Prospective Validation of the Clinical Usefulness of Reflex Fluorescence In Situ Hybridization Assay

Prospective Validation of the Clinical Usefulness of Reflex Fluorescence In Situ Hybridization Assay in Patients With Atypical Cytology for the Detection of Urothelial Carcinomaof the BladderBruce J.Schlomer,Richard Ho,Arthur Sagalowsky,*Raheela Ashfaq†and Yair Lotan‡From the Departments of Urology(BJS,RH,AS,YL)and Pathology(RA),University of Texas Southwestern Medical Center,Dallas,TexasAbbreviationsand AcronymsFISHϭfluorescence in situ hybridizationNPVϭnegative predictive value PPVϭpositive predictive value RCCϭrenal cell carcinomaUCϭurothelial carcinomaSubmitted for publication May20,2009.Study received institutional review board ap-proval.Supported by a grant from Abbott,Inc.*Financial interest and/or other relationship with The Journal of Urology,AUA-South Central Section Board of Directors and Bioniche.†Financial interest and/or other relationship with Ikonisys,MTM,Prediction Science,CSI and Caris.‡Correspondence:UT Southwestern Depart-ment of Urology,Moss Building,8th Floor,Suite 112,5323Harry Hines Blvd.,Dallas,Texas75390-9110(telephone:214-648-0483;FAX:214-648-8786;e-mail:Yair.Lotan@). See Editorial on page11.For another article on a related topic see page339.Purpose:We previously prospectively evaluated the clinical usefulness of aflu-orescence in situ hybridization assay for treating patients with atypical cytology results at risk for bladder cancer.This study is a prospective validation the usefulness offluorescence in situ hybridization in this setting.Materials and Methods:Between June2007and January2009every patient who underwent cystoscopy and cytology with atypical cytology underwent a reflex UroVysion®test.A comprehensive review was then performed to evaluate clin-ical and pathological data on each patient.Results:The study population comprised108patients with no history of cancer and108who underwent cystoscopy for cancer surveillance.In patients with cystoscopically visualized lesions UroVysion had a positive predictive value of 100%but there were false-negative results.In patients with equivocal cystoscopy and a history of cancer all4high grade tumors were detected and there were no false-negativefindings.In patients with equivocal cystoscopy and no prior cancer the positive predictive value was100%and there were no false-negative results. In patients with negative cystoscopy the UroVysion test detected all cancers but the positive predictive value was10%and29%in patients with and without a history of cancer,respectively.Conclusions:This prospective study of a reflexfluorescence in situ hybridization assay in patients with atypical cytology validates our previousfindings.In patients with atypical cytology and obvious tumor on cystoscopy the assay was unnecessary but it was beneficial in those with equivocal or negative cystoscopy results.Thefluorescence in situ hybridization assay identified all urothelial carcinoma tumors in patients with equivocal or negative cystoscopy.In patients with equivocal or negative cystoscopy and atypical cytology,a reflexfluorescence in situ hybridization assay may help avoid unnecessary evaluation while identi-fying those who would need further evaluation.Key Words:in situ hybridization,fluorescence;cytology;carcinoma,transitional cell;urothelium;urinary bladderB LADDER cancer is the4th most com-mon cancer in men and5th most com-mon cancer overall.1Cystoscopy andcytology are the current tests mostcommonly used for the detection andsurveillance of urothelial carcinoma.Cystoscopy has a false-negative rateof up to30%.2,3In addition,erythem-atous or suspicious areas in the blad-der on cystoscopy are often not asso- 0022-5347/10/1831-0062/0Vol.183,62-67,January2010 THE JOURNAL OF UROLOGY®Printed in U.S.A. Copyright©2010by A MERICAN U ROLOGICAL A SSOCIATION DOI:10.1016/j.juro.2009.08.157ciated with malignancy especially in patients with no history of cancer.4,5Cytology often is used as an adjunct to cystoscopy to help detect carcinoma in situ and upper tract lesions.Unfortunately cytology has a poor sensitivity for low grade tumors and frequently the results are inconclusive for malignancy.6–8A cy-tology report with atypicalfindings creates a dilemma for clinicians and patients,especially for patients with negative cystoscopy,suspicious cystoscopy or a history of bladder cancer.The clinician has an option to ob-serve the patient with the potential for missed cancer or to biopsy every patient with the attendant risks of anesthesia and potential bladder injury.Multiple urine based bladder markers have been evaluated to see if they assist in detecting bladder cancer.9Currently markers are not recommended by a recent panel of experts to assist in the detection of urothelial carcinoma.10A problem with most of the urine based markers that have been studied is a lack of validation after initial reports and the absence of studies to determine their additive role in clinical decision making.UroVysion is a multitarget FISH assay that de-tects aneuploidy of chromosomes3,7and17,and loss of the9p21band in cells in urine specimens.11 The FISH assay has demonstrated improved sensi-tivity compared to cytology for detecting UC but it has a lower specificity.11,12To improve the under-standing of the role of urine based tumor markers in clinical practice,prospective studies are needed with subsequent validation.We previously hypothesized that a reflex FISH assay could help identify patients with atypical cytology who harbor cancer.13In that study we implemented a prospective protocol in which every patient with an atypical cytology result underwent a reflex FISH assay.The study in120 patients with atypical cytology showed that UroVysion was useful in patients with equivocal or negative cys-toscopy but unnecessary in those with an obvious tu-mor on cystoscopy.This study is a prospective validation of our pre-vious study.A reflex UroVysion test was used in all patients with atypical cytology results during a19-month period.We evaluated the sensitivity,specific-ity,PPV and NPV based on the presence or absence of bladder cancer history and cystoscopicfindings. MATERIALS AND METHODSInstitutional review board approval was obtained for this study.Between June2007and January2009every pa-tient with an atypical cytology result underwent a reflex FISH assay.All cytology specimens were interpreted by an experienced cytopathologist.A cytology result was con-sidered atypical if it was not unequivocally positive or negative.Thefigure shows atypical cells with correspond-ing FISH results.Cystoscopy results were recorded as positive if an obvious tumor could be seen,equivocal if an erythematous or suspicious area was described but no obvious tumor was seen,and negative if cystoscopy showed normal mucosa.A comprehensive review of the clinical and pathological information was then performed on each patient.Clinical stage was assigned by the oper-ative surgeon according to the1997TNM system and tumor grade was assigned according to the2003WHO guidelines.The presence of UC was confirmed by surgical pathology from transurethral resection specimen but the absence of UC was not confirmed by biopsy in all patients. Absence of UC after initial cystoscopy was assessed by1or more methods such as negative history of gross hematu-ria,negative urinalysis,negative cytology and/or FISH, negative cystoscopy and negative bladder biopsy.In terms of sample collection and preparation voided or barbotage urine was collected in a sterile container with nofixative of50%alcohol.One ThinPrep®slide was pre-pared using a ThinPrep2000processor.Residual material was refrigerated and retained for the reflex FISH assay. The slide was stained using a Papanicolaou stain and interpreted by a trained cytopathologist.Reflex FISH as-say was performed on the residual specimen when the cytology evaluation was atypical.The cytopathologist was blinded to the cystoscopy and biopsy results.The prepara-tion of the ThinPrep slide and procedure for the UroVysion FISH assay were described previously.13RESULTSThere were216patients with atypical cytology dur-ing the study period including those with(108)and without(108)a history of UC.The overall average age of the216patients was65.7years(range29.8to 95.8)and175(81.0%)were men.For patients with-out a history of cancer cystoscopy was performed due to hematuria(68,63.0%),bladder mass on imaging (18,16.7%),irritative voiding(6,5.6%),renal mass (3,2.8%),urinary tract infection(2,1.9%),prostate cancer(1,0.9%)and none reported(10,9.3%).The patients were then grouped by whether there was a papillary or solid lesion,an equivocal lesion(red lesion or patch)or no lesion on cystoscopy.Average followup for patients with and without a history of UC was11.2(range0to22.5)and6.2months(range 0to22.9),respectively.Followup cystoscopy was performed in69%of the patients with a history of UC and22%of those without a history of UC.In patients with a history of UC at least1followup cystoscopy was performed in87,4underwent cys-tectomy,2died and15were lost to followup.In patients without a history of UC at least1followup cystoscopy was performed in29,3underwent cys-tectomy,3died and23were lost to followup.There were45patients without a history of cancer who had a negative initial evaluation and no followup cystos-copy.These patients were seen in clinic,and had a negative history for gross hematuria(33),negative urinalysis(9)and negative cytology(3).REFLEX FLUORESCENCE IN SITU HYBRIDIZATION ASSAY AND ATYPICAL CYTOLOGY63FISH and History of CancerTable 1shows the outcomes of FISH and cystoscopy in patients with a history of cancer.Of the 108patients with a history of cancer and atypical cytol-ogy 23had positive cystoscopy,18had equivocal cystoscopy and 67had negative cystoscopy.All pa-tients with a positive cystoscopy result had cancer.There were 5patients with a positive cystoscopy andfalse-negative FISH.In patients with equivocal le-sions or negative cystoscopy the FISH assay was positive in all high grade tumors with a PPV of 50%and 10%,respectively.There were no false-negative findings.Table 2shows the sensitivity,specificity,PPV and NPV of the UroVysion assay in patients with a his-tory of cancer categorized by cystoscopicfindings.Examples of 2cases referred per reflex protocol for atypical urine cytology.A ,atypical cytology.B ,positive UroVysion test of atypical cells in part A .C ,atypical cytology.D ,negative UroVysion test of atypical cells in part C .Table 1.FISH outcomes in patients with cancer historyNo.PtsNo.Ca (%)No.Ta Low GradeNo.Ta High GradeNo.TIS No.T1High GradeNo.T2or GreaterNo.Neg Biopsy No.OtherPapillary/nodular lesion:Pos 1414(100)2822200Neg55(100)3010100Borderline 22(100)1100000Uninformative 22(100)02*00000Equivocal/red lesion:Pos 84(50)0130010Neg600000000Borderline 200000000Uninformative 200000000Neg cystoscopy:Pos 202(10)0100041(upper tract UC)Neg3900000010Borderline 700000010Uninformative1*Of these patients 1had initial uninformative biopsy due to small amount of tissue but on followup biopsy had Ta high grade UC.REFLEX FLUORESCENCE IN SITU HYBRIDIZATION ASSAY AND ATYPICAL CYTOLOGY64Uninformative (5)or borderline FISH (11)results were excluded from analysis.The sensitivity and NPV were 100%for patients with equivocal or neg-ative cystoscopy.FISH Without History of CancerTable 3shows the outcomes of FISH and cystoscopy in patients with no history of cancer.Of the 108patients without a history of cancer and atypical cytology 20had positive cystoscopy,17had equivo-cal cystoscopy and 71had negative cystoscopy.In the patients with positive cystoscopy FISH detected 8tumors but there were 7false-negative results including 4high grade tumors.There were also 2borderline and 1uninformative results in patients with high grade or invasive disease.In patients with equivocal lesions the FISH assay detected both patients with high grade cancers.There were 2patients with RCC with negative re-sults on FISH but there were no false-negative FISH results in patients with urothelial carcinoma.In patients with a negative cystoscopy both had urothe-lial carcinoma detected with 5false-positive results.There were no false-negative FISH results in pa-tients with a negative cystoscopy but there was 1patient who had clear cell RCC.Table 4shows the sensitivity,specificity,PPV and NPV of FISH in patients without a history of cancer categorized by cystoscopic findings.Uninformative (4)or borderline (7)FISH results were excluded from analysis as well as patients with a cancer other than urothelial or prostate.The sensitivity and NPV were 100%for patients with an equivocal or nega-tive cystoscopy.Because of 6false-negative results (1patient had metastatic breast cancer)the sensi-tivity and NPV in patients with positive cystoscopy were poor at 57%and 25%,respectively.The PPV for patients with equivocal cystoscopy was 100%but was only 29%for those with negative cystoscopy due to 5false-positive results.DISCUSSIONThis study was a prospective validation of whether a reflex FISH assay would improve the clinical treat-ment and decisions of patients with atypical cytology.In the prior study the FISH assay was unnecessary in patients with obvious tumors on cystoscopy but was beneficial in those with equivocal or negative cystos-copy results.13In that study the FISH assay identified all high grade cancer in patients with equivocal or negative cystoscopy and atypical cytology while po-Table 2.FISH performance characteristics in patients with cancer historyPosEquivocal NegNo.pts191459No./total No.(%):Sensitivity 14/19(73.6)4/4(100)2/2(100)Specificity 0/06/10(60)39/57(68.4)PPV 14/14(100)4/8(50)2/20(10)NPV0/56/6(100)39/39(100)Patients with borderline and uninformative FISH results,and those with nonu-rothelial or nonprostate cancer were excluded from analysis.Table 3.FISH outcomes in patients with no cancer historyNo.PtsNo.Ca (%)No.Ta Low GradeNo.Ta High GradeNo.TIS No.T1High GradeNo.T2or GreaterNo.Neg Biopsy No.Other Papillary/nodular lesion:Pos 88(100)1202201Prostate CaNeg97(77.8)2400021Metastatic breast Ca Borderline 22(100)0000101High grade urachal Ca Uninformative 11(100)0001000Equivocal/red lesion:Pos 22(100)0110000Neg131(7.7)0000081Clear cell RCC Borderline 11(100)0000001Clear cell RCC Uninformative 100000000Neg cystoscopy:Pos 72(28.6)0011000Neg581(1.7)0000031Clear cell RCC Borderline 400000010Uninformative2Table 4.FISH performance characteristics in patients with no cancer historyPosEquivocal NegNo.pts171565No./total No.(%):Sensitivity 8/14(57.1)2/2(100)2/2(100)Specificity 2/2(100)12/12(100)57/62(91.9)PPV 8/8(100)2/2(100)2/7(28.6)NPV2/8(25)12/12(100)57/57(100)Patients with borderline and uninformative FISH results,and those with non-urothelial cancer were excluded from analysis.REFLEX FLUORESCENCE IN SITU HYBRIDIZATION ASSAY AND ATYPICAL CYTOLOGY 65tentially avoiding biopsy in patients with a negative FISH result.In this study we confirmed that UroVysion is not necessary or beneficial in patients with a visualized tumor because it would not change the treatment as all patients would undergo biopsy or transurethral resection.On the other hand the FISH assay was useful in patients with equivocal lesions on cystoscopy as all6with UC had a positive FISH result and there were no false-negative FISH results.In the19patients with equivocal cystoscopy and negative FISH assay no UC were diagnosed suggesting that a significant number of bladder bi-opsies can be safely avoided in these patients.How-ever,because there were some patients who had high grade visible tumors and a negative FISH,a negative FISH does not exclude the possibility of high grade UC.In our previous study only1case of low grade Ta UC would have been missed in pa-tients with equivocal cystoscopy and negative FISH results.In patients with a negative cystoscopy all4with UC had a positive FISH result,including1with high grade upper tract UC.There were no false-negative results.There was a significant number of false-positive results in patients with negative cystoscopy and overall only4of27(15%)with a positive FISH and negative cystoscopy were diagnosed with UC.It was also reassuring that there were no false-nega-tive results.The results of this validation study along with our initial study suggest that a reflex FISH assay is not needed in patients with a visualized lesion on cys-toscopy but that it can be useful in those with equiv-ocal or negative cystoscopy.With equivocal cystos-copy patients with a negative FISH result can safely avoid a biopsy and have followup cystoscopy instead. In patients with negative cystoscopy and positive FISH there was a15%risk offinding high grade UC. Therefore,a biopsy and upper tract study should be strongly considered.While there are several urine based tumor mark-ers for the detection and surveillance of bladder cancer,they are not currently recommended by guideline panels.10The European Association of Urology guidelines on nonmuscle invasive bladder cancer noted that it is unclear whether tests offer additional information for the detection and man-agement of nonmuscle invasive bladder cancer.14 One of the main reasons that markers are not rec-ommended by guidelines panels is the absence of studies identifying the added benefit of markers in clinical scenarios.Furthermore,to our knowledge there are no validation studies evaluating the role of markers in the detection or surveillance of bladder UC.Most previous studies evaluating common markers have focused on optimizing sensitivity and specificity.While markers have an improved sensi-tivity compared to cytology they also have a lower specificity.9,11,12,15The consequence of sensitivity in the50%to80%range is the possibility of missed cancer,and specificity that ranges from70%to85% results in false-positive outcomes.Thus,clinicians are uncertain about the meaning of positive and negative marker results.There is a need for studies that identify clinical scenarios in which markers assist clinicians in decision making.Our initial study and the current validation suggest that pa-tients with atypical cytology benefit from a reflex UroVysion FISH assay.While the test has a consid-erable cost,the ability to avoid bladder biopsies in cases of equivocal lesions and the early detection of high grade cancer justify this added cost.Prior studies have evaluated urine based tumor markers in the setting of atypical or negative cytol-ogy.16,17Skacel et al retrospectively evaluated the UroVysion FISH assay in120urine specimens,and found a sensitivity of100%,89%and60%in patients with suspicious,atypical and negative cytology,re-spectively.16Ferra et al retrospectively evaluated the UroVysion FISH assay in160urine samples with suspicious cytology,and found a sensitivity of 68%,specificity of40%,PPV of57%and NPV of 52%.18Unlike our study this group excluded atypi-cal cytology from analysis and did not stratify re-sults based on cystoscopicfindings or history of can-cer.All patients did have followup cystoscopy while in our study this was not the case,which could explain some of the difference in NPV as some can-cers may have been missed in our study since fol-lowup cystoscopy was not performed in every pa-tient.The BTA stat®test has also been evaluated in patients with suspicious cytology.17The study did not stratify patients by cystoscopicfindings.Fur-thermore,the false-negative rate was35.7%(19of 28),which is a high rate of missed cancer.Yoder et al evaluated the prospective use of a reflex FISH pro-tocol in250patients with negative or equivocal cy-tology with a focus on anticipatory positives.19This study focused on anticipatory positive results and approximately27%of patients under bladder carci-noma surveillance with negative cystoscopy had a positive FISH result.In approximately65%of this anticipatory positive group recurrent bladder urothe-lial carcinoma developed within29months.A limitation of our study was that we did not have long-term followup in our cohort,and followup cys-toscopy was performed in only22%of patients with no history of cancer and69%of those with a history of cancer.Thus,the low PPV in patients with nega-tive cystoscopy could represent those in whom can-cer will develop in the future.This study was not designed to evaluate anticipatory positives and was designed to determine how a clinician should ap-proach the patient at testing.Future studies willREFLEX FLUORESCENCE IN SITU HYBRIDIZATION ASSAY AND ATYPICAL CYTOLOGY 66evaluate the issue of false vs anticipatory positives further.There were9uninformative and18borderline results in216patients.The uninformative result occurs when there are insufficient cells to run the assay accurately,which is a limitation of the reflex protocol and the assay itself.In the setting of an uninformative result clinicians have the option of repeating a FISH assay with a fresh urine sample if they believe that the added information of a positive or negative result would influence treatment.Bor-derline results were frequently associated with can-cer but we did not include them in the analysis of sensitivity,specificity,PPV and NPV because it would introduce a bias to assume these were posi-tive or negative results.Again it seems most reason-able to repeat the test in patients with negative or equivocal cystoscopy.The test is not needed in pos-itive cystoscopy.Afinal consideration is that this study was per-formed in patients with atypical cytology.There is also likely variation in how pathologists define atyp-ical cytology.Our pathologists rarely label cells as suspicious and,thus,atypical cells and suspicious cells are categorized together.Other institutions may more rigorously distinguish patients with atyp-ical and suspicious cells.18This could change the prevalence of bladder cancer in the population and as a result will change the performance of the test. On the other hand,neither atypical nor suspicious cells exclude malignancy and,as such,require a decision by the clinician whether to evaluate or ig-nore the outcome of cytology.Due to the high spec-ificity of cytology we do not believe the UroVysion assay offers much additional benefit in patients with a positive cytology.Those patients are likely to have cancer,and a more careful evaluation with biopsy and upper tract imaging is recommended.This study also does not evaluate the benefits of the UroVysion assay in patients with a negative cytol-ogy.Future studies will be necessary to evaluate the role of markers in patients with negative cytology. CONCLUSIONSThis prospective validation study of a reflex FISH assay in patients with atypical cytology validates our previousfindings that reflex FISH is unneces-sary in patients with an obvious tumor on cystos-copy,but is beneficial in those with atypical cytology and equivocal or negative cystoscopy.The sensitiv-ity of the FISH assay to detect UC in patients with atypical cytology and equivocal or negative cystos-copy was100%with a specificity of60%to100% depending on history of UC and presence of cysto-scopic lesions.REFERENCES1.Jemal A,Siegel R,Ward E et al:Cancer statis-tics,2008.CA Cancer J Clin2008;58:71.2.Daniltchenko DI,Riedl CR,Sachs MD et al:Long-term benefit of5-aminolevulinic acidfluorescence assisted transurethral resection of superficial blad-der cancer:5-year results of a prospective random-ized study.J Urol2005;174:2129.3.Denzinger S,Burger M,Walter B et al:Clinically relevant reduction in risk of recurrence of super-ficial bladder cancer using5-aminolevulinic acid-inducedfluorescence diagnosis:8-year results of prospective randomized study.Urology2007;69: 675.4.Svatek RS,Lee D and Lotan Y:Correlation of office-based cystoscopy and cytology with histo-logic diagnosis:how good is the reference stan-dard?Urology2005;66:65.5.Hedelin H,Holmang S and Wiman L:The cost of bladder tumour treatment and follow-up.Scand J Urol Nephrol2002;36:344.6.Paez A,Coba JM,Murillo N et al:Reliability of the routine cytological diagnosis in bladder can-cer.Eur Urol1999;35:228.7.Raitanen MP,Aine R,Rintala E et al:Differences between local and review urinary cytology indiagnosis of bladder cancer.An interobservermulticenter analysis.Eur Urol2002;41:284.8.Nabi G,Greene DR and O’Donnell M:How im-portant is urinary cytology in the diagnosis ofurological malignancies?Eur Urol2003;43:632.9.Lotan Y and Roehrborn CG:Sensitivity and spec-ificity of commonly available bladder tumor mark-ers versus cytology:results of a comprehensiveliterature review and meta-analyses.Urology2003;61:109.10.Lokeshwar VB,Habuchi T,Grossman HB et al:Bladder tumor markers beyond cytology:Interna-tional Consensus Panel on bladder tumor mark-ers.Urology2005;66:35.11.Sarosdy MF,Schellhammer P,Bokinsky G et al:Clinical evaluation of a multi-targetfluorescent insitu hybridization assay for detection of bladdercancer.J Urol2002;168:1950.12.Sarosdy MF,Kahn PR,Ziffer MD et al:Use of amultitargetfluorescence in situ hybridization as-say to diagnose bladder cancer in patients withhematuria.J Urol2006;176:44.13.Lotan Y,Bensalah K,Ruddell T et al:Prospectiveevaluation of the clinical usefulness of reflexfluorescence in situ hybridization assay in pa-tients with atypical cytology for the detection ofurothelial carcinoma of the bladder.J Urol2008;179:2164.14.Babjuk M,Oosterlinck W,Sylvester R et al:EAUguidelines on non-muscle-invasive urothelial car-cinoma of the bladder.Eur Urol2008;54:303.15.Grossman HB,Messing E,Soloway M et al:Detection of bladder cancer using a point-of-careproteomic assay.JAMA2005;293:810.16.Skacel M,Fahmy M,Brainard JA et al:Multitar-getfluorescence in situ hybridization assay de-tects transitional cell carcinoma in the majority ofpatients with bladder cancer and atypical or neg-ative urine cytology.J Urol2003;169:2101.17.Raitanen MP,Aine R,Kylmala T et al:The di-lemma of suspicious urine cytology in patientsbeing followed for bladder cancer.Ann ChirGynaecol2001;90:256.18.Ferra S,Denley R,Herr H et al:Reflex UroVysiontesting in suspicious urine cytology cases.CancerCytopathol2009;117:7.19.Yoder BJ,Skacel M,Hedgepeth R et al:ReflexUroVysion testing of bladder cancer surveillancepatients with equivocal or negative urine cytol-ogy:a prospective study with focus on the naturalhistory of anticipatory positivefindings.Am J ClinPathol2007;127:295.REFLEX FLUORESCENCE IN SITU HYBRIDIZATION ASSAY AND ATYPICAL CYTOLOGY67。