder washing;Bladder barbotage;Instrumented

*Corresponding author. Güliz A. Barkogy, Loyola University Medical Center, 2IL 60153; Tel.: 1 (708) 327-2572; fax: 1

E-mail address: gbarkan@lumc.edu (G

2213-2945/$36 � 2014 American Societyhttp://dx.doi.org/10.1016/j.jasc.2014.09.20

Introduction Adequacy criteria are well established in some areas of cytopathology to prevent false nega-tive diagnoses. To date, no such criteria have been proposed and validated for urinary tract specimens. Ouraim was to determine a cellularity cutoff point that significantly affects the sensitivity of detecting high-grade or in situ urothelial carcinoma (HGUC or UCIS) in bladder barbotage/washing specimens.Materials and methods Bladder barbotage specimens collected in liquid-based media were selected. Spec-imens diagnosed as “positive for HGUC” (with histologic confirmation) composed the study group, withnegative cases as control specimens. Samples were serially diluted and ThinPrep slides of decreasing cellu-larity were made and reviewed for diagnosis and cellularity. In a retrospective validation study, we identifiedcases with a “negative for malignancy” bladder barbotage/washing and a surgical pathology diagnosis ofUCIS or HGUC (ie, false negative cytology). Cellularity was assessed.Results A distinct difference in sensitivity was noted at a cutoff point of 2644 (20 per 10 high-powerfields) urothelial cells. Sensitivities increased for atypical or higher (68.3% versus 100%) and HGUC(43.3% versus 88.0%) after application of this cutoff point with high statistical significance (P Z 0.001and 0.0001, respectively). For the retrospective review, cases below the cutoff point were reclassified as un-satisfactory, and sensitivity rose from 76.3% to 84.8% (P Z 0.0027).Conclusions Our results indicate that, in the absence of atypical or malignant cells, an adequate bladderbarbotage specimen should have a minimum of 2644 (20 per 10 high-power fields) well-visualized, well-preserved urothelial cells to increase the positive predictive value of this test.� 2014 American Society of Cytopathology. Published by Elsevier Inc. All rights reserved.

an, MD, Department of Pathol-160 S. First Avenue, Maywood,(708)-327-2620..A. Barkan).

of Cytopathology. Published by Elsev6

Introduction

Urinary tract cytology (UTC) is widely used for the evaluationof hematuria, irritative voiding symptoms, and in the follow-

ier Inc. All rights reserved.

2 J. Prather et al.

up of urothelial carcinoma (UC).1 The continued use of UTCin the detection ofUC, despite numerous proposed alternatives(BTA [Bladder tumor antigen] STAT and BTA TRAK [Pol-ymedco, Courtlandt Manor, NY], Nuclear Matrix Protein/NMP22 [Matritech Inc., Newton, MA], Immunocytology/ImmunoCyte/uCytþ [Diagnocure, Sainte-Foy, Quebec,Canada], and others2), is due to its high specificity, rangingfrom 78% to 100%.3 Nonetheless, the reported sensitivity ofUTC for the detection of UC is low, especially in studiesincluding low-grade UCs, for which reported detection ratescan be as low as 10%.3 Even for high-grade UC (HGUC), thereported sensitivity of UTC ranges from 38% to 100%.3-5 Thislarge variation in reported sensitivities of UTC in the detectionof HGUC is most likely related to multiple factors, includingthe diagnostic criteria used for the grading of UC on histologicevaluation,6 the type of specimen used (voided urine, bladderbarbotage, etc),7 cytopreparatory method, the experience andskill of the interpreter, and the quality of the sample. Thequalityof the specimen can affect the sensitivity ofUTCandmay be, atleast in part, responsible for its low sensitivity in the detectionof HGUCs reported in some studies. However, the impact ofthe quality of the specimen on the performance of UTC cannotbe assessed in the absence of quantifiable adequacy criteria.Unfortunately, such adequacy criteria have not been defined todate.

The aim of this study was to determine what constitutesan adequate bladder washing/barbotage specimen. To reachthis goal, we determined the minimum cellularity that al-lows a cytologic diagnosis to be rendered by using a 2-stepapproach. In the first step, we serially diluted residualbladder barbotage specimens known to be positive for ma-lignancy to determine the minimum cellularity at which thecytologic preparation could still be diagnosed as abnormal(atypical urothelial cells [AUC] and above). In the secondstep of this study, we used the cellularity criteria determinedin the first step to retrospectively assess the impact on thesensitivity in a large cohort of previously diagnosed bladderbarbotage specimens. To our knowledge, this is the firstevidence-based study to determine adequacy criteria forbladder washing/barbotage cytology.

Materials and methods

Prospective determination of adequacy criteria

Nineteen bladder barbotage specimens were selected for thestudy. Cases were selected based on the cytologic diagnosis,the availability of residual material, and adequate histologicfollow-up. The study group was composed of 10 consecu-tive specimens diagnosed as “positive for HGUC” that hadhistologic confirmation as either HGUC or UC in situ(UCIS). The control group consisted of 9 consecutive casesdiagnosed as negative for UC or diagnosed as atypical orsuspicious with reactive or low-grade UC on a subsequenthistology specimen. In our institution, the bladder barbotage

specimens are collected by urologists and processed in ourlaboratory following a standard protocol that has been inplace for over 15 years.8 Briefly, vigorous irrigation of thebladder with 80 to 100 ml3 of sterile saline solution isperformed during cystoscopy and the resultant fluid is thensent to the cytology laboratory, where the volume and grossappearance of the fluid are noted. The 50 to 100 ml3 of fluidare then centrifuged for 5 to 10 minutes at 2000 rpm, thenfixed with 30 ml of CytoLyt solution (Hologic, Inc, Bed-ford, Mass), and recentrifuged for another 2 minutes at 2000rpm. The resulting cell sediment is processed using aThinPrep 3000 Processor (Hologic, Inc) and stained ac-cording to the Papanicolaou (Pap) method. On average, wereceive 1600 UTC per year in our laboratory.

The residual cellular samples left over in the containerafter the preparation of the diagnostic ThinPrep slide wereserially diluted with repetitive 2-fold dilutions in PreservCytsolution (Hologic, Inc) to achieve the desired cellularity.ThinPrep slides were prepared from each dilution using theThinPrep 3000 Processor according to the abovementionedprocedure. The slides were then reviewed by 1 of the au-thors (J.P.) to check for the presence of cellular material;acellular slides were discarded. The resulting 6 to 11 slidesper case, together with the original diagnostic slide, for atotal of 146 slides (85 from cases positive for UC, and 61from control cases), were included in the study. The slideswere deidentified, assigned a study number, and placed in arandom order into trays of 20 slides that were circulatedamong 5 observers. The 5 observers participating in thestudy were a cytotechnologist, a cytopathology fellow, and3 board-certified cytopathologists with 10 to 20 years ofexperience and interest in UTC. The observers were onlyinformed of the age, sex, and medical history of the patient,but not of the exact design of the study or of the compo-sition of the cases entered into the study. Each slide wasdiagnosed independently as negative for HGUC, AUC,suspicious for HGUC, or positive for HGUC. All diagnoseswere entered into a spreadsheet. Slides that had the samediagnosis rendered by �3 observers were considered diag-nostically concordant. Slides with discordant diagnoseswere reviewed around a multiheaded microscope to achievea consensus diagnosis.

In each slide, 1 observer counted the number of urothelialcells (excluding nonurothelial cells such as inflammatorycells and squamous cells) in 10 consecutive nonadjacent40�objective high-power fields (hpfs) on an OlympusBX40 microscope (Olympus America, Center Valley, Pa)with 10�/FN22 ocular. Counting was done along the hor-izontal diameter of the ThinPrep preparation, including thecenter of the slide, and the cell counts were averaged per 10hpfs. The formula used for calculating the cellularity of thewhole preparation was n � 1322, where n represents theaverage cell count per hpf and 1322 represents the numberof hpf per 20-mm diameter ThinPrep slide for a FN22ocular/40� objective.9,10 For cases with <10 cells per 10hpfs, the cell count was extended to 40 hpfs over the entire

Table 1 Prospective study.

Cellularity Sensitivities

AUCþ HGUC

<10 per 10 hpfs 60.5 37.2�10 per 10 hpfs 95.2 76.2P value 0.0001 0.0004<20 per 10 hpfs 68.3 43.3�20 per 10 hpfs 100.0 88.0P value 0.001 0.0001

Abbreviations: AUCþ, any diagnosis higher than AUC (including suspi-cious for high grade urothelial carcinoma and high grade urothelial car-cinoma); hpfs, high-power fields; HGUC, high grade urothelialcarcinoma.Values are percentages.

Bladder cytology adequacy criteria 3

ThinPrep slide. As large clusters (>10 cells) of benignurothelial cells can often be difficult to count and can alsoimpart a misleadingly high count, any group of benignurothelial cells with >10 cells was not included in the count.

Retrospective validation study

We searched our anatomic pathology database (Copath 6.0;Sunquest Information Systems, Tucson, Ariz) for casesdiagnosed as UCIS or HGUC on biopsies or resectionspecimens between January 1, 2001 and December 31,2012. We then identified all cases that had a bladder bar-botage/washing cytology performed within a 3-month in-terval of the surgical pathology diagnosis of UCIS orHGUC. Cases with a cytologic diagnosis of “negative formalignancy” (ie, so-called false negative diagnoses) wereidentified and their cytologic preparations were retrievedand reviewed to assess their cellularity using the methodoutlined earlier. Cases with cellularity values falling belowthe level determined in the previous phase of the study to beadequate, were reclassified as inadequate (unsatisfactory)rather than negative for malignancy, and the impact ofexcluding these cases from the sensitivity calculations wasdetermined.

Impact of adequacy criteria on bladder washing/barbotage unsatisfactory rates

To determine the impact of the quantitative adequacycriteria determined by the previously described prospectiveand retrospective studies, 100 consecutive bladder washing/barbotage specimens accessioned starting from January 1,2012 were retrieved and reviewed by 2 pathologists. Caseswith cellularities below the cutoff point determined wereconsidered inadequate/nondiagnostic.

The nondiagnostic rate obtained was then compared tothe laboratory’s current nondiagnostic rate of bladderwashing/barbotage rate of 1.5%.

Statistical analysis

Using the histopathologic diagnosis as a gold standard, 2 �2 contingency tables were made and sensitivity was calcu-lated using the standard formula (number of true positivesdivided by the sum of the number of true positives andnumber of false negatives). Using free calculators (Quick-Calcs; GraphPad Software, Inc, La Jolla, Calif), Fisher exacttest with a 2-tailed P value was used to determine the sta-tistical significance of the observed changes in sensitivity.A P value of < 0.05 was considered significant.

Results

Of a total of 85 UC slides, 19 (22.4%) received a consensusdiagnosis of “negative,” 15 (17.6%) of AUC, 3 (3.5%) of

suspicious for HGUC, and 48 (56.5%) of positive forHGUC. To achieve a consensus diagnosis, 23 of 85 had tobe reviewed over a multiheaded microscope and discussed.In the remaining 62 cases, there was 3-way agreement in 32cases, 4-way agreement in 21, and 5-way agreement in9. Table 1 outlines the sensitivity of the consensus diagnosisof both AUC or higher (AUCþ) and HGUC.

The data was analyzed for 2 cutoff points: 10 and 20urothelial cells per 10 hpfs of which, the second (20 uro-thelial cells per 10 HPFs, corresponding to 2644 totalurothelial cells per ThinPrep slide) appeared to be clini-cally useful with a sensitivity of 100% at the threshold ofAUC.

In the retrospective review, a total of 431 cases had asurgical biopsy/resection specimen with the diagnosis ofHGUC or UCIS and a bladder barbotage specimen within a3-month time span. The original cytology diagnosesrendered on these specimens were as follows: 102 nega-tive, 57 AUC, 54 suspicious for HGUC, and 218 HGUC.The AUC, suspicious for HGUC, and HGUC casescombine to total 329 AUCþ cases. In our institution,during the 12-year span covered by this study (2001-2012),the sensitivity of urine barbotage specimens for detectingUC was 76.3% using AUCþ as a threshold and 50.6%using positive for HGUC as a threshold. Of the 102 urinarycytology cases that were originally “false negative” (ie,were interpreted as negative for malignancy), 69 wereavailable for review. Of these 69 cases, 18 (26%) and 30(43%) were considered inadequate/nondiagnostic using the10 and 20 cells per 10 hpfs cutoff points, respectively.After removing cases with cellularities below the cutoffvalue of 20 urothelial cells per 10 hpfs (considered inad-equate/unsatisfactory/nondiagnostic based on the pro-spective part of our study) from further consideration, werecalculated the sensitivity of UTC (Table 2). The sensi-tivity rose from 76.3% to 84.8% using AUCþ as athreshold and from 50.6% to 56.2% using positive forHGUC as a threshold. The increase in sensitivity is sta-tistically significant only when using the threshold ofAUCþ (P Z 0.0027).

Table 2 Retrospective review.

Sensitivities

AUCþ HGUC

Original sensitivity 76.3 50.6Using 20/10 hpfs 84.8 56.2P value 0.0027 0.12

Values are percentages.Abbreviations as in Table 1.

4 J. Prather et al.

Discussion

In this study, we found that a cutoff value of 20 urothelial cellsper 10 hpfs showed a statistically significant increase insensitivity for the cytologic diagnosis of UC in both phases ofour studydthe prospective dilutional study and the retro-spective review. This suggests that an adequate bladder bar-botage specimen should have�20 urothelial cells per 10 hpfswhen processed by the ThinPrep method. The applicability ofthis cellularity threshold to other urinary cytopreparatorymethods (filter, cytospin, SurePath [BD, Franklin Lakes,N.J.], or other liquid-based preparations) requires furtherstudy.Of note, none of the cases included in the study sufferedfrom previously described limiting factors such as presence ofabundant lubricant, abundant inflammation, or numerous redblood cells. Because the presence of any of these factors mayaffect the specimen’s cellularity, by adhering to and cloggingthe ThinPrep membrane and therefore interfering with thetransfer of the cells onto the slide, as well as the visualizationof the cells that are present on the slide, they should beregarded as additional adequacy criteria.

We did not address the adequacy of voided urine sam-ples. These frequently hypocellular samples may havedifferent adequacy requirements. The cellularity of voidedurine specimens depend on a variety of factors, includingthe type of voided urine (first-morning voided urine speci-mens tend to be more cellular than random voided urinespecimens), the pathologic process (infections, mechanicalirritation by catheters or stones, and urothelial neoplasiamay result in increased exfoliation) and potentially thevolume of urine from which the cytologic slide is made. Forvoided specimens, it remains difficult to determine what anoptimal cellularity is because no evidence-based studieshave been performed on such specimens to date. In addition,because it may not be possible to make modifications to thecollection that would increase the cellularity, the value oflabeling such specimens as inadequate is debatable.

Compared with voided urine specimens, bladderwashing/barbotage specimens have a higher cellular yieldand lack contamination from nonurothelial cells such asvaginal squamous cells, which may be responsible for theirhigher sensitivity.11

Earlier studies on UTC were done using different cyto-preparatory methods, including direct sediment smears,

filter, and cytospin preparations. As early as 1974, Morseand Melamed12 found widely variable numbers of bothurothelial cells and cancer cells in voided urine specimensand suggested the need for further studies to assess theminimum quantity of urine and number of urothelial cellsnecessary for diagnosis. Unfortunately, such studies werenot performed, possibly due to the difficulty of obtainingaccurate counts of cells in slides prepared as sedimentsmears, filter, and cytospin slides, which frequently showthick, overlapping cellular areas, poor preservation, andobscuring inflammation or blood.4

Currently, UTC specimens are processed using liquid-based preparations (LBP) in most laboratories. The stan-dardized processing of LBP usually results in slides withrelatively uniformly distributed cells, higher cellularity, andimproved preservation of cellular morphology, all of whichcan lead to easier and less time-consuming screening. Theimproved visualization of the cells in LBP makes assessingadequacy by counting urothelial cells easier and mayimprove the sensitivity of UTC, when using AUCþ as adiagnostic threshold.13 However, to our knowledge, a spe-cific number of urothelial cells required to further reduce thefalse negative rate of UTC has not been determined to date.Bastacky et al14 reviewed the slides of a subset of thecytohistologic discrepant cases found in their study forspecimen adequacy and considered specimens as subopti-mal if they had <15 intermediate or basal urothelial cells,obscuring blood/inflammation, and/or poor cellular preser-vation.14 Applying these criteria, the UTC specimen from18 of 31 cases (58%) with false negative cytology wereconsidered suboptimal.14 The adequacy criteria proposed byBastacky et al14 are cited by the 2004 Papanicolaou Societyof Cytopathology Practice Guidelines Task Force Recom-mendations.15 It is worth mentioning, however, that themajority of urinary specimens in the Bastacky et al14 studywere voided urine samples (72%), and that the investigatorsdid not report the exact mechanism by which this value (15intermediate or basal urothelial cells) was obtained. It istherefore difficult to compare this number, which may beuseful in determining the adequacy of voided urine speci-mens, to the much higher number of urothelial cells (2644)that we found to be required for adequate bladder barbotage/washing specimens.

Our study has several limitations. First, although wedetermined the adequate cellularity cutoff for bladderwashing/barbotage specimens through an experimentalapproach using serial dilutions, we have applied this ade-quacy criterion retrospectively rather than prospectively toclinical samples to determine its impact on sensitivity. Asfor any study including a retrospective review of cases,prospective confirmation of the validity of this criterion isneeded. Second, all cell counts were performed by a singleobserver, which did not allow us to assess the magnitude ofthe interobserver variability of such counts. Previous studieshave shown that the cellularity assessment is good (“nearexcellent”) in cervical ThinPrep samples,16 but additional

Bladder cytology adequacy criteria 5

studies may be needed to confirm this good reproducibilityfor UTC samples. Third, as mentioned above, this studyonly addressed the adequacy of instrumented UTC (bladderwashing/barbotage specimens) processed by the ThinPrepmethod. Therefore the adequacy criteria derived from thisstudy may be not be applicable to voided urine specimensand may need to be modified for bladder washing/barbotagespecimens UTC prepared with cytopreparatory methodsother than the ThinPrep method. Cytologic preparations,such as cytocentrifugation specimens,17 SurePath, and otherLBP, have different sizes of areas covered by cellular ma-terial, may have different cell density, and may be affecteddifferently by the presence of excessive mucus,18 blood,19

and lubricant20 in the sample. Fourth, we have onlyincluded urothelial cells in our cell counts; the exclusion ofsquamous cells from the cell counts may affect rare caseswith extensive squamous metaplasia, in which sufficientnumbers of urothelial cells may not be present. In thesesituations cystoscopic/clinical correlation is needed. Ofnote, technical factors such as operator of the cystoscope,method of performing the barbotage/washing (ie, How“vigorous” was the irrigation performed? How close was thescope to the mucosa when the washing is being done?) mayinfluence the specimen collection.

Quantitative adequacy criteria have been established fora number of cytologic specimens including Pap tests21 andthyroid aspirates.22 The study of the effect of cellularity oncervical cytology sensitivity was performed using serialdilutions,21 whereas the thyroid adequacy criteria weredetermined by retrospectively reviewing ThinPrep slidesfrom thyroid aspirates in cases with surgical excisionfollow-up.22 Herein we applied both methods to bladderwash specimens in order to establish more robust adequacycriteria. Quantitative adequacy criteria are used to excludeunsatisfactory specimens from further diagnostic consider-ations and have been found to be useful to reduce thenumber of false negative cytology results, increase thesensitivity of the test, and increase the confidence of clini-cians and patients in the test. Therefore, these criteria, whichare relatively easy to apply and are relatively reproducible(especially when accompanied by cellularity reference im-ages23), have been included in the standardized Bethesdareporting criteria for Pap tests and thyroid fine-needleaspirates.9,24

The cell counting method that we have used in this studywas similar to that recommended by the Bethesda Systemfor cervical liquid-based specimens: we counted cells inminimum of 10 fields of view (including the center of theslide) with a 40� objective along the horizontal diameter ofthe slide. A recent article found that more accurate cellu-larity results can be obtained with a modification of thiscounting method, the so-called SKML (Stichting Kwali-teitsbewaking Medische Laboratoriumdiagnostiek) protocol,in which 10 microscope fields along the 2 axes of the slide(5 along the horizontal and 5 along the vertical diameter ofthe slide) are counted.10 Further studies are needed to

determine whether this modification of the cell countingmethod offers significant advantages for UTC sample ade-quacy assessment.

The clinical impact of implementing these adequacycriteria on bladder washing/barbotage specimens is that apercentage of specimens will be considered unsatisfactory(or possibly satisfactory but limited by low cellularity).Based on our data, this percentage would be around 5%.Currently, without defined adequacy criteria, about 1.5%of our bladder washing/barbotage specimens are diagnosedas inadequate/nondiagnostic due to extremely low cellu-larity or obscuring factors. The application of the proposedquantitative adequacy criteria would therefore result in arather large increase in unsatisfactory rates. It may beuseful to define 2 thresholds of adequacy: satisfactory butlimited by low cellularity when the sample contains be-tween 10 and 20 urothelial cells per 10 hpfs and unsatis-factory/nondiagnostic when it contains <10 urothelial cellsper 10 hpfs.

This increase in unsatisfactory/nondiagnostic rates posessome clinical management problems because it would raisethe question what the further management of cases diagnosedas inadequate/nondiagnostic using the proposed adequacycriteria would be. This clearly would be a clinical judgmentcall that the urologist has to make, based on the patient’shistory, cystoscopic findings, and concomitant histopatho-logic findings. In most cases, in patients with low likelihoodof malignancy, such as patients undergoing work-up for he-maturia, who have benign cystoscopic findings or possiblyhave another identifiable reason for hematuria, it would bereasonable to forgo a repeat cystoscopy. On the other hand, ifthe patient has a history of UCIS or HGUC, has suspiciouscystoscopy findings, but a nondiagnostic biopsy (“denudedurothelium”), a repeat cystoscopy or more timely surveillancemay be appropriate. We hope that reports including assess-ments of adequacy (cellularity) of bladder barbotage/washingspecimens will give constant feedback to urologists and willhelp modify their collection technique to increase cellularyield and ultimately reduce the unsatisfactory rates. Similarresults have been achieved with well-defined adequacycriteria used for Pap tests that resulted in better specimenquality and rather low inadequate rates through constantfeedback given to the clinicians.

Conclusions

Our results indicate that, in the absence of atypical or ma-lignant cells, an adequate bladder barbotage specimenshould have a minimum of 2644 (20 per 10 hpfs) well-visualized, well-preserved urothelial cells to increase thesensitivity of this test.

Funding sources

No specific funding was disclose.

6 J. Prather et al.

Conflict of interest disclosures

The authors made no disclosures.

Acknowledgments

The authors thank Elizabeth Sheil and Maria Vivo for theirtechnical expertise in preparing the study material.

References

1. Barkan GA, Wojcik EM. Genitourinary cytopathology (kidney and uri-nary tract). Cancer Treat Res. 2014;160:149e183.

2. Caraway NP, Katz RL. A review on the current state of urine cytologyemphasizing the role of fluorescence in situ hybridization as an adjunctto diagnosis. Cancer Cytopathol. 2010;118:175e183.

3. Yafi FA, Brimo F, Auger M, Aprikian A, Tanguay S, Kassouf W. Isthe performance of urinary cytology as high as reported historically?A contemporary analysis in the detection and surveillance of bladdercancer. Urol Oncol. 2014;32:27.e1e27.e6.

4. Hoda RS. Non-gynecologic cytology on liquid-based preparations: amorphologic review of facts and artifacts. Diagn Cytopathol. 2007;35:621e634.

5. Konety BR, Metro MJ, Melham MF, Salup RR. Diagnostic value ofvoided urine and bladder barbotage cytology in detecting transitionalcell carcinoma of the urinary tract. Urol Int. 1999;62:26e30.

6. Curry JL, Wojcik EM. The effects of the current World Health Orga-nization/International Society of Urologic Pathologists bladderneoplasm classification system on urine cytology results. Cancer.2002;96:140e145.

7. Hundley AF, Maygarden S, Wu JM, Visco AG, Connolly A. Adequacyof urine cytology specimens: an assessment of collection techniques.Int Urogynecol J Pelvic Floor Dysfunct. 2007;18:997e1001.

8. Bakhos R, Shankey TV, Flanigan RC, Fisher S, Wojcik EM. Compar-ative analysis of DNA flow cytometry and cytology of bladder wash-ings: review of discordant cases. Diagn Cytopathol. 2000;22:65e69.

9. Solomon D, Nayar R. The Bethesda System for Reporting CervicalCytology: Definitions, Criteria, and Explanatory Notes. 2nd ed. NewYork, NY: Springer; 2004.

10. Siebers AG, van der Laak JA, Huberts-Manders R, Vedder JE,Bulten J. Accurate assessment of cell density in low cellular liquid-based cervical cytology. Cytopathology. 2013;24:216e221.

11. Murphy WM, Crabtree WN, Jukkola AF, Soloway MS. The diagnosticvalue of urine versus bladder washing in patients with bladder cancer. JUrol. 1981;126:320e322.

12. Morse N, Melamed MR. Differential counts of cell populations in uri-nary sediment smears from patients with primary epidermoid carci-noma of bladder. Acta Cytol. 1974;18:312e315.

13. Voss JS, Kipp BR, Krueger AK, et al. Changes in specimen prepara-tion method may impact urine cytologic evaluation. Am J Clin Pathol.2008;130:428e433.

14. Bastacky S, Ibrahim S, Wilczynski SP, Murphy WM. The accuracy ofurinary cytology in daily practice. Cancer. 1999;87:118e128.

15. Layfield LJ, Elsheikh TM, Fili A, Nayar R, Shidham V, PapanicolaouSociety of Cytopathology. Review of the state of the art and recom-mendations of the Papanicolaou Society of Cytopathology for urinarycytology procedures and reporting: the Papanicolaou Society ofCytopathology Practice Guidelines Task Force. Diagn Cytopathol.2004;30:24e30.

16. Haroon S, Samayoa L, Witzke D, Davey D. Reproducibility of cervi-covaginal ThinPrep cellularity assessment. Diagn Cytopathol. 2002;26:19e21.

17. Piaton E, Faynel J, Hutin K, Ranchin MC, Cottier M. Conventionalliquid-based techniques versus Cytyc Thinprep processing of urinarysamples: a qualitative approach. BMC Clin Pathol. 2005;5:9.

18. Kenyon S, Sweeney BJ, Happel J, Marchilli GE, Weinstein B,Schneider D. Comparison of BD Surepath and ThinPrep Pap systemsin the processing of mucus-rich specimens. Cancer Cytopathol.2010;118:244e249.

19. Sweeney BJ, Haq Z, Happel JF, Weinstein B, Schneider D. Compari-son of the effectiveness of two liquid-based Papanicolaou systems inthe handling of adverse limiting factors, such as excessive blood. Can-cer. 2006;108:27e31.

20. Holton T, Smith D, Terry M, Madgwick A, Levine T. The effect oflubricant contamination on ThinPrep (Cytyc) cervical cytologyliquid-based preparations. Cytopathology. 2008;19:236e243.

21. Studeman KD, Ioffe OB, Puszkiewicz J, Sauvegeot J, Henry MR. Ef-fect of cellularity on the sensitivity of detecting squamous lesions inliquid-based cervical cytology. Acta Cytol. 2003;47:605e610.

22. Michael CW, Pang Y, Pu RT, Hasteh F, Griffith KA. Cellular adequacyfor thyroid aspirates prepared by ThinPrep: how many cells areneeded? Diagn Cytopathol. 2007;35:792e797.

23. Moore D, Pugh-Cain D, Walker T. Cervical smear adequacy: cellu-larity references were found to increase both interobserver agreementand unsatisfactory rate. Cytopathology. 2009;20:161e168.

24. Ali SZ, Cibas ES. The Bethesda System for Reporting Thyroid Cytopa-thology: Definitions, Criteria, and Explanatory Notes. 1st Edition. NewYork, NY: Springer; 2010.