a panel of three markers hyper- and hypomethylated in ... · 1978 clin cancer res; 20(7) april 1,...

Predictive Biomarkers and Personalized Medicine

A Panel of Three Markers Hyper- and Hypomethylated inUrine Sediments Accurately Predicts Bladder CancerRecurrence

Sheng-Fang Su1,3, Andr�e Luís de Castro Abreu1, Yoshitomo Chihara1, Yvonne Tsai1, Claudia Andreu-Vieyra1,Siamak Daneshmand1, Eila C. Skinner4, Peter A. Jones1, Kimberly D. Siegmund2, and Gangning Liang1

AbstractPurpose: The high risk of recurrence after transurethral resection of bladder tumor of nonmuscle invasive

disease requires lifelong treatment and surveillance. Changes in DNA methylation are chemically stable,

occur early during tumorigenesis, and can be quantified in bladder tumors and in cells shed into the urine.

Some urinemarkers have been used to help detect bladder tumors; however, their use in longitudinal tumor

recurrence surveillance has yet to be established.

Experimental Design: We analyzed the DNA methylation levels of six markers in 368 urine sediment

samples serially collected from 90 patients with noninvasive urothelial carcinoma (Tis, Ta, T1; grade low-

high). The optimum marker combination was identified using logistic regression with 5-fold cross-

validation, and validated in separate samples.

Results: A panel of three markers discriminated between patients with and without recurrence with the

area under the curve of 0.90 [95% confidence interval (CI), 0.86–0.92] and 0.95 (95% CI, 0.90–1.00),

sensitivity and specificity of 86%/89% (95%CI, 74%–99% and 81%–97%) and 80%/97% (95%CI, 60%–

96% and 91%–100%) in the testing and validation sets, respectively. The three-marker DNA methylation

test reliably predicted tumor recurrence in 80%ofpatients superior to cytology (35%) and cystoscopy (15%)

while accurately forecasting no recurrence in 74% of patients that scored negative in the test.

Conclusions:Given their superior sensitivity and specificity in urine sediments, a combination of hyper-

andhypomethylatedmarkersmayhelp avoidunnecessary invasive exams and reveal the importance ofDNA

methylation in bladder tumorigenesis. Clin Cancer Res; 20(7); 1978–89. �2014 AACR.

IntroductionBladder cancer was one of the 10 most prevalent malig-

nancies inmales in 2011 ranking fourth and eighth in termsof deaths and new cases, respectively (1, 2). Nonmuscleinvasive bladder cancer (NMIBC) accounts for 80% of allthe cases, and can be further classified into mucosa only(Ta), carcinoma in situ (Tis), and lamina propria invading(T1) lesions (3, 4). The primary treatment for NMIBC istransurethral resection of bladder tumor (TURBT) with orwithout intravesical chemo or immunotherapy; however,

more than 50%of patients recur after the TURBTprocedure,with the highest rate of recurrence occurring in patients withhigh-risk disease (5, 6). As a result, patients require frequentand lifelong monitoring following TURBT, making bladdercancer one of the most costly types of cancer to manage.

The current gold standard for monitoring of bladdercancer recurrence involves the use of cystoscopy and cytol-ogy (2, 3). Disease surveillance is cumbersome because ofthe invasive nature of cystoscopic examination and the lowsensitivity of urinary cytology in the detection of low-gradetumors (7). Recently, efforts have been devoted to findbetter markers of disease diagnosis and prognosis in sam-ples collected by noninvasive methods, such as urine sedi-ments (8). The addition of nuclearmatrix protein 22 (NMP-22), bladder tumor antigen, or UroVysion FISH has showntohelp increase the sensitivity of cytology (9).However, dueto their inconsistent performance in terms of specificity orsensitivity, the markers proposed to date have not beenwidely adopted in routine clinical practice (10). Therefore,there is a need tofind reliablemarkers tomonitor recurrencein TURBT patients, which in turn, may improve diseasemanagement.

Epigenetic changes, namely changes in chromatin structurethat regulate gene expression, occur during tumorigenesis

Authors' Affiliations : Departments of 1Urology and 2PreventiveMedicine;3Program in Genetic, Molecular, and Cellular Biology, USC Norris Com-prehensive Cancer Center, Keck School of Medicine, University of South-ern California, Los Angeles; and 4Department of Urology, School of Med-icine, University of Stanford, Stanford, California

Note: Supplementary data for this article are available at Clinical CancerResearch online (http://clincancerres.aacrjournals.org/).

Corresponding Authors: Gangning Liang, Department of Urology, USCNorris Comprehensive Cancer Center, 1441 Eastlake Ave., NOR 7346, LosAngeles, CA 90089. Phone: 323-865-0470; Fax: 323-865-0102 E-mail:[email protected]; and Kimberly D. Siegmund, E-mail: [email protected]

doi: 10.1158/1078-0432.CCR-13-2637

�2014 American Association for Cancer Research.

ClinicalCancer

Research

Clin Cancer Res; 20(7) April 1, 20141978

on October 7, 2020. © 2014 American Association for Cancer Research. clincancerres.aacrjournals.org Downloaded from

http://clincancerres.aacrjournals.org/

(11). Aberrant DNA methylation including increases anddecreases at specific loci is the most common epigeneticchange in tumorigenesis and it can be detected in prema-lignant lesions (12–15). Changes in DNA methylation arechemically stable and can be quantified, whichmakes thempotentially good tumor markers (16, 17). Inactivation oftumor suppressor genes by gain of DNA methylation(hypermethylation) or global loss of DNA methylation(hypomethylation), which activates genes that are nor-mally not expressed, have both been observed in bladdertumors (13, 18–20). Further studies have also shown thatmethylation changes found in urine sediments mirrorthose found in tumor tissues, indicating cancer-specificfeatures (19, 21–23).We previously identified both hyper- and hypomethy-

lated regions in bladder tumors and their premalignantlesions (12, 13). We demonstrated that a specific LINE1element, which is located within the MET oncogene (L1-MET) and activates an alternate transcript of MET, washypomethylated, and that the promoter of ZO2 (tightjunction protein 2)was hypermethylated in bladder tumorsas well as in adjacent histologically normal urothelium,suggesting that epigenetic changes precede morphologicchanges, a phenomenon that might be involved in malig-nant predisposition termed "epigenetic field defect" (12,13). We also found a group of genes that showed methyl-ation changes both in bladder tumors and urine sedimentsfrom patients with bladder cancer (21, 24). On the basis ofthese studies, we hypothesized that DNA methylationchanges in urine sediments from TURBT patients couldbe used to detect early bladder cancer recurrence. To testour hypothesis, we collected urine samples from TURBT

patients at follow-up visits over a 7-year period and assessedthe methylation status of a panel of markers, includingcancer-specific hypermethylated markers (HOXA9, SOX,NPY), an epigenetic driver gene (IRAK3; ref. 25), and fielddefect markers (ZO2 and L1-MET; refs. 12, 13). Our resultsshow that the combination of SOX1, IRAK3, and L1-METprovides better resolution than cytology and cystoscopy inthe detection of early recurrence.Overall, our results suggesta critical role of the balance between hyper- and hypo-DNAmethylation in bladder carcinogenesis, and provide a non-invasive and cost-effective way to assess patients post-TURBT, which may help inform treatment direction andlimit the use of invasive procedures such as cystoscopies.

Materials and MethodsPatients and sample collection

The study population includes patients under surveil-lance for tumor recurrence following TURBT for noninva-sive urothelial carcinoma (Tis, Ta, T1; grade low-high).Urine samples were obtained from 90 such patients withNMIBC at each available clinical follow-up visit. Patient’sage ranged from 41 to 96 years, with a median age of 69years. Urine collection at follow-up visits was performed atthe Department of Urology, USC Norris ComprehensiveCancer Center (Los Angeles, CA) from2004 to 2011 accord-ing to the institutional guidelines, in compliance withInstitutional Review Board-approved protocols. Patients athigh risk of recurrence (Tis, high-grade Ta/T1 disease) hadreceived prior intravesical therapy with Bacillus Calmette-Guerin (BCG) or mitomycin C at the discretion of thetreating physician. A total of 368 samples were collectedunder patient informed consent at follow-up visits over aperiod ranging from 5 to 89 months (Fig. 1A). Thebaseline clinicopathological characteristics of the patientsshowed no significant differences between the studygroups (Table 1).

Tumor recurrence was defined as biopsy-proven bladdercancer occurring subsequent to complete resection of thevisible primary tumor. Severe atypia concomitant withpapillary lesions detected by cytology and cystoscopy wasrecorded as recurrence only when the biopsy results wereabsent. Over the collection period, 34 patients had tumorrecurrence, whereas 56 patients were not diagnosed withrecurrence through the last follow-up visit. The clinicalcharacteristics of 34 recurrent tumors are summarizedin Table 2. Out of the 34 patients with recurrence, 31provided a urine sample at the time of diagnosis. Tumorswere characterized according to the criteria of the AmericanJoint Committee on Cancer (World Health Organization/International Society of Urological Pathology (ISUP);ref. 26) and staging was based on the tumornodemetastasisclassification (International Union Against Cancer; ref. 4)across the entire study period.

DNA extraction from urine sediments and DNAmethylation analysis by pyrosequencing

Urine specimens (�50 mL) included samples from both"urine" and "bladder wash." The bladder wash was

Translational RelevanceNonmuscle invasive bladder cancer, characterized by

a high rate of recurrence, is a relatively high-cost diseasein cancer management. Urinary DNA methylationchanges have shown their stable, reliable, and earlyappearance inbladder carcinogenesis.We longitudinallyanalyze DNA methylation changes in urine sedimentsserially collected from patients that underwent bladdertumor resections at the time of follow-up visits. Ourresults show that the combination of a transcriptionfactor (SOX1), a specific LINE-1 element, and a keyepigenetic driver gene interleukin-1 receptor-associatedkinase 3 (IRAK3) provides better resolution than cytol-ogy and cystoscopy in the detection of early recurrence.Therefore, our markers may help avoid unnecessaryinvasive exams during clinical tumor surveillance in acost-effective manner. We provide new insights into thevalue of incorporating both hyper- and hypo-DNAmethylation markers into the screening of urine sedi-ments for personalized following and monitoring oftumor recurrence in transurethral resection of bladdertumor (TURBT) patients.

Urinary DNA Methylation Markers Predict Bladder Cancer Recurrence

www.aacrjournals.org Clin Cancer Res; 20(7) April 1, 2014 1979



D

C

BA

TURBT TURBT

TURBTTURBT

+

R

+++

+

R

BCG/

BCG

BCG/BCG BCG BCGBCG BCG BCG

+

+

+

0 5

HOXA9 HOXA9SOX1 SOX1NPY NPYIRAK3 IRAK3ZO2 ZO2L1-MET L1-MET

HOXA9 HOXA9SOX1 SOX1NPY NPYIRAK3 IRAK3ZO2 ZO2L1-MET L1-MET

Hypermethylation Hypermethylation

Hypermethylation Hypermethylation

Hypomethylation Hypomethylation

Hypomethylation Hypomethylation

10 15 20 25 30Follow-up (month)

100

80

60

40

20

0

100

80

60

40

20

0–6 0 6 12 18 24 30

Follow-up (month)

AUC=0.92595%Cl=0.86–0.96P < 0.0001

AUC=0.94995%Cl=0.91–0.99P < 0.0001

AUC=0.995%Cl=0.83–0.95P < 0.0001

AUC=0.94295%Cl=0.92–0.98P < 0.0001

AUC=0.93495%Cl=0.87–0.97P < 0.0001

AUC=0.95295%Cl=0.88–0.97P < 0.0001

HOXA9

NPY IRAK3

Z02 L1-MET

SOX1

1-Specificity

1-Specificity

1-Specificity

1-Specificity

1-Specificity

1-Specificity

0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

0.0 0.2 0.4 0.6 0.8 1.0

0.0 0.2 0.4 0.6 0.8 1.00.0 0.2 0.4 0.6 0.8 1.0

0.0 0.2 0.4 0.6 0.8 1.0

0.0

0.2

0.4

0.8

1.0

0.6

0.0

0.2

0.4

0.8

1.0

0.6

0.0

7873

7258 7145

7844

Time (month) Time (month) –6 0 2.5 5.5 8.5 15 21.5 34

Time (month) Time (month)–6 100

100

80

60

40

20

0

100

80

60

40

20

0–6 –8 0 8 16 24 32 400 6 12 18 24 30 36 42 48

Follow-up (month) Follow-up (month)

12 19 26.5 35.5 42 49 –8 0 27 33 39 42

Cytology Cytology

Cytology Cytology

Cystoscopy Cystoscopy

Cystoscopy Cystoscopy

DN

A m

eth

ylat

ion

(%

)

DN

A m

eth

ylat

ion

(%

)

Biopsy Biopsy

Biopsy Biopsy

Treatment Treatment

Treatment Treatment

Mitomycin

mitomycin mitomycin

DN

A m

eth

ylat

ion

(%

)

DN

A m

eth

ylat

ion

(%

)

+

0 3 6 10 16 22 29

0.2

0.4

0.8

1.0

0.6

0.0

0.2

0.4

0.8

1.0

0.6

0.0

0.2

0.4

0.8

1.0

0.6

0.0

0.2

0.4

0.8

1.0

0.6

Sen

sitiv

ityS

ensi

tivity

Sen

sitiv

ity

Sen

sitiv

ityS

ensi

tivity

Sen

sitiv

ity

0 10 20 30 40 50Follow-up time (month) N = 368Follow-up

Follow-up with Recurrence

60 70 80 90

90

80

Recu

rrence

(N = 34)

No

Recu

rrence

(N = 56)

70

60

50

40

30

20

10

0

Pat

ien

ts

Su et al.

Clin Cancer Res; 20(7) April 1, 2014 Clinical Cancer Research1980



collected at the timeof cystoscopyby thenurses or urologist.The same sample underwent cytology and DNA methyla-tion analysis in a double-blinded fashion. The samples thatunderwent DNA methylation analysis were stored at 4�Cuntil cells were pelleted by centrifugation for 10 minutes at1,500 g. DNA was then extracted from urine sediments aspreviously described and stored at 4�C (21). DNA wasbisulfite converted using EZ DNA Methylation Kit (ZymoResearch) according to the manufacturer’s instructions. SixDNAmethylation markers were selected from our previousstudy (12, 13); the regions of interest were PCR amplifiedusing biotin-labeled primers (Supplementary Table S2) andanalyzed by pyrosequencing, a high-throughput and quan-

titative tool for DNA sequence detection. The percentage ofmethylated cytosines divided by the sumofmethylated andunmethylated cytosines was measured using PSQ HS96(Qiagen) as previously described (13).

Statistical analysisReceiver operating characteristic (ROC) curves summa-

rize the accuracy ofDNAmarkers in urine sediment from87independent samples, selected at the time of the last follow-up visit (nonrecurrent patients), or at the time of firstrecurrence (patients with recurrence). A subset of 83patients with complete data on all markers was used tobuild a multivariable predictor model. We used stepwise

Table 1. The clinicopathological characteristics of 90 TURBT patients

CharacteristicNo recurrenceN ¼ 56

RecurrenceN ¼ 34

Age, yMedian 71 69Range 42–96 41–87

Sex, no. (%)Male 48 (86) 27 (79)Female 8 (14) 7 (21)

Histology TCC, no. (%) 56 (100) 34 (100)Number of tumors, no. (%)Unifoci 19 (34) 18 (53)Multifoci 16 (29) 12 (35)Missing 21 (37) 4 (12)

T Stage, no. (%)Tis 2 (4) 1 (3)Ta 37 (66) 19 (56)T1 17 (30) 14 (41)

Tumor gradea, no. (%)Low 26 (46) 17 (50)High 30 (54) 17 (50)

Concomitant CIS, no. (%) 11 (20) 7 (21)Treatment, no. (%)Adjuvant BCG 36 (64) 20 (59)Adjuvant chemotherapy instillation 11 (19) 12 (35)

Follow-up time since TURBT, y 5.5 (0.6–9.7) 4.7 (0.4–26)Study follow-up time, y 3.5 (0.4–7.1) 3.6 (0.5–7.4)Total urines analyzed, no. 208 160Urines analyzed/patient, no.Mean (�SD) 3.7 � 1.8 4.7 � 2.1Range 2–9 2–10

Abbreviations: TCC, transitional cell carcinoma; CIS, carcinoma in situ; no., number; (%), percentage.aGrade 1 and 2 are reported as low grade. Grade 3 and more are reported as high grade.

Figure 1. A panel of six DNA methylation markers tested in urine sediments from TURBT patients was positively correlated with bladder tumor recurrenceand showed high sensitivity and specificity. A, timeline of longitudinally collected urine sediment samples frompatients with bladder cancer tumor resections.Each patient's starting point, denoted by time 0, refers to the first follow-up visit in the study when a urine sample was collected. A follow-up visitmarked in red indicates the time of recurrence. B, ROC curves of HOXA9, SOX1, NPY, IRAK3, ZO2, and L1-MET were created using 31 urine sedimentsof TURBT patients at first recurrence and 56 urine sediments from the last follow-up of recurrence-free patients. C and D, long-term DNA methylationanalysis in TURBT patients and its relationship with clinical status in patients who had no recurrence (C) and patients who had recurrence (D). �, negative;�, suspicious; þ, positive (biopsy-proven bladder tumor); R, recurrence.





Table 2. The clinical characteristics of 34 patients with recurrence bladder cancer

Baseline RecurrenceCharacteristic N ¼ 34

Histology TCC, no. (%) 34 (100) 30 (88)Number of tumors, no. (%)Unifoci 18 (53) 19 (56)Multifoci 12 (35) 12 (35)Missing 4 (12) 3 (9)

T Stage, no. (%)Tis 1 (3) 4 (12)Ta 19 (56) 20 (59)T1 14 (41) 2 (6)T2 0 1 (3)Missing 0 7 (20)

Tumor gradea, no. (%)Low 17 (50) 16 (47)High 17 (50) 13 (38)Missing 0 5 (15)

Treatment, no. (%)Adjuvant BCG 20 (59) 15 (44)Adjuvant chemotherapy instillation 12 (35) 5 (15)

Baseline Recurrence

Patient number Number of tumors T stage Grade Number of tumors T stage Grade

4843 Multifoci T1 High Multifoci Ta High5137 Unifoci Ta Low Unifoci Ta Low6664 Multifoci T1 High Unifoci Missing Low6675 Unifoci Ta Low Multifoci Ta Low6762 Missing Ta Low Unifoci Ta Low6804 Multifoci T1 High Unifoci Missing Low6851 Unifoci T1 High Unifoci Ta Low7145 Multifoci T1 High Multifoci CIS High7258 Unifoci T1 High Unifoci CIS High7346 Unifoci Ta High Unifoci Ta High7397 Unifoci T1 Low Multifoci Ta Low7592 Unifoci Ta High Unifoci Ta High7662 Missing Ta Low Multifoci Ta High7716 Multifoci Ta High Multifoci Ta High7718 Unifoci T1 Low Unifoci T1 High7728 Unifoci T1 Low Multifoci Missing NA7743 Unifoci CIS CIS Unifoci Ta Low7774 Unifoci Ta Low Multifoci Ta Low7792 Multifoci Ta Low Missing Missing NA7809 Multifoci Ta Low Unifoci CIS High7810 Unifoci Ta Low Multifoci Ta Low7817 Multifoci Ta Low Multifoci Ta Low7859 Multifoci T1 High Unifoci Ta High7873 Multifoci Ta Low Unifoci Missing NA7891 Missing T1 High Unifoci CIS High7896 Unifoci Ta Low Unifoci Ta Low8659 Unifoci T1 High Missing Missing NA8792 Unifoci T1 High Unifoci T2 High

(Continued on the following page)

Su et al.




logistic regression, selecting variables to add or subtractbased on the Akaike Information Criterion (AIC). The riskscore was obtained using logistic regression, and representsthe probability of a positive result (recurrence) on the log-odds scale.On this scale, a score of 0 represents a probabilityof 0.5 (50% chance) for a patient having recurrence. Thissuggests that the best cutoff of the risk score to predictrecurrence is 0, with scores > 0 having a more than 50%chance of being from a recurrent patient, and scores < 0having a less than 50% chance of being from a recurrencepatient. The risk score was computed using 83 patients withcomplete data on all markers (29 samples taken at the timeof first recurrence after TURBT, and 54 samples frompatients who were recurrence free at the last time of urinecollection). The three-marker panel was selected using theforward and backward stepwise variable selection proce-dure. The AIC is the optimality criterion used for modelselection.When comparing twomodels, themodelwith thelowest AIC is preferred. We compare the AIC of the modelwith no variables to the AIC of all 1-variable models, andadd the variable reducing the AIC themost. This is repeated,by adding thenext variable that further reduces theAIC. Thisforward step is repeated once more, with the addition of abackward step that evaluates thepossibility of removingoneof the variables already in themodel. For each new step, theaddition/removal of a variable is considered, providing ameans of "stepping" through models with different com-bination of variables, to search for the best predictivemodel. The procedure endswhen themodel with the lowestAIC is found.Sensitivity and specificity were estimated using 5-fold

cross-validation, repeating the model selection for eachsubdivision of the data. Five-fold cross-validation was usedto obtain the reported (less biased) estimates of sensitivityand specificity. Model selection was performed using for-ward and backward stepwise selection on four fifths of thedataset, and the predictive ability assessed on the fifth thatwas not used for variable selection, an independent datasubset. This was repeated five times, each time holding aseparate fifth of the dataset out for validation, and perform-ing a newmodel selection on the remaining four fifths. The

five data subsets consisted of four groups of 17 (6 recur-rences/11 nonrecurrences) and one group of 15 (5 recur-rences/10 nonrecurrences). The final model was then eval-uated on the remaining samples from our dataset to eval-uate the performance of themarkers providing the best fit toour training data. Control samples (n¼ 134) included visitsbefore the last follow-up visit where the patient was notdiagnosed with bladder cancer; case samples (n ¼ 25)included recurrences occurring after the first recurrence andsamples at the initial clinic visit when the patient presentedwith bladder cancer.

ResultsDNA methylation analysis in urine sediments

To evaluatewhether hypermethylation ofHOXA9, SOX1,NPY, IRAK3, and ZO2, and hypomethylation of L1-METcould be detected in urine sediments, we analyzed urinesamples collected from patients with bladder tumors (n ¼20) and from age-matched cancer-free controls (n ¼ 20)using pyrosequencing. The results show that DNA methyl-ationofHOXA9 (P<0.0001), SOX1 (P¼0.0017),NPY (P¼0.005), IRAK3 (P < 0.0001), and ZO2 (P < 0.0001) wassignificantly increased,whereasmethylation of L1-MET (P <0.0001)was significantly decreased inurine sediments frompatients with cancer compared with healthy donors, indi-cating that the DNA methylation status in urine sedimentsmirror that of the tumor (Supplementary Fig. S1).

Longitudinal study of DNA methylation changes inurine sediments collected from TURBT patients at thetime of follow-up visits

To examine whether aberrant DNA methylation of fivehypermethylated and one hypomethylated markers inurine sediments is associated with tumor recurrence, weanalyzed their DNA methylation status in 368 urine sedi-ments collected in follow-up visits followed under stan-dard care amongst patients that had undergone priortumor resections. Figure 1A shows the representativetime-dependent methylation analysis. Patients withoutrecurrence had longer median follow-up time than therecurrence group (Table 1). The Spearman correlation of

Table 2. The clinical characteristics of 34 patients with recurrence bladder cancer (Cont'd )

8928 Missing T1 High Missing Missing NA9216 Multifoci Ta Low Unifoci Ta Low9532 Unifoci Ta Low Unifoci Ta Low9536 Unifoci Ta High Unifoci T1 High9626 Unifoci Ta Low Multifoci Ta Low9627 Unifoci Ta High Multifoci Ta Low

Abbreviations: TCC, transitional cell carcinoma; CIS, carcinoma in situ; no., number; (%), percentage.aGrade 1 and 2 are reported as low grade. Grade 3 and more are reported as high grade.

Baseline Recurrence

Patient number Number of tumors T stage Grade Number of tumors T stage Grade





DNA methylation level for each marker was then calcu-lated (Supplementary Fig. S2). Individual DNA methyla-tion marker success rates averaged 98.9% across all sam-ples (94.9–100%). Next, the DNA methylation levels ofthese markers in 31 urine sediments from patients collect-ed at the time of first recurrence was compared with that of56 samples from the last follow-up visit of patients whodid not recur within the study period (SupplementaryTable S1). Our results show that the six markers individ-ually displayed high sensitivity and specificity in recurrencedetection as evidenced by the ROC curves and area underthe curve (AUC) values [0.93–0.95, 95% confidence inter-val (CI) shown in Fig. 1B]. In the group of patients withoutbladder tumor recurrence, urine sediment samples showedconsistent DNA methylation levels throughout the dura-tion of surveillance; markers methylated in tumorsdecreased in methylation levels, whereas the marker hypo-methylated in tumors (L1-MET) increased and maintainedhigh methylation levels after tumor resection (Fig. 1C;patients 7,873 and 7,844). In contrast, patients who hadbladder tumor recurrence displayed changes in the DNAmethylation status of all six markers at the time of clin-ically defined recurrence. DNA methylation levels ofhypermethylated markers continued to increase untilrecurrence was confirmed with a positive cystoscopy andbiopsy 19 and 33 months after the first urine sample wasobtained. The elevated methylation levels decreased fol-lowing resection surgery (Fig. 1D; patients 7,258 and7,145). Our results demonstrate that the methylationlevels of these markers displayed a clear trend in thesamples obtained at follow-up visits leading to the con-firmation of recurrence and showed not only a signifi-cant correlation with recurrence (P < 0.0001), but also a

possible predictive value as methylation changes could bedetected before clinical evidence of recurrence.

A three-marker panelTo determine the combination of markers capable of

detecting tumor recurrence in urine sediments with thehighest sensitivity and specificity, we built a model ofmultiple markers by logistic regression, using 29 samplestaken at the time of first recurrence and 54 samples frompatients who were recurrence free at the last time of urinecollection. Three markers SOX1, IRAK3, and L1-MET werefound to provide the best possible marker combination(risk score ¼ �0.37608 þ 0.17095 � SOX1 þ 0.21604 �IRAK3 – 0.09887 � L1-MET). Scores above zero predictrecurrence. Ninety-four percent of patients with no recur-rence showed negative scores (95% CI, 88%–100%) and93% of patients with recurrence showed positive scores(95% CI, 84%–100%; Fig. 2A). The 5-fold cross-validationanalysis estimated an AUC of 0.90 (95% CI, 0.86–0.92)with sensitivity of 86% (95%CI, 74%–99%) and specificityof 89% (95% CI, 81%–97%) for a risk score cutoff of zero(Fig. 2B and Supplementary Fig. S3B). This three-genemodel was then validated in the remaining samples takenfrom the same patient cohort using 25 samples taken at avisit where known urothelial carcinoma was present (TU,nine recurrences after the first recurrence, and 16 at the timeof entry into the study), and 134 samples taken at visitsbefore the last follow-up from patients who had not devel-oped cancer during a given follow-up time (CU). Notably,the three-marker model showed an AUC of 0.95 (95% CI,0.90–1.00) with high sensitivity (80%; 95%CI, 60%–96%)and specificity (97%; 95% CI, 91%–100%) in the internalvalidation set (Fig. 2C and Supplementary Fig. S3C). The

No recurrence recurrence–10

0

10

20

30

Ris

k sc

ore

CU TU–10

0

10

20

30

Ris

k sc

ore

B

Sensitivity = 93%Specificity = 94%

(N = 45) (N = 29) (N = 134) (N = 25)

CASensitivity = 80%Specificity = 97%

******

No recurrence Recurrence CU TUN N 134

43* 80

2025

Positive Positive6 25% %11 86

54

1-Specificity(false-positive)

1-Specificity(false-positive)

Sensitivity(true-positive)

Sensitivity(true-positive)

29

^

Figure 2. A three-marker signatureshowed high sensitivity andspecificity in detecting tumorrecurrence. A, the risk score of�0.37608 þ 0.17095 � SOX1 þ0.21604 � IRAK3 – 0.09887 � L1-MET was calculated in the urinesediments of TURBT patients withno recurrence at the last follow-upand with recurrence. B, 5-foldcross-validation showed asensitivity of 86% (95% CI, 74%–

99%) and specificity of 89% (95%CI, 81%–97%). C, this three-marker model was validated in aseparate urine sediment samplesthat included urine sediments fromrecurrence-free patients before thelast follow-up visit (CU) and urinesediments of patients with knownurothelial carcinoma (TU) with thesensitivity of 80% (95% CI, 60%–

96%) and specificity of 97% (95%CI, 91%–100%). Risk scoresabove the cutoff value (red dashedline) denote positive scores,whereas those below signifynegative scores.

Su et al.




DNA methylation scores within the no recurrence andrecurrence groups showed no correlation with any of theprimary tumor characteristics (Supplementary Table S3).However, a positive correlation was found between DNAmethylation status and grade of the primary tumor in therecurrence group as well as stage (Ta vs. T1) at recurrence(P<0.05; Supplementary Table S4). These results demon-strate that the combination of a tumor-specific hyper-methylated marker, SOX1, an epigenetic driver, IRAK3,and a field defect-associated hypomethylated marker, L1-MET, can detect disease recurrence with high sensitivityand specificity.

Power of prediction of recurrenceTo evaluate whether methylation of the three-marker

model predicts recurrence in our longitudinal study sam-ples, we screened DNA methylation and calculated riskscores given by the combination of SOX1, IRAK3, andL1-MET in every urine sample obtained at follow-up visitsfrom 90 TURBT patients. Risk scores over follow-up fallprimarily below the cutoff in samples frompatients withoutrecurrence (Fig. 3A) and often above the cutoff in samplesfrom patients with recurrence (Fig. 3B). Positive DNAmethylation scores were found in 90% of the samples(34/38) at the time of recurrence diagnosis, exceeding thesensitivity of both cytology (16%) and cystoscopy (8%) forthe same visits to the clinic (Fig. 4A). To quantify theprediction value of the three markers, we analyzed riskscores in patents/samples in the period before recurrence(Fig. 4B and Supplementary Fig. S4A) or at any time visits(Fig. 4C and Supplementary Fig. S4B). Eighty percent ofpatients (16/20) whose urine samples showed a history ofpositiveDNAmethylation scores developed recurrence later(95% CI, 62%–98%). Out of the 70 patients who did nothave a history of positive DNA methylation scores, 52(74%) did not recur (95% CI, 64%–85%; Fig. 4B). Ourresults indicate that the three-marker signature detected inurine sediments of follow-up visits can reliably predictrecurrence in 80% of patients, which is superior to the35% (95% CI, 14%–56%) and 15% (95% CI, 0%–31%)predicted by cytology and cystoscopy, respectively (Fig. 4B).Sample-level charts report the percentage of samples byDNA methylation score from patients with or withoutrecurrence. The results demonstrate that the three-markermodel can successfully detect current and subsequent recur-rence in90%(64/71)ofDNAmethylation-positive samples(95% CI, 83%–97%), whereas these same samples showed30% (95%CI, 19%–40%) suspicious plus positive in cytol-ogy and 44% (95%CI, 32%–55%) suspicious plus positivein cystoscopy (Supplementary Fig. S4B).

DiscussionMarkers that can be detected in urine sediments provide a

noninvasive method to test for the presence of bladdertumor cells and premalignant cell populations in theurinary tract (22). Some U.S. Food and Drug Administra-tion-approved tests, such as NMP-22, ImmunoCyt, and

UroVysion, have been used for the surveillance of bladdercancer, and have shown a higher degree of sensitivity thancytology. However, the following situationsmake them lessthan ideal for comprehensive utilization and general adop-tion into the clinical practice: (i) thesemarkers have a lowerspecificity than cytology, (ii) the specificity of NMP-22 andImmunoCyt are influenced by other urinary benign condi-tions, (iii) they are not meant to replace urinary cytologyand cystoscopy, but to complement those surveillancemethods, and (iv) they are expensive, labor intensive, andprovide marginal improvement in disease detection (3,10, 27, 28). Although some of these markers are currentlyused to predict the responses to intravesical therapies likeBCG, further studies in a larger population and consistentperformance assessment are still needed (29). In addi-tion, some new investigational urine markers suchas microsatellite alterations and gene mutations (e.g.,

Figure 3. The risk scores given by the combination of three DNAmethylationmarkers in the continuously monitored urine showed distinctpatterns in recurrence and no recurrence patients. A and B, DNAmethylation levels of the three-marker combination were used tocalculate the risk score for recurrence in the urine sediment samples fromTURBT patients who had no recurrence (A) or had recurrence (B) and oftwo individual patients. V, TURBT operation; R, recurrence; risk score ¼� 0.37608þ 0.17095� SOX1þ 0.21604� IRAK3 – 0.09887� L1-MET.The red dashed line indicates the cutoff value. The orange arrowrepresents positive scores before recurrence.





fibroblast growth factor receptor 3; FGFR3) have not beenwidely deployed as a routine screening method for recur-rence (30–32).

Changes in DNA methylation are chemically stable,occur early during tumorigenesis, and can be quantifiedon high-throughput platforms, which make them poten-

tially good tumor markers. Many studies have shown thataberrant DNAmethylation of a single or a combination ofmarkers in urine sediments of patients carrying bladdercancer stably reflects their methylation status in bladdertumors independently of the presence of hematuria,bladder infections, or other bladder benign conditions,

A

B C

RecurrenceNo recurrencePositiveSuspicious

N/ANegative

Before recurrence Anytime

At recurrence

(95% CI, 0.64–0.85)

(95% CI, 0.71–0.89)

(95% CI, 0.66–0.86)

(95% CI, 0.62–0.98)

(95% CI, 0.14–0.56)

(95% CI, 0.0–0.31)

(95% CI, 0.59–0.80)

(95% CI, 0.77–0.93)

(95% CI, 0.71–0.89)

(95% CI, 0.77–0.99)

(95% CI, 0.27–0.61)

(95% CI, 0.55–0.86)

Figure 4. Three DNA methylation markers help predict the risk of recurrence of bladder tumors in 80% of patients. A, percentage of urine sediments that hadpositive scores (DNA methylation score calculated to be higher than cutoff values) at the time of recurrence (38 samples, 29 patients). B and C,pie charts summarize all patients in the period before recurrence (B), or at anytime (C) and the comparisonwith cytology and cystoscopy reports in these samegroups of patients. A patient-level positive score represents a history of positive DNA methylation scores at any eligible visits. Patient-level chartsreport the percentage of recurrence-free patients in those without a history of positive samples (negative predictive value, NPV) and percentage of patientswith recurrence in those with a history of DNA methylation positive samples (positive predictive value, PPV).

Su et al.




thereby establishing DNA methylation screening of urinesediments as a promising noninvasive approach for blad-der cancer detection (10, 19, 21, 33–37). However, moststudies have focused on finding correlations between themethylation status of markers present in the primarytumor or urine sediments at the time of diagnosis (beforeTURBT) and recurrence (38). Although some of suchmarkers showed positive correlations with the number,size, grade, and stage of primary tumors and prior recur-rence history, others did not, likely due to the variation ofthe study population or the sample collection conditions(39–43). The variety of methods used to detect methyl-ation, the fact that only one sample was evaluated bypatient, and the reduced number of control samples usedin the different studies, have made it difficult to accuratelypredict recurrence.More recently, it has been proposed that longitudinal

collection and testing of urine sediments may help assessthe prognostic and recurrence predictive value of markers(43, 44). Several studies undertook this approach byusing DNA methylation analysis, microsatellite markers,and a FGFR3 mutation assay (45, 46). Although thesemarkers were highly sensitive, they displayed low speci-ficity in some cases comparable with that of cytology or ahigh rate of false-positive results (47, 48). The three-marker model proposed in this study may circumventthe specificity problem. As far as we know, we are the firstgroup using multiple DNA methylation markers to direct-ly test risk value and monitor recurrence in serial urinesamples from patients with a history of noninvasiveurothelial carcinoma. SOX1, IRAK3, and L1-MET had arecurrence predictive power far superior to that of cytol-ogy and cystoscopy (80% vs. 35% vs. 15% accuracy), andtherefore, they could supplement visits that reveal cyto-logically or cystoscopically atypical or suspicious results.NPVs of the three-marker panel were slightly lower thanthose obtained by cytology and cystoscopy, largely due tothe definition of recurrence in our study. Patients with"no recurrence" displayed negative cytologic or cystosco-pical results.In addition, the three markers we identified here may

also contribute to functional changes during tumorigen-esis. For example, IRAK3 shows significantly decreasedexpression and promoter methylation in various cancertypes, and our laboratory identified IRAK3 as a key driverfor cancer cell survival through the activation of survivin(25). Some of the limitations of our study are that themechanisms underlying bladder tumor recurrence arestill unclear and that the samples of the validation setwere not ideal. However, our analysis revealed certainfunctional roles of these methylation markers. We min-

imized the chances of over-reporting sensitivity/specific-ity in our study by evaluating the risk score on samplesdifferent from those used to derive the statistical model.Undoubtedly, validation of these urine markers in alarger, independent patient cohort with appropriate fol-low-up visits is needed. However, the length of thefollow-up time for each individual patient could be adifficult point for such a long-term study.

In conclusion, our study provides new insights into thevalue of a combination of hypermethylated and hypo-methylated tumor-specific markers to screen urine sedi-ments from patients following bladder tumor resections.To our knowledge, this is the first study to analyze multipleurine sediment samples collected over the course of manyyears by DNA methylation markers for bladder tumorrecurrence. This studyprovides evidence that amarker panelmay helpminimize the frequency of cystoscopy for patientswith a negative score. We suggest that patients with apositive urinary methylation test but no clinical evidenceof bladder cancer disease should still be closely monitoredbecause they carry a high risk of recurrence.

Disclosure of Potential Conflicts of InterestP.A. Jones is a consultant/advisory board member of Astex, Lilly, and

Zymo. No potential conflicts of interest were disclosed by the otherauthors.

Authors' ContributionsConception and design: Y. Chihara, E.C. Skinner, P.A. Jones, G. LiangDevelopment of methodology: A.L. de C. Abreu, Y. Chihara, Y.C. Tsai,G. LiangAcquisitionofdata (provided animals, acquired andmanagedpatients,provided facilities, etc.): A.L. de C. Abreu, Y. Chihara, E.C. Skinner,G. LiangAnalysis and interpretation of data (e.g., statistical analysis, biosta-tistics, computational analysis): S.-F. Su, A.L. de C. Abreu, Y. Chihara,S. Daneshmand, E.C. Skinner, P.A. Jones, K.D. Siegmund, G. LiangWriting, review, and/or revision of the manuscript: S.-F. Su, Y. Chihara,C. Andreu-Vieyra, S. Daneshmand, P.A. Jones, K.D. Siegmund, G. LiangAdministrative, technical, or material support (i.e., reporting ororganizing data, constructing databases): S.-F. Su, A.L. de C. Abreu,Y. Chihara, G. LiangStudy supervision: Y. Chihara, P.A. Jones, G. Liang

AcknowledgmentsThe authors thank Ravi Agarwal for carefully reading and revising the

manuscript;Dr. SueEllenMartin andMoli Chen for urine sample processing;themembers of the Cytopathology Laboratory of the KeckMedical Center ofUSC for their assistance with this project; and Hui Shen and Drs. Terry Kellyand Jueng Soo You for the helpful discussion.

Grant SupportThis work was supported by NCI (RO1 CA083867-PAJ, RO1 CA

124518-GL, and RO1 CA097346-KDS) and part of P30CA014089-tissueprocurement.

Received September 27, 2013; revised December 19, 2013; acceptedJanuary 9, 2014; published online April 1, 2014.

References1. Siegel R, Ward E, Brawley O, Jemal A. Cancer statistics, 2011:

The impact of eliminating socioeconomic and racial disparitieson premature cancer deaths. CA Cancer J Clin 2011;61:212–36.

2. Morgan TM, Clark PE. Bladder cancer. Curr Opin Oncol 2010;22:242–9.

3. Babjuk M, Oosterlinck W, Sylvester R, Kaasinen E, B€ohle A, Palou-Redorta J, et al. EAU guidelines on non-muscle-invasive urothelial





carcinoma of the bladder, the 2011 update. Eur Urol 2011;59:997–1008.

4. Sobin LH, Gospodarowicz MK, Christian W. TNM classification ofmalignant tumours. 7th ed. Wiley-Blackwell; 2009.

5. Shelley MD, Mason MD, Kynaston H. Intravesical therapy for super-ficial bladder cancer: A systematic review of randomised trials andmeta-analyses. Cancer Treat Rev 2010;36:195–205.

6. Mill�an-Rodríguez F, Ch�echile-Toniolo G, Salvador-Bayarri J, Palou J,Algaba F, Vicente-Rodríguez J. Primary superficial bladder cancer riskgroups according to progression, mortality and recurrence. J Urol2000;164:680–4.

7. Lintula S, Hotakainen K. Developing biomarkers for improved diag-nosis and treatment outcome monitoring of bladder cancer. ExpertOpin Biol Ther 2010;10:1169–80.

8. Sturgeon CM, Duffy MJ, Hofmann BR, Lamerz R, Fritsche HA, Gaar-enstroom K, et al. National Academy of Clinical Biochemistry Labo-ratory Medicine Practice Guidelines for use of tumor markers in liver,bladder, cervical, and gastric cancers. Clin Chem 2010;56:e1–48.

9. Parker J, Spiess PE. Current and emerging bladder cancer urinarybiomarkers. ScientificWorldJournal 2011;11:1103-12.

10. Reinert T. Methylation markers for urine-based detection of bladdercancer: The next generation of urinary markers for diagnosis andsurveillance of bladder cancer. Adv Urol 2012;2012:503271.

11. Jones PA. Functions of DNA methylation: Islands, start sites, genebodies and beyond. Nat Rev Genet 2012;13:484–92.

12. Wolff EM, Chihara Y, Pan F, Weisenberger DJ, Siegmund KD, SuganoK, et al. Unique DNAmethylation patterns distinguish noninvasive andinvasive urothelial cancers and establish an epigenetic field defect inpremalignant tissue. Cancer Res 2010;70:8169–78.

13. Wolff EM, Byun HM, Han HF, Sharma S, Nichols PW, Siegmund KD,et al. Hypomethylation of a LINE-1 promoter activates an alternatetranscript of the MET oncogene in bladders with cancer. PLoS Genet2010;6:e1000917.

14. Niwa T, Tsukamoto T, Toyoda T, Mori A, Tanaka H, Maekita T, et al.Inflammatory processes triggered by Helicobacter pylori infectioncause aberrant DNA methylation in gastric epithelial cells. Cancer Res2010;70:1430–40.

15. Ushijima T, Hattori N. Molecular pathways: Involvement of Helicobac-ter pylori-triggered inflammation in the formation of an epigenetic fielddefect, and its usefulness as cancer risk and exposure markers. ClinCancer Res 2012;18:923–9.

16. Wolff EM, Liang G, Jones PA. Mechanisms of Disease: Genetic andepigenetic alterations that drive bladder cancer. Nat Clin Pract Urol2005;2:502–10.

17. Laird PW.Thepower and thepromise ofDNAmethylationmarkers.NatRev Cancer 2003;3:253–66.

18. Kim WJ, Kim YJ. Epigenetic biomarkers in urothelial bladder cancer.Expert Rev Mol Diagn 2009;9:259-69.

19. Reinert T, Modin C, Castano FM, Lamy P, Wojdacz TK, Hansen LL,et al. Comprehensive genome methylation analysis in bladdercancer: Identification and validation of novel methylated genes andapplication of these as urinary tumor markers. Clin Cancer Res2011;17:5582–92.

20. Vallot C, Stransky N, Bernard-Pierrot I, H�erault A, Zucman-Rossi J,Chapeaublanc E, et al. A novel epigenetic phenotype associated withthe most aggressive pathway of bladder tumor progression. J NatlCancer Inst 2011;103:47–60.

21. Friedrich MG, Weisenberger DJ, Cheng JC, Chandrasoma S, Sieg-mund KD, Gonzalgo ML, et al. Detection of methylated apoptosis-associated genes in urine sediments of bladder cancer patients. ClinCancer Res 2004;10:7457–65.

22. Seifert HH, Schmiemann V, Mueller M, Kazimirek M, Onofre F, Neu-hausen A, et al. In situ detection of global DNA hypomethylation inexfoliative urine cytology of patients with suspected bladder cancer.Exp Mol Pathol 2007;82:292–7.

23. Kim YK, KimWJ. Epigenetic markers as promising prognosticators forbladder cancer. Int J Urol 2009;16:17–22.

24. ChiharaY,Kanai Y, FujimotoH,SuganoK,KawashimaK, LiangG, et al.Diagnostic markers of urothelial cancer based on DNA methylationanalysis. BMC Cancer 2013;13:275.

25. DeCarvalhoDD, SharmaS, You JS, SuSF, TaberlayPC,Kelly TK, et al.DNA methylation screening identifies driver epigenetic events of can-cer cell survival. Cancer Cell 2012;21:655–67.

26. Edge SB, Byrd DR, Compton CC, Fritz AG, Greene FL, Trotti A, et al.AJCC Cancer Staging Manual, 7th ed.. New York: Springer-Verlag;2010.

27. Kamat AM, Karam JA, Grossman HB, Kader AK, Munsell M, DinneyCP. Prospective trial to identify optimal bladder cancer surveillanceprotocol: Reducing costs while maximizing sensitivity. BJU Int2011;108:1119–23.

28. Whitson J, Berry A, Carroll P, Konety B. A multicolour fluorescence insitu hybridization test predicts recurrence in patients with high-risksuperficial bladder tumours undergoing intravesical therapy. BJU Int2009;104:336–9.

29. Kamat AM, Dickstein RJ, Messetti F, Anderson R, Pretzsch SM,Gonzalez GN, et al. Use of fluorescence in situ hybridization to predictresponse to bacillus Calmette-Gu�erin therapy for bladder cancer:Results of a prospective trial. J Urol 2012;187:862–7.

30. Tilki D, Burger M, Dalbagni G, Grossman HB, Hakenberg OW, Palou J,et al. Urine markers for detection and surveillance of non-muscle-invasive bladder cancer. Eur Urol 2011;60:484–92.

31. Steiner G, Schoenberg MP, Linn JF, Mao L, Sidransky D. Detection ofbladder cancer recurrence by microsatellite analysis of urine. Nat Med1997;3:621–4.

32. Kompier LC, Lurkin I, van der AaMN, van Rhijn BW, van der Kwast TH,Zwarthoff EC. FGFR3, HRAS, KRAS, NRAS and PIK3CA mutations inbladder cancer and their potential as biomarkers for surveillance andtherapy. PLoS One 2010;5:e13821.

33. Costa VL, Henrique R, Danielsen SA, Duarte-Pereira S, Eknaes M,Skotheim RI, et al. Three epigenetic biomarkers, GDF15, TMEFF2, andVIM, accurately predict bladder cancer from DNA-based analyses ofurine samples. Clin Cancer Res 2010;16:5842–51.

34. Chung W, Bondaruk J, Jelinek J, Lotan Y, Liang S, Czerniak B, et al.Detection of bladder cancer using novel DNA methylation biomarkersin urine sediments. Cancer Epidemiol Biomarkers Prev 2011;20:1483–91.

35. Dulaimi E,UzzoRG,GreenbergRE,Al-SaleemT,CairnsP.Detectionofbladder cancer in urine by a tumor suppressor gene hypermethylationpanel. Clin Cancer Res 2004;10:1887–93.

36. Lin HH, Ke HL, Huang SP, Wu WJ, Chen YK, Chang LL. Increasesensitivity in detecting superficial, low grade bladder cancer by com-bination analysis of hypermethylation of E-cadherin, p16, p14,RASSF1A genes in urine. Urol Oncol 2010;28:597–602.

37. Chan MW, Chan LW, Tang NL, Tong JH, Lo KW, Lee TL, et al.Hypermethylation of multiple genes in tumor tissues and voidedurine in urinary bladder cancer patients. Clin Cancer Res 2002;8:464–70.

38. Oliveira AI, Jer�onimo C, Henrique R. Moving forward in bladder cancerdetection anddiagnosis: The role of epigenetic biomarkers. Expert RevMol Diagn 2012;12:871–8.

39. Friedrich MG, Chandrasoma S, Siegmund KD, Weisenberger DJ,Cheng JC, Toma MI, et al. Prognostic relevance of methylation mar-kers in patients with non-muscle invasive bladder carcinoma. Eur JCancer 2005;41:2769–78.

40. Tada Y, Wada M, Taguchi K, Mochida Y, Kinugawa N, Tsuneyoshi M,et al. The association of death-associated protein kinase hypermethy-lation with early recurrence in superficial bladder cancers. Cancer Res2002;62:4048–53.

41. Zhao Y, Guo S, Sun J, Huang Z, Zhu T, Zhang H, et al. Methylcap-seqreveals novel DNA methylation markers for the diagnosis and recur-rence prediction of bladder cancer in a Chinese population. PLoS One2012;7:e35175.

42. Negraes PD, Favaro FP, Camargo JL, Oliveira ML, Goldberg J, RainhoCA, et al. DNAmethylation patterns in bladder cancer andwashing cellsediments: Aperspective for tumor recurrence detection. BMCCancer2008;8:238.

43. Reinert T, Borre M, Christiansen A, Hermann GG, Orntoft TF, DyrskjøtL. Diagnosis of bladder cancer recurrence based on urinary levels ofEOMES, HOXA9, POU4F2, TWIST1, VIM, and ZNF154 hypermethyla-tion. PLoS ONE 2012;7:e46297.


Su et al.



44. Hoque MO, Begum S, Topaloglu O, Chatterjee A, Rosenbaum E, VanCriekinge W, et al. Quantitation of promoter methylation of multiplegenes in urine DNA and bladder cancer detection. J Natl Cancer Inst2006;98:996–1004.

45. Roupret M, Hupertan V, Yates DR, Comperat E, Catto JW, Meuth M,et al. A comparison of the performance of microsatellite and methyl-ation urine analysis for predicting the recurrence of urothelial cellcarcinoma, and definition of a set of markers by Bayesian networkanalysis. BJU Int 2008;101:1448-53.

46. Zuiverloon TC, van der Aa MN, van der Kwast TH, Steyerberg EW,Lingsma HF, Bangma CH, et al. Fibroblast growth factor receptor 3

mutation analysis on voided urine for surveillance of patients with low-grade non-muscle-invasive bladder cancer. Clin Cancer Res 2010;16:3011–8.

47. Brems-Eskildsen AS, Zieger K, Toldbod H, Holcomb C, Higuchi R,Mansilla F, et al. Prediction and diagnosis of bladder cancer recurrencebased on urinary content of hTERT, SENP1, PPP1CA, and MCM5transcripts. BMC Cancer 2010;10:646.

48. Zuiverloon TC, Beukers W, van der Keur KA, Munoz JR, Bangma CH,Lingsma HF, et al. A methylation assay for the detection of non-muscle-invasive bladder cancer (NMIBC) recurrences in voided urine.BJU Int 2012;109:941–8.





2014;20:1978-1989. Clin Cancer Res Sheng-Fang Su, André Luís de Castro Abreu, Yoshitomo Chihara, et al. Sediments Accurately Predicts Bladder Cancer RecurrenceA Panel of Three Markers Hyper- and Hypomethylated in Urine

Updated version

http://clincancerres.aacrjournals.org/content/20/7/1978

Access the most recent version of this article at:

Material

Supplementary

http://clincancerres.aacrjournals.org/content/suppl/2015/07/17/20.7.1978.DC1

Access the most recent supplemental material at:

Cited articles

http://clincancerres.aacrjournals.org/content/20/7/1978.full#ref-list-1

This article cites 46 articles, 12 of which you can access for free at:

Citing articles

http://clincancerres.aacrjournals.org/content/20/7/1978.full#related-urls

This article has been cited by 2 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected]

To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://clincancerres.aacrjournals.org/content/20/7/1978To request permission to re-use all or part of this article, use this link



http://clincancerres.aacrjournals.org/content/suppl/2015/07/17/20.7.1978.DC1

http://clincancerres.aacrjournals.org/content/20/7/1978.full#ref-list-1

http://clincancerres.aacrjournals.org/content/20/7/1978.full#related-urls

http://clincancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]



a panel of three markers hyper- and hypomethylated in ... · 1978 clin cancer res; 20(7) april 1,...

Documents