Urine cell-based DNA methylation classifier for monitoring bladder cancer

Urine cell-based DNA methylation classifier for monitoring bladder cancer Background: Current standard methods used to detect and monitor bladder cancer (BC) are invasive or have low sensitivity. This study aimed to develop a urine methylation biomarker classifier for BC monitoring and validate this classifier in patients in follow-up for bladder cancer (PFBC). Methods: Voided urine samples (N = 725) from BC patients, controls, and PFBC were prospectively collected in four centers. Finally, 626 urine samples were available for analysis. DNA was extracted from the urinary cells and bisulfite modificated, and methylation status was analyzed using pyrosequencing. Cytology was available from a subset of patients (N = 399). In the discovery phase, seven selected genes from the literature (CDH13, CFTR, NID2, SALL3, TMEFF2, TWIST1, and VIM2) were studied in 111 BC and 57 control samples. This training set was used to develop a gene classifier by logistic regression and was validated in 458 PFBC samples (173 with recurrence). Results: A three-gene methylation classifier containing CFTR, SALL3, and TWIST1 was developed in the training set (AUC 0.874). The classifier achieved an AUC of 0.741 in the validation series. Cytology results were available for 308 samples from the validation set. Cytology achieved AUC 0.696 whereas the classifier in this subset of patients reached an AUC 0.768. Combining the methylation classifier with cytology results achieved an AUC 0.86 in the validation set, with a sensitivity of 96%, a specificity of 40%, and a positive and negative predictive value of 56 and 92%, respectively. Conclusions: The combination of the three-gene methylation classifier and cytology results has high sensitivity and high negative predictive value in a real clinical scenario (PFBC). The proposed classifier is a useful test for predicting BC recurrence and decrease the number of cystoscopies in the follow-up of BC patients. If only patients with a positive combined classifier result would be cystoscopied, 36% of all cystoscopies can be prevented. Keywords: Cytology, Biomarkers, Bladder cancer, DNA methylation, Non-invasive diagnosis, Urine Background pathological characteristics, non-muscle-invasive bladder Seventy to 80% of patients with bladder cancer (BC) cancer (NMIBC) patients can be classified into three present with non-muscle-invasive tumors, either con- different prognostic groups [1]. A minority of patients fined to the mucosa [stage Ta and carcinoma in situ (20–30%) have low-risk tumors with a recurrence rate of (CIS)] or submucosa (stage T1). Based on clinical and 20–30%, without progression. The second and also the largest group, the intermediate-risk group, consists of patients who frequently develop a non-muscle-invasive * Correspondence: LMENGUAL@clinic.cat Antoine G. van der Heijden and Lourdes Mengual contributed equally to recurrence (40–60%) but seldom progress to this work. muscle-invasive disease. Finally, a small group of pa- Laboratory and Department of Urology, Hospital Clinic of Barcelona, tients has a relatively aggressive NMIBC at presentation. IDIBAPS, University of Barcelona, Barcelona, Spain Hospital Clínic de Barcelona, Centre de Recerca Biomèdica CELLEX, office The 5-year recurrence rate in this group is as high as B22, C/Casanova, 143, 08036 Barcelona, Spain 68% despite maximum intravesical treatment. Full list of author information is available at the end of the article © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. van der Heijden et al. Clinical Epigenetics (2018) 10:71 Page 2 of 10 Furthermore, up to 34% of these high-risk patients will four different centers [Hospital Clínic of Barcelona develop muscle-invasive bladder cancer (MIBC) [2]. For (Spain); Radboud University Medical Center in Nijmegen this reason, an intensive follow-up schedule is (The Netherlands); St. John Emergency Hospital, mandatory in patients with intermediate- or high-risk Bucharest (Romania); MD Anderson Cancer Center, NMIBC. Houston, Texas, (USA)], from October 2010 to February The follow-up schedule consists of urethrocystoscopy 2012. Participating centers were asked to collect and and urine cytology. Depending on the patient’s risk pro- prepare the cell pellet by urine centrifugation and freeze file, the European Association of Urology guidelines rec- them for a final processing at the Hospital Clinic of ommend up to 15 urethrocystoscopies during the first Barcelona or Radboud University Medical Center, 5 years of follow-up [3]. Nijmegen. We took a two-stage approach with a discovery Urethrocystoscopy is considered the gold standard, phase (or training set) and a validation phase (or testing but is invasive, expensive, and moreover misses up to set) (Additional file 1: Figure S1). In the discovery phase, 15% of the papillary and up to 30% of the flat recur- the inclusion criteria for the cases were patients of both rences [4, 5]. Urine cytology, on the other hand, has a sexes, 18 years of age or older, and patients with histo- high specificity (SP) but lacks sensitivity (SN) especially pathological confirmation of BC at any grade or stage. in low-risk tumors [6]. Additionally, the interobserver Without being mandatory, we recommended patients to and intraobserver reproducibility of cytology is poor [7]. have cytology at cystoscopy or during the period between Recently, several non-invasive methods, NMP-22, blad- cystoscopy and surgery. Patients with a prior endovesical der tumor antigen, and UroVysion FISH, have shown to chemotherapeutic or immunotherapeutic treatment could help increase the sensitivity of urine cytology. However, be included. The exclusion criteria were the absence of due to limited specificity or sensitivity, the markers pro- histological confirmation of BC and patients with other posed to date have not been widely adopted in daily clin- urological malignancies (prostate, kidney, urinary tract tu- ical practice. Therefore, there is a clear clinical need to mors). The inclusion criteria for the controls were patients find reliable markers to monitor the recurrence in of both sexes, 18 years of age or older, and with NMIBC [8]. non-malignant urologic pathology (infection, lithiasis, DNA methylation has been recognized to be import- urinary incontinence, BPH) or non-urologic pathology. ant in developmental biology and cancer etiology in gen- The exclusion criterion for controls was a histological eral [9]. DNA methylation occurs principally at CpG confirmation of any urological malignancy. dinucleotides. These CpG dinucleotides are distributed The validation phase was designed as a cross-sectional throughout the genome, and the majority is normally study including PFBC, i.e., the indicated population for methylated. Some regions in the genome have a high the test in daily clinical practice. For efficiency reasons, CpG density and are called CpG islands. Hypermethyla- we oversampled patients with a recurrence because we tion of normally unmethylated CpG islands in the pro- focused on sensitivity instead of specificity (see also the moter regions of tumor suppressor genes represses its results in the “Reducing the number of follow-up cystos- transcription in human tumors [9, 10]. Therefore, aber- copies by using the three-gene methylation/cytology rant DNA methylation is a potential biomarker for diag- combined classifier” section). PFBC with a prior endove- nosis, prognosis, and monitoring of disease after therapy sical chemotherapeutic or immunotherapeutic treatment [11]. Recently, it was shown that the combination of could be included. Without being mandatory, we recom- SOX1, IRAK3,and L1-MET provides better resolution mended PFBC to have cytology at cystoscopy. The ex- than cytology and cystoscopy in the detection of early clusion criterion was a histological confirmation of any recurrence [12]. The objective of the present study is to other urological malignancy. investigate whether a set of methylation markers can A total number of 725 voided urine samples were pro- lead to the development of a voided urine test that pre- spectively collected by the four participating institutions. dicts tumor presence and may be used to stratify BC pa- From the total number of urines collected, 99 (13%) tients according to their risk of recurrence, thus were excluded from the study because of technical prob- allowing the reduction of the number of cystoscopies in lems during the sample collection, storage, or analysis. the follow-up of BC. Finally, 626 urines were used: 111 from BC patients and 57 from controls for the discovery phase and 458 from Methods PFBC (of whom 173 had a recurrence) for the validation Patients and clinical samples phase (Tables 1 and 2). The grade and stage of the tu- After Institutional Review Board approval and obtaining mors were determined according to WHO 2004 criteria patients’ informed consents, we prospectively collected and TNM 2002 classification, respectively [13, 14]. freshly voided urine samples from BC patients, controls, Tumors were classified according to their risk of re- and patients in follow-up for bladder cancer (PFBC) at currence and progression into three categories: high-risk van der Heijden et al. Clinical Epigenetics (2018) 10:71 Page 3 of 10 Table 1 Clinicopathological and demographic characteristics of Table 1 Clinicopathological and demographic characteristics of the study population classified by the study phase the study population classified by the study phase (Continued) Discovery phase Validation phase Discovery phase Validation phase Training set Testing set Training set Testing set N bladder cancer (%) N R-PFBC (%) T1 LG – 22 (8) Gender T1 HG – 82 (29) Male 86 (77) 135 (78) T2 HG – 3 (1) Female 25 (23) 38 (22) Tx LG – 2 (1) Age Tx HG – 2 (1) Mean 72 68 Subtotals 285 Range 39–98 26–99 Total 168 458 LG low-grade, HG high-grade, TURBT transurethral resection bladder tumor, Stage and grade BPH benign prostate hyperplasia, CIS/Tis carcinoma in situ, BC bladder cancer, Tis 7 (6) 13 (8) R-PFBC recurrent patients in follow-up for bladder cancer, NR-PFBC non- recurrent patients in follow-up for bladder cancer Ta LG 26 (23) 61 (35) Ta HG 11 (10) 20 12) (HR) NMIBC (any of the following: T1, HG/G3 tumors, T1 LG 20 (18) 35 (20) or CIS), non-high-risk (nHR) NMIBC (all the other cases of NMIBC), and muscle-invasive bladder cancer (MIBC) T1 HG 22 (20) 44 (25) (T2–4). None of the included patients had an upper > T2 LG 1 (1) – urinary tract tumor. > T2 HG 24 (22) – Subtotals 111 173 Urine sample processing N control (%) Urine samples were collected before cystoscopy, the day Gender before the transurethral resection of the bladder tumor (TURBT), or the day before cystectomy. From all pa- Male 29 (51) – tients and controls, only one single sample was included. Female 28 (49) – Age For urine cytology Mean 60 – Urine cytology was performed according to Papanico- Range 22–82 – laou staining and evaluated by expert pathologists in Urinary condition each center blinded to the patient’s clinical history. The results were considered as positive or negative. No cen- BPH 11 (19) – tral cytology review was performed. Urolithiasis 13 (23) – Incontinence 2 (4) – For methylation studies Benign bladder disease 1 (2) – Voided urine samples (50 to 100 ml) were collected in Urinary tract infections 12 (21) – sterile containers containing 4 ml of 0.5 M EDTA, pH Non-urological diseases 18 (32) – 8.0. Urines were immediately stored at 4 °C and proc- essed within the next 24 h. The samples were centri- Subtotals 57 – fuged at 1000×g for 10 min, at 4 °C. The cell pellets N NR-PFBC (%) were frozen at − 80 °C. Gender Male – 217 (76) DNA isolation, bisulfite treatment, and PCR Female – 68 (24) DNAs from the urinary cell pellets were extracted using Age QIAamp DNA Mini Kit (Qiagen) according to the manufacturer’s instructions and quantified with a Nano- Mean – 69 Drop1000 (NanoDrop Technologies, Wilmington, DE, Range – 26–92 USA). DNA extraction was performed in each center ex- Stage and grade previous TURBT cept for Bucharest, whose cell pellets were sent in dry Tis – 21 (7) ice to the Radboud University Medical Center, Nijmegen Ta LG – 100 (35) (The Netherlands) for DNA extraction. Ta HG – 53 (19) One microgram of genomic DNA was used for the bi- sulfite modification using EpiTect Bisulfite kit (Qiagen, van der Heijden et al. Clinical Epigenetics (2018) 10:71 Page 4 of 10 Table 2 Clinicopathological and demographic characteristics of the study population classified by the participating center Hospital Clinic Barcelona Radboud University Saint John Emergency MD Anderson Cancer Total Medical Center, Nijmegen Clinical Hospital Bucharest Center Houston Discovery phase Bladder cancer urine samples Stage Tis 5 – 11 7 Ta 14 5 5 13 37 T1 19 2 21 – 42 >T2 9 – 16 – 25 Grade LG 20 2 17 8 47 HG 27 5 26 6 64 Subtotal 47 7 43 14 111 Control urine samples BPH 8 3 –– 11 Urolithiasis 13 –– – 13 Incontinence 1 1 –– 2 Benign bladder 45 1 – 10 disease Urinary tract 18 4 – 13 infection Non-urological 6 –– 28 diseases Subtotal 33 17 5 2 57 Validation phase R-PFBC URINE SAMPLES Stage Tis 2 1 8 2 13 Ta 13 15 47 6 81 T1 4 2 55 18 79 Grade LG 14 8 74 – 96 HG 5 10 36 26 77 Subtotal 19 18 110 26 173 NR-PFBC urine samples Stage previous TURBT Tis 5 6 6 4 21 Ta 32 56 36 26 150 Ta + CIS 1 –– 23 T1 36 16 25 19 96 T1 + CIS 5 –– 38 T2 3 –– – 3 Tx 4 –– – 4 Grade previous TURBT LG 34 35 38 19 126 HG 52 43 29 35 159 Subtotal 86 78 67 54 285 TOTAL 185 120 225 96 626 LG low-grade, HG high-grade, TURBT transurethral resection bladder tumor, BPH benign prostate hyperplasia, CIS/Tis carcinoma in situ, BC bladder cancer, R-PFBC recurrent patients in follow-up for bladder cancer, NR-PFBC non-recurrent patients in follow-up for bladder cancer. van der Heijden et al. Clinical Epigenetics (2018) 10:71 Page 5 of 10 Inc.) following the manufacturer’s instructions. The inclusion criterion was p value of < 0.1. Risk probability modified DNA was eluted with 20 μl Tris-HCL (1 mM, of presenting BC was calculated in the training set. We pH 8.0) and stored at − 80 °C before further processing. established cutoff point value (≥ 0.464) allowing 15% Bisulfite modifications were performed in Hospital false negatives in the tumor group (SN = 85%). In the Clinic of Barcelona, Spain (training set), and in Radboud subset of samples in which cytology results were avail- University Medical Center, Nijmegen, The Netherlands able, the cutoff point value that yielded 85% SN in the (testing set). training set was 0.688, and in the combined model A total of seven DNA methylation markers, i.e., (methylation test + cytology results), it was 0.617. If the CDH13, CFTR, NID2, SALL3, TMEFF2, TWIST1, and predicted probability value derived from each classifier VIM2, were selected from four recently published BC in each of the samples was higher than the cutoff point studies [15–18]. The sequences of the primers used to value, the samples were classified for each of the classi- amplify the regions of interest of these genes and the fiers as tumor sample. The performance of the models PCR conditions are shown in Additional file 2: Table S1. was evaluated in a testing set by means of AUROC using PCR primers were designed using the PyroMark Assay pROC R-package [19]. Student’s t test was used to evalu- Design software v2.0 (Qiagen). PCR was performed in a ate statistical differences in DNA methylation. Statistical volume of 25 μl with 2 μl of converted genomic DNA, significance was established at p value of 0.05. 0.6 U Ampli Taq Gold 360 DNA polymerase (Thermo- R-software and SPSS v23.0 were used for calculations. fisher), 0.8 μl of a mix of Primer-F and biotinylated Primer-R at 10 μM, 2 μl MgCl 25 mM, and 0.5 μl Results dNTPs 10 mM. Amplification was performed according Training set to the following thermocycling conditions: denaturation DNA methylation of all seven selected genes was signifi- at 95 °C for 10 min, followed by 45 cycles of 95 °C for cantly increased in urine sediments from BC patients 30 s, the optimal Tm for 30 s, and 72 °C for 1 min; and compared to controls (Fig. 1). To determine the combin- a final extension at 72 °C for 7 min. The formation of ation of markers capable of detecting BC in urine PCR products with accurate size was confirmed by sediments with the highest accuracy, we built a model of resolving PCR samples (1 μl) by 2% agarose gel elec- multiple markers by logistic regression. The best pos- trophoresis, with visualization by ethidium bromide sible biomarker combination based on AUC was pro- staining. vided by the combination of CFTR, SALL3, and TWIST1. This three-gene methylation classifier achieved Pyrosequencing for quantitative methylation an AUC = 0.874 (Fig. 2a); at a fixed overall SN of 85%, Biotin-labeled single-stranded amplicons were isolated the classifier provides a SP of 68%. Moreover, the SN of from 20 μl of the PCR product according to the protocol the three-gene methylation classifier increases through using the Pyromark Q96 Work Station and pyrose- the BC risk groups and grading (Table 3). quenced with 0.3 μM sequence primer using PSQ96MD System (Biotage AB). Additional file 2: Table S1 shows Testing set the sequences of the primer sets used for bisulfite se- To examine whether the three-gene methylation BC quencing. The percent methylation for each of the CpGs diagnostic classifier was able to identify recurrences in a was calculated using PyroQ CpG Software (Qiagen). The clinical setting, the classifier was validated in an inde- differences in the percentage of methylation were calcu- pendent multicenter international series of 458 urine lated between BC vs. control and recurrent vs. sediments from patients in follow-up for bladder cancer non-recurrent PFBC (R-PFBC and NR-PFBC, respect- (PFBC), of whom 173 had a recurrence. Recurrent PFBC ively) samples. A high correlation in the methylation (R-PFBC; N = 173) displays higher percentages of DNA percentages of the CpG dinucleotides in the same is- methylation compared with non-recurrent PFBC land was observed (Additional files 3 and 4:Figures (NR-PFBC; N = 285) (Fig. 3). SN of the three-gene S2 and S3). For this reason, hypermethylation was an- methylation BC classifier increased in the validation alyzed in all genes at the first CpG dinucleotide series (SN = 90%), while SP drops (SP = 31%), as evi- present (Additional file 5:Table S2). denced by the ROC curves and AUC value (AUC = 0.741) (Table 3 and Fig. 2a). Figure 4a depicts the risk Data analysis probabilities derived from the three-gene methylation Univariable and multivariable logistic regression analyses classifier in R-PFBC and NR-PFBC. were used to examine the associations between BC and DNA methylation status of urinary sediments. A forward Comparison of test performance with urine cytology stepwise logistic regression was performed to determine A total of 399 urine cytologies (91 from the training and the best classifier between BC and control samples. The 308 from the testing set) were performed. In both van der Heijden et al. Clinical Epigenetics (2018) 10:71 Page 6 of 10 Fig. 1 Percentage of DNA methylation in bladder cancer and control urine sediments for the seven selected genes analyzed in the discovery phase. The number of samples in each group is given in brackets. Abbreviations: BC, bladder cancer; C, control training and testing set, SN of the three-gene methyla- three-gene methylation classifier than that of the urine tion classifier (86 and 93%, respectively) was higher than cytology in training (40 and 29%, respectively) and that of the urine cytology (54 and 46%, respectively) testing set (82 and 69%, respectively). Contrary, posi- in this subset of samples (Additional files 6 and 7: tive predictive value (PPV) is higher for the urine cy- Table S3 and Figure S4). In the testing set, this means tology than for the methylation classifier in training that 55% of the recurrences (68 out of 124) were (98 and 89%, respectively) and testing set (85 and detected by the three-gene classifier but were missed by 50%, respectively). Cytology had a SP of 93 and 94% urine cytology. On the other hand, 12 recurrences were while the three-gene methylation classifier achieved a missed by the three-gene classifier of which half was SP, in this subset of samples, of 47 and 27%, in the training detected by cytology (Additional file 8: Figure S5). and testing set, respectively. In the testing set, 120 Negative predictive value (NPV) is also higher for the NR-PFBC samples were positive by the three-gene Fig. 2 ROC curves in the training and testing series for (a) the three-gene BC methylation classifier and (b) the combined three-gene methylation/cytology classifier van der Heijden et al. Clinical Epigenetics (2018) 10:71 Page 7 of 10 Table 3 Diagnostic performance of the three-gene methylation methylation classifier. Nine of them also had positive classifier in the training and testing set of samples (at fixed urine cytology (Additional file 8:Figure S5). After 1year, sensitivity of 85% in the training set) two NR-PFBC with a positive test (who had a negative cy- Training set Testing set tology) had a tumor recurrence. Overall N samples 111 BC/57 C 173 R-PFBC/285 NR-PFBC Combination of the three-gene methylation classifier with AUC 0.874 0.741 urine cytology Combining the three-gene methylation classifier and SN (%) 84.68 89.6 urine cytology results showed an improved diagnostic SP (%) 68.42 30.53 performance in the training (AUC 0.858) and as well as PPV (%) 83.93 43.91 in the testing set (AUC 0.86) (Fig. 2b). A SN of 96% and NPV (%) 69.64 82.86 a NPV of 92% are achieved in the testing set. Of note, in Non-high-risk NMIBC HG tumors, a 100% SN and NPV are achieved N samples 26 BC/57 C 61 R-PFBC/285 NR-PFBC (Additional file 6: Table S3). The risk probabilities derived from the combined classifier in R-PFBC and SN (%) 73.08 88.52 NR-PFBC are shown in Fig. 4b. SP (%) 68.42 30.53 PPV (%) 51.35 21.43 Reducing the number of follow-up cystoscopies by using NPV (%) 84.78 92.55 the three-gene methylation/cytology combined classifier High-risk NMIBC In our study, we oversampled patients with BC. We N samples 60 BC/57 C 112 R-PFBC/285 NR-PFBC therefore cannot directly calculate predictive values SN (%) 86.67 90.18 from the test results. In the hospitals participating in SP (%) 68.42 30.53 thestudy,recurrenceisdetectedinapproximately 10% of all follow-up cystoscopies performed (90% of PPV (%) 74.29 33.78 patients previously diagnosed with NMIBC are NPV (%) 82.98 88.78 without recurrence at the time of follow-up cystoscopy). MIBC In order to calculate the predictive values that reflect N samples 25 BC/57C – values in real clinical practice, we assumed the distribution SN (%) 92 – of recurrent vs. non-recurrent to be 10 vs. 90%. For this, SP (%) 68.42 – we multiplied the NR-PFBC samples by 7. Using the SN and SP that we found in the study, the PPV and NPV in PPV (%) 56.1 – the validation phase become 15 and 99%, respectively. If NPV (%) 95.12 – patients with a negative classifier will not undergo a cyst- Low-grade oscopy, this means that more than a third (~ 36%) of all N samples 47 BC/57 C 96 R-PFBC/285 NR-PFBC cystoscopies can be prevented at the cost of 4% of recur- SN (%) 76.6 90.62 rences remaining undiagnosed which all were LG tumors. SP (%) 68.42 30.53 PPV (%) 66.67 30.53 Discussion NPV (%) 78 90.62 In the present study, a set of DNA methylation markers to predict the presence of bladder cancer (BC) in urine High grade samples has been selected. The best possible marker N samples 64 BC/57 C 77 R-PFBC/285 NR-PFBC combination to discriminate BC from controls was the SN (%) 90.62 88.31 combination CFTR, SALL3, and TWIST1. We confirmed SP (%) 68.42 30.53 that these genes (and specifically CpG dinucleotides PPV (%) 76.32 25.56 analyzed here) are hypermethylated in the bladder can- NPV (%) 86.67 90.62 cer tissue using methylation data from the TCGA Research Network [20] (Additional file 9: Figure S6). LG low-grade, HG high-grade, AUC area under the curve, MIBC muscle-invasive bladder cancer, NMIBC non-muscle invasive bladder cancer, NPV negative This supports their use as diagnostic markers in urine predictive value, PPV positive predictive value, SN sensitivity, SP specificity, BC samples. In the training set, the three-gene methylation bladder cancer, C control, R-PFBC recurrent patients in follow-up for bladder cancer, NR-PFBC non-recurrent patients in follow-up for bladder cancer. classifier achieved an AUC 0.874 while in the testing set, an AUC 0.741 was achieved to discriminate recurrent from non-recurrent patients in follow-up for bladder cancer (PFBC). These results improved significantly in van der Heijden et al. Clinical Epigenetics (2018) 10:71 Page 8 of 10 Fig. 3 Percentage of DNA methylation in recurrent and non-recurrent PFBC urine sediments for the three genes of the classifier in the validation phase. The number of samples in each group is given in brackets. Abbreviations: NR-PFBC, non-recurrent patients in follow-up for bladder cancer; R-PFBC, recurrent patients in follow-up for bladder cancer the testing set when cytology results were included in [21]. However, the results were significantly better in the analysis (AUC 0.86). thesubgroupof activesmokers.Unfortunately,we did TWIST1 hypermethylation in BC was first described not collect information about tobacco smoking, and by Renard and co-workers [16]. In a case-control study therefore, we cannot perform a subset analysis for (n = 145 cases/321 controls), they detected TWIST1 and smoking behavior. NID2 hypermethylation in urine sediments of BC pa- Yu and co-workers previously found in a case-control tients using methylation-specific PCR. This two-gene study that CFTR and SALL3, out of 59 genes that were panel achieved a SN of 90%, SP of 93%, PPV of 86%, screened, were the most frequently methylated genes to and NPV of 95%. Nevertheless, the group of Fantony predict the presence of BC in urine [15]. However, in published conflicting results. They found only a SN of this study, no PFBC were included. In daily clinical 67% and SP of 69% for this two-gene urine panel practice, this group is especially of interest. It is not very ab Fig. 4 Box plots showing the individual risk probabilities derived from (a) the three-gene methylation classifier and (b) the combined three-gene methylation/cytology classifier for recurrent and non-recurrent PFBC in the cross-sectional study. Dots above the cutoff value (dashed line) denote positive samples, whereas those below signify negatives scores. Cutoff values for the three-gene methylation classifier and the combined three-gene methylation/cytology classifier are ≥ 0.464 and ≥ 0.617, respectively. Abbreviations: NR-PFBC, non-recurrent patients in follow-up for bladder cancer; R-PFBC, recurrent patients in follow-up for bladder cancer van der Heijden et al. Clinical Epigenetics (2018) 10:71 Page 9 of 10 likely that biomarkers will replace ureterocystoscopy in validation series, and we had to make an estimation to patients with macroscopic hematuria referred to the ur- calculate the number of cystoscopies that could be ologist. But in PFBC, urinary biomarkers with a high SN skipped. Secondly, 13% of the samples had to be ex- and high NPV could make a difference, i.e., lower the cluded due to technical failures. Thirdly, intravesical number of follow-up cystoscopies. Our combined three-gene treatments in NMIBC patients may have influenced methylation/cytology classifier achieves a high SN and NPV, methylation patterns. Finally, we have evaluated only a for both high-risk and non-high-risk NMIBC. Consequently, limited number of hypermethylated genes with diagnos- more than a third of all cystoscopies could be prevented. tic value previously described in the literature. The SP of the combined classifier in the testing set, using PFBC samples, is expected to be lower than the SP Conclusions of the training set, using BC and control samples. In conclusion, this combined three-gene methylation/cy- Possible reasons for methylation observed in PFBC sam- tology classifier can reduce the number of follow-up ples are small tumors not yet detected by cystoscopy, re- cystoscopies in PFBC. This approach may improve the sidual tumor cells at the resection site, or epigenetically patients’ quality of life. For a definitive conclusion, repli- changed urothelial cells at the resection site or else in cation of the classifier in another series of patients and the bladder also known as epigenetic field defect [22]. In cost-effectiveness studies are needed. patients with persistent hypermethylation, which does not recur within 18 months, the presence of an Additional files epigenetic field defect is most likely. Wolff and co-workers suggested that the aberrant methylation is Additional file 1: Figure S1. Flowchart of the entire study. A total of caused by a generalized epigenetic alteration in the seven hypermethylated genes, differentially expressed between BC patients whole bladder urothelium, and this widespread methyl- and controls (n = 168), were determined in the discovery phase. With these results, a three-gene methylation classifier was developed. This three-gene ated urothelium may be the cause of the high recurrence classifier was tested in a cross-sectional study (validation phase; n =458). rate in NMIBC [22]. Samples with available cytology results in each phase are indicated. Abbrevia- The discovery of highly sensitive methylation markers tions: BC, bladder cancer; C, control; R-PFBC, recurrent patients in follow-up for bladder cancer; NR-PFBC, non-recurrent patients in follow-up for bladder allows us to lower the number of follow-up cystoscopies cancer. (PPTX 75 kb) in more than a third of all follow-up patients. In the Additional file 2: Table S1. Primer sequences used in PCR and clinical situation, patients supply a urine sample before pyrosequencing. (DOCX 13 kb) cystoscopy to perform the test. If the combined test is Additional file 3: Figure S2. Pearson correlation coefficient heat map positive, patients will undergo a cystoscopy. In our of the percentage of methylation for every CpG dinucleotide in the seven genes. Every CpG sites are correlated (via the Pearson correlation) validation series, 60% of NR-PFBC had a positive com- with all others. Correlations are scaled by the color of the corresponding bined test and should undergo a cystoscopy. This is not cell. Parameters are represented in the same order on the x- and y-axes. a major problem since in normal daily practice, (PPTX 1232 kb) follow-up patients would have undergone a cystoscopy Additional file 4: Figure S3. Representative pyrograms showing gene methylation patterns in DNA urine samples from a bladder cancer anyway. If the combined test is negative, cystoscopy patient. Percentage of methylation is indicated above peaks (gray could be skipped. columns) corresponding to the CpG sites in this region. (PPTX 183 kb) However, a methylation test has also financial and lo- Additional file 5: Table S2. Percentage of methylation for each CpG gistic implications, which means that a cost-effectiveness dinucleotide in the seven selected genes in control and bladder cancer urine samples. Underlined in grey the CpG site used for methylation analysis is necessary. Using our three-gene methylation/ analysis. Abbreviations: SDV; Standard Deviation. (DOCX 21 kb) cytology classifier, 4% of PFBC are wrongly diagnosed as Additional file 6: Table S3. Diagnostic performance of the three-gene not having a recurrence; all of them had LG NMIBC. Of methylation classifier, cytology, and the combined methylation/cytology note, cystoscopy, our gold standard, misses up to 15% of classifier in the training and testing subset of samples with cytology available. Abbreviations: LG, Low Grade; HG, High Grade; AUC, area under the papillary and up to 30% of the flat lesions [4]. curve; MIBC, muscle invasive bladder cancer; NMIBC, Non-Muscle Invasive The strengths of this study lie in the fact that we have Bladder Cancer; NPV, Negative Predictive Value; PPV, Positive Predictive chosen for a two-stage approach using PFBC in the val- Value; SN, Sensitivity; SP, Specificity; BC, Bladder Cancer; C, Control; R- PFBC, Recurrent Patients in Follow up for Bladder Cancer; NR-PFBC, Non idation phase. Furthermore, the use of voided urine Recurrent Patients in Follow up for Bladder Cancer. (DOCX 20 kb) samples to analyze the DNA methylation status allows Additional file 7: Figure S4. Sensitivity, negative and positive predictive the development of a non-invasive BC diagnostic tool values of urine cytology, the three-gene methylation classifier, and the with an easy translation into clinical practice. However, combined three-gene methylation/cytology classifier in the testing set (N = 308). Overall specificity was 94% for urine cytology, 27% for the three-gene some limitations should be mentioned. To avoid ineffi- methylation classifier, and 40% for the combined three-gene methylation/ ciency, patients with a recurrence were oversampled by cytology classifier. Abbreviations: LG, low-grade; HG, high-grade; NMIBC also recruiting patients who were scheduled for a nHR, non-muscle-invasive bladder cancer non-high risk; NMIBC HR, non-- muscle-invasive bladder cancer high risk; NPV, negative predictive value; TURBT of a proven bladder tumor. Consequently, the PPV, positive predictive value. (PPTX 189 kb) number of NR-PFBC was misrepresented in the van der Heijden et al. Clinical Epigenetics (2018) 10:71 Page 10 of 10 Received: 1 December 2017 Accepted: 3 May 2018 Additional file 8: Figure S5. Flow diagram of participants in the cross-sectional study according a) to the three-gene methylation classifier and cytology results and b) to the combined three-gene methylation/cytology References classifier. Abbreviations: R-PFBC, recurrent patients in follow-up for bladder 1. Oosterlinck W, Lobel B, Jakse G, et al. Guidelines on bladder cancer. Eur cancer; NR-PFBC, non-recurrent patients in follow-up for bladder cancer; Cytol, Urol. 2002;41:105–12. cytology; NA, non-available; Test, combined three-gene methylation/cytology 2. Fernandez-Gomez J, Madero R, Solsona E, et al. Predicting nonmuscle invasive classifier. (PPTX 85 kb) bladder cancer recurrence and progression in patients treated with bacillus Additional file 9: Figure S6. DNA methylation profiles for bladder Calmette-Guerin: the CUETO scoring model. J Urol. 2009;182:2195–203. cancer and control tissue samples for the three-gene classifier. Data obtained 3. Babjuk M, Bohle A, Burger M, et al. EAU Guidelines on Non-Muscle-invasive from Wanderer Web page: http://maplab.imppc.org/wanderer/.The Urothelial Carcinoma of the Bladder: update 2016. Eur Urol. 2016;71:447–61. red arrow indicates the CpG dinucleotide analyzed in each of the 4. Grossman HB, Gomella L, Fradet Y, et al. A phase III, multicenter comparison three genes. (PPTX 203 kb) of hexaminolevulinate fluorescence cystoscopy and white light cystoscopy for the detection of superficial papillary lesions in patients with bladder cancer. J Urol. 2007;178:62–7. Abbreviations 5. Fradet Y, Grossman HB, Gomella L, et al. A comparison of AUC: Area under the curve; BC: Bladder cancer; BPH: Benign prostatic hexaminolevulinate fluorescence cystoscopy and white light cystoscopy for hyperplasia; C: Control; CIS/Tis: Carcinoma in situ; HG: High-grade; HR: High-risk; the detection of carcinoma in situ in patients with bladder cancer: a phase LG: Low-grade; MIBC: Muscle-invasive bladder cancer; NA: Not available; III, multicenter study. J Urol. 2007;178:68–73. nHR: Non-high risk; NMIBC HR: Non-muscle-invasive bladder cancer high risk; 6. Brown FM. Urine cytology. It is still the gold standard for screening? Urol NMIBC nHR: Non-muscle-invasive bladder cancer non-high risk; NMIBC: Non- Clin North Am. 2000;27:25–37. muscle-invasive bladder cancer; NPV: Negative predictive value; NR-PFBC: Non- 7. Sherman AB, Koss LG, Adams SE. Interobserver and intraobserver differences recurrent patients in follow-up for bladder cancer; PPV: Positive predictive value; in the diagnosis of urothelial cells. Comparison with classification by R-PFBC: Recurrent patients in follow-up for bladder cancer; SN: Sensitivity; computer. Anal Quant Cytol. 1984;6:112–20. SP: Specificity; TURBT: Transurethral resection of the bladder tumor 8. Parker J, Spiess PE. Current and emerging bladder cancer urinary biomarkers. ScientificWorldJournal. 2011;11:1103–12. Acknowledgements 9. Esteller M. Epigenetics in cancer. N Engl J Med. 2008;358:1148–59. We thank the technical support from the staff of the Servei Veterinari de 10. Saxonov S, Berg P, Brutlag DL. A genome-wide analysis of CpG Genètica Molecular, Facultat de Veterinària, Universitat Autònoma de Barcelona. dinucleotides in the human genome distinguishes two distinct classes of Part of the work was developed at the building Centre de Recerca Biomèdica promoters. Proc Natl Acad Sci U S A. 2006;103:1412–7. Cellex, Barcelona. Furthermore, funding from CERCA Programme/Generalitat de 11. Shames DS, Minna JD, Gazdar AF. DNA methylation in health, disease, and Catalunya is gratefully acknowledged. cancer. Curr Mol Med. 2007;7:85–102. 12. Su SF, Castro Abreu AL, Chihara Y, et al. A panel of three markers hyper- and hypomethylated in urine sediments accurately predicts bladder cancer Funding recurrence. Clin Cancer Res. 2014;20:1978–89. This work was supported by grants from the Dutch Cancer Society. 13. Sobin LH, Wittekind CH. TNM classification of malignant tumours. International union against cancer. 6th ed. New York: Wiley; 2002. Availability of data and materials 14. Lopez-Beltran A, Sauter G, Gasser T, et al. Tumours of the urinary system. In: Please contact the corresponding author for data requests. Eble JN, Sauter G, Epstein JI, Sesterhenn IA, editors. Pathology and genetics of tumours of the urinary system and male genital organs. World Health Authors’ contributions Organization classification of tumours. Lyon: IARC Press; 2004. p. 89–157. AGvdH, LM, LALMK, MJR, JAW, and AA contributed to the conception and 15. Yu J, Zhu T, Wang Z, et al. A novel set of DNA methylation markers in urine design. AGvdH, LM, MI-T, CCMvR-vdW, MB, LALMK, MJR, JAW, and AA sediments for sensitive/specific detection of bladder cancer. Clin Cancer contributed to the methodology, collection, and assembly of data. All Res. 2007;13:7296–304. authors contributed to the data analysis and interpretation and manuscript 16. Renard I, Joniau S, van Cleynenbreugel B, et al. Identification and validation writing and reviewing. All authors approved the final manuscript. of the methylated TWIST1 and NID2 genes through real-time methylation- specific polymerase chain reaction assays for the noninvasive detection of primary bladder cancer in urine samples. Eur Urol. 2010;58:96–104. Ethics approval and consent to participate 17. Brait M, Begum S, Carvalho AL, et al. Aberrant promoter methylation of The study was approved by the Institutional Review Boards of the Hospital multiple genes during pathogenesis of bladder cancer. Cancer Epidemiol Clínic of Barcelona (Spain); Radboud University Medical Center in Nijmegen Biomark Prev. 2008;17:2786–94. (The Netherlands), St. John Emergency Hospital, Bucharest (Romania); and 18. Costa VL, Henrique R, Danielsen SA, et al. Three epigenetic biomarkers, MD Anderson Cancer Center, Houston, Texas, (USA). Patients’ informed GDF15, TMEFF2, and VIM, accurately predict bladder cancer from DNA- consents were obtained before the sample collection. based analyses of urine samples. Clin Cancer Res. 2010;16:5842–51. 19. Robin X, Turck N, Hainard A, et al. pROC: an open-source package for R and Competing interests S+ to analyze and compare ROC curves. BMC Bioinformatics. 2011;12:77. The authors declare that they have no competing interests. 20. Diez-Villanueva A, Mallona I, Peinado MA. Wanderer, an interactive viewer to explore DNA methylation and gene expression data in human cancer. Epigenetics Chromatin. 2015;8:22. Publisher’sNote 21. Fantony JJ, Abern MR, Gopalakrishna A, et al. Multi-institutional external Springer Nature remains neutral with regard to jurisdictional claims in validation of urinary TWIST1 and NID2 methylation as a diagnostic test for published maps and institutional affiliations. bladder cancer. Urol Oncol. 2015;33:387–6. 22. Wolff EM, Chihara Y, Pan F, et al. Unique DNA methylation patterns Author details distinguish noninvasive and invasive urothelial cancers and establish an Department of Urology Radboud University Medical Center, Nijmegen, The epigenetic field defect in premalignant tissue. Cancer Res. 2010;70:8169–78. Netherlands. Laboratory and Department of Urology, Hospital Clinic of Barcelona, IDIBAPS, University of Barcelona, Barcelona, Spain. CIBERehd, Plataforma de Bioinformática, Centro de Investigación Biomédica en red de Enfermedades Hepáticas y Digestivas, Barcelona, Spain. Saint John Emergency Clinical Hospital, Bucharest, Romania. MD Anderson Cancer Center, Houston, Texas, USA. Hospital Clínic de Barcelona, Centre de Recerca Biomèdica CELLEX, office B22, C/Casanova, 143, 08036 Barcelona, Spain. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Clinical Epigenetics Springer Journals
Loading next page...
 
/lp/springer_journal/urine-cell-based-dna-methylation-classifier-for-monitoring-bladder-jP4F3mSEya
Publisher
BioMed Central
Copyright
Copyright © 2018 by The Author(s).
Subject
Biomedicine; Human Genetics; Gene Function
ISSN
1868-7075
eISSN
1868-7083
D.O.I.
10.1186/s13148-018-0496-x
Publisher site
See Article on Publisher Site

Abstract

Background: Current standard methods used to detect and monitor bladder cancer (BC) are invasive or have low sensitivity. This study aimed to develop a urine methylation biomarker classifier for BC monitoring and validate this classifier in patients in follow-up for bladder cancer (PFBC). Methods: Voided urine samples (N = 725) from BC patients, controls, and PFBC were prospectively collected in four centers. Finally, 626 urine samples were available for analysis. DNA was extracted from the urinary cells and bisulfite modificated, and methylation status was analyzed using pyrosequencing. Cytology was available from a subset of patients (N = 399). In the discovery phase, seven selected genes from the literature (CDH13, CFTR, NID2, SALL3, TMEFF2, TWIST1, and VIM2) were studied in 111 BC and 57 control samples. This training set was used to develop a gene classifier by logistic regression and was validated in 458 PFBC samples (173 with recurrence). Results: A three-gene methylation classifier containing CFTR, SALL3, and TWIST1 was developed in the training set (AUC 0.874). The classifier achieved an AUC of 0.741 in the validation series. Cytology results were available for 308 samples from the validation set. Cytology achieved AUC 0.696 whereas the classifier in this subset of patients reached an AUC 0.768. Combining the methylation classifier with cytology results achieved an AUC 0.86 in the validation set, with a sensitivity of 96%, a specificity of 40%, and a positive and negative predictive value of 56 and 92%, respectively. Conclusions: The combination of the three-gene methylation classifier and cytology results has high sensitivity and high negative predictive value in a real clinical scenario (PFBC). The proposed classifier is a useful test for predicting BC recurrence and decrease the number of cystoscopies in the follow-up of BC patients. If only patients with a positive combined classifier result would be cystoscopied, 36% of all cystoscopies can be prevented. Keywords: Cytology, Biomarkers, Bladder cancer, DNA methylation, Non-invasive diagnosis, Urine Background pathological characteristics, non-muscle-invasive bladder Seventy to 80% of patients with bladder cancer (BC) cancer (NMIBC) patients can be classified into three present with non-muscle-invasive tumors, either con- different prognostic groups [1]. A minority of patients fined to the mucosa [stage Ta and carcinoma in situ (20–30%) have low-risk tumors with a recurrence rate of (CIS)] or submucosa (stage T1). Based on clinical and 20–30%, without progression. The second and also the largest group, the intermediate-risk group, consists of patients who frequently develop a non-muscle-invasive * Correspondence: LMENGUAL@clinic.cat Antoine G. van der Heijden and Lourdes Mengual contributed equally to recurrence (40–60%) but seldom progress to this work. muscle-invasive disease. Finally, a small group of pa- Laboratory and Department of Urology, Hospital Clinic of Barcelona, tients has a relatively aggressive NMIBC at presentation. IDIBAPS, University of Barcelona, Barcelona, Spain Hospital Clínic de Barcelona, Centre de Recerca Biomèdica CELLEX, office The 5-year recurrence rate in this group is as high as B22, C/Casanova, 143, 08036 Barcelona, Spain 68% despite maximum intravesical treatment. Full list of author information is available at the end of the article © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. van der Heijden et al. Clinical Epigenetics (2018) 10:71 Page 2 of 10 Furthermore, up to 34% of these high-risk patients will four different centers [Hospital Clínic of Barcelona develop muscle-invasive bladder cancer (MIBC) [2]. For (Spain); Radboud University Medical Center in Nijmegen this reason, an intensive follow-up schedule is (The Netherlands); St. John Emergency Hospital, mandatory in patients with intermediate- or high-risk Bucharest (Romania); MD Anderson Cancer Center, NMIBC. Houston, Texas, (USA)], from October 2010 to February The follow-up schedule consists of urethrocystoscopy 2012. Participating centers were asked to collect and and urine cytology. Depending on the patient’s risk pro- prepare the cell pellet by urine centrifugation and freeze file, the European Association of Urology guidelines rec- them for a final processing at the Hospital Clinic of ommend up to 15 urethrocystoscopies during the first Barcelona or Radboud University Medical Center, 5 years of follow-up [3]. Nijmegen. We took a two-stage approach with a discovery Urethrocystoscopy is considered the gold standard, phase (or training set) and a validation phase (or testing but is invasive, expensive, and moreover misses up to set) (Additional file 1: Figure S1). In the discovery phase, 15% of the papillary and up to 30% of the flat recur- the inclusion criteria for the cases were patients of both rences [4, 5]. Urine cytology, on the other hand, has a sexes, 18 years of age or older, and patients with histo- high specificity (SP) but lacks sensitivity (SN) especially pathological confirmation of BC at any grade or stage. in low-risk tumors [6]. Additionally, the interobserver Without being mandatory, we recommended patients to and intraobserver reproducibility of cytology is poor [7]. have cytology at cystoscopy or during the period between Recently, several non-invasive methods, NMP-22, blad- cystoscopy and surgery. Patients with a prior endovesical der tumor antigen, and UroVysion FISH, have shown to chemotherapeutic or immunotherapeutic treatment could help increase the sensitivity of urine cytology. However, be included. The exclusion criteria were the absence of due to limited specificity or sensitivity, the markers pro- histological confirmation of BC and patients with other posed to date have not been widely adopted in daily clin- urological malignancies (prostate, kidney, urinary tract tu- ical practice. Therefore, there is a clear clinical need to mors). The inclusion criteria for the controls were patients find reliable markers to monitor the recurrence in of both sexes, 18 years of age or older, and with NMIBC [8]. non-malignant urologic pathology (infection, lithiasis, DNA methylation has been recognized to be import- urinary incontinence, BPH) or non-urologic pathology. ant in developmental biology and cancer etiology in gen- The exclusion criterion for controls was a histological eral [9]. DNA methylation occurs principally at CpG confirmation of any urological malignancy. dinucleotides. These CpG dinucleotides are distributed The validation phase was designed as a cross-sectional throughout the genome, and the majority is normally study including PFBC, i.e., the indicated population for methylated. Some regions in the genome have a high the test in daily clinical practice. For efficiency reasons, CpG density and are called CpG islands. Hypermethyla- we oversampled patients with a recurrence because we tion of normally unmethylated CpG islands in the pro- focused on sensitivity instead of specificity (see also the moter regions of tumor suppressor genes represses its results in the “Reducing the number of follow-up cystos- transcription in human tumors [9, 10]. Therefore, aber- copies by using the three-gene methylation/cytology rant DNA methylation is a potential biomarker for diag- combined classifier” section). PFBC with a prior endove- nosis, prognosis, and monitoring of disease after therapy sical chemotherapeutic or immunotherapeutic treatment [11]. Recently, it was shown that the combination of could be included. Without being mandatory, we recom- SOX1, IRAK3,and L1-MET provides better resolution mended PFBC to have cytology at cystoscopy. The ex- than cytology and cystoscopy in the detection of early clusion criterion was a histological confirmation of any recurrence [12]. The objective of the present study is to other urological malignancy. investigate whether a set of methylation markers can A total number of 725 voided urine samples were pro- lead to the development of a voided urine test that pre- spectively collected by the four participating institutions. dicts tumor presence and may be used to stratify BC pa- From the total number of urines collected, 99 (13%) tients according to their risk of recurrence, thus were excluded from the study because of technical prob- allowing the reduction of the number of cystoscopies in lems during the sample collection, storage, or analysis. the follow-up of BC. Finally, 626 urines were used: 111 from BC patients and 57 from controls for the discovery phase and 458 from Methods PFBC (of whom 173 had a recurrence) for the validation Patients and clinical samples phase (Tables 1 and 2). The grade and stage of the tu- After Institutional Review Board approval and obtaining mors were determined according to WHO 2004 criteria patients’ informed consents, we prospectively collected and TNM 2002 classification, respectively [13, 14]. freshly voided urine samples from BC patients, controls, Tumors were classified according to their risk of re- and patients in follow-up for bladder cancer (PFBC) at currence and progression into three categories: high-risk van der Heijden et al. Clinical Epigenetics (2018) 10:71 Page 3 of 10 Table 1 Clinicopathological and demographic characteristics of Table 1 Clinicopathological and demographic characteristics of the study population classified by the study phase the study population classified by the study phase (Continued) Discovery phase Validation phase Discovery phase Validation phase Training set Testing set Training set Testing set N bladder cancer (%) N R-PFBC (%) T1 LG – 22 (8) Gender T1 HG – 82 (29) Male 86 (77) 135 (78) T2 HG – 3 (1) Female 25 (23) 38 (22) Tx LG – 2 (1) Age Tx HG – 2 (1) Mean 72 68 Subtotals 285 Range 39–98 26–99 Total 168 458 LG low-grade, HG high-grade, TURBT transurethral resection bladder tumor, Stage and grade BPH benign prostate hyperplasia, CIS/Tis carcinoma in situ, BC bladder cancer, Tis 7 (6) 13 (8) R-PFBC recurrent patients in follow-up for bladder cancer, NR-PFBC non- recurrent patients in follow-up for bladder cancer Ta LG 26 (23) 61 (35) Ta HG 11 (10) 20 12) (HR) NMIBC (any of the following: T1, HG/G3 tumors, T1 LG 20 (18) 35 (20) or CIS), non-high-risk (nHR) NMIBC (all the other cases of NMIBC), and muscle-invasive bladder cancer (MIBC) T1 HG 22 (20) 44 (25) (T2–4). None of the included patients had an upper > T2 LG 1 (1) – urinary tract tumor. > T2 HG 24 (22) – Subtotals 111 173 Urine sample processing N control (%) Urine samples were collected before cystoscopy, the day Gender before the transurethral resection of the bladder tumor (TURBT), or the day before cystectomy. From all pa- Male 29 (51) – tients and controls, only one single sample was included. Female 28 (49) – Age For urine cytology Mean 60 – Urine cytology was performed according to Papanico- Range 22–82 – laou staining and evaluated by expert pathologists in Urinary condition each center blinded to the patient’s clinical history. The results were considered as positive or negative. No cen- BPH 11 (19) – tral cytology review was performed. Urolithiasis 13 (23) – Incontinence 2 (4) – For methylation studies Benign bladder disease 1 (2) – Voided urine samples (50 to 100 ml) were collected in Urinary tract infections 12 (21) – sterile containers containing 4 ml of 0.5 M EDTA, pH Non-urological diseases 18 (32) – 8.0. Urines were immediately stored at 4 °C and proc- essed within the next 24 h. The samples were centri- Subtotals 57 – fuged at 1000×g for 10 min, at 4 °C. The cell pellets N NR-PFBC (%) were frozen at − 80 °C. Gender Male – 217 (76) DNA isolation, bisulfite treatment, and PCR Female – 68 (24) DNAs from the urinary cell pellets were extracted using Age QIAamp DNA Mini Kit (Qiagen) according to the manufacturer’s instructions and quantified with a Nano- Mean – 69 Drop1000 (NanoDrop Technologies, Wilmington, DE, Range – 26–92 USA). DNA extraction was performed in each center ex- Stage and grade previous TURBT cept for Bucharest, whose cell pellets were sent in dry Tis – 21 (7) ice to the Radboud University Medical Center, Nijmegen Ta LG – 100 (35) (The Netherlands) for DNA extraction. Ta HG – 53 (19) One microgram of genomic DNA was used for the bi- sulfite modification using EpiTect Bisulfite kit (Qiagen, van der Heijden et al. Clinical Epigenetics (2018) 10:71 Page 4 of 10 Table 2 Clinicopathological and demographic characteristics of the study population classified by the participating center Hospital Clinic Barcelona Radboud University Saint John Emergency MD Anderson Cancer Total Medical Center, Nijmegen Clinical Hospital Bucharest Center Houston Discovery phase Bladder cancer urine samples Stage Tis 5 – 11 7 Ta 14 5 5 13 37 T1 19 2 21 – 42 >T2 9 – 16 – 25 Grade LG 20 2 17 8 47 HG 27 5 26 6 64 Subtotal 47 7 43 14 111 Control urine samples BPH 8 3 –– 11 Urolithiasis 13 –– – 13 Incontinence 1 1 –– 2 Benign bladder 45 1 – 10 disease Urinary tract 18 4 – 13 infection Non-urological 6 –– 28 diseases Subtotal 33 17 5 2 57 Validation phase R-PFBC URINE SAMPLES Stage Tis 2 1 8 2 13 Ta 13 15 47 6 81 T1 4 2 55 18 79 Grade LG 14 8 74 – 96 HG 5 10 36 26 77 Subtotal 19 18 110 26 173 NR-PFBC urine samples Stage previous TURBT Tis 5 6 6 4 21 Ta 32 56 36 26 150 Ta + CIS 1 –– 23 T1 36 16 25 19 96 T1 + CIS 5 –– 38 T2 3 –– – 3 Tx 4 –– – 4 Grade previous TURBT LG 34 35 38 19 126 HG 52 43 29 35 159 Subtotal 86 78 67 54 285 TOTAL 185 120 225 96 626 LG low-grade, HG high-grade, TURBT transurethral resection bladder tumor, BPH benign prostate hyperplasia, CIS/Tis carcinoma in situ, BC bladder cancer, R-PFBC recurrent patients in follow-up for bladder cancer, NR-PFBC non-recurrent patients in follow-up for bladder cancer. van der Heijden et al. Clinical Epigenetics (2018) 10:71 Page 5 of 10 Inc.) following the manufacturer’s instructions. The inclusion criterion was p value of < 0.1. Risk probability modified DNA was eluted with 20 μl Tris-HCL (1 mM, of presenting BC was calculated in the training set. We pH 8.0) and stored at − 80 °C before further processing. established cutoff point value (≥ 0.464) allowing 15% Bisulfite modifications were performed in Hospital false negatives in the tumor group (SN = 85%). In the Clinic of Barcelona, Spain (training set), and in Radboud subset of samples in which cytology results were avail- University Medical Center, Nijmegen, The Netherlands able, the cutoff point value that yielded 85% SN in the (testing set). training set was 0.688, and in the combined model A total of seven DNA methylation markers, i.e., (methylation test + cytology results), it was 0.617. If the CDH13, CFTR, NID2, SALL3, TMEFF2, TWIST1, and predicted probability value derived from each classifier VIM2, were selected from four recently published BC in each of the samples was higher than the cutoff point studies [15–18]. The sequences of the primers used to value, the samples were classified for each of the classi- amplify the regions of interest of these genes and the fiers as tumor sample. The performance of the models PCR conditions are shown in Additional file 2: Table S1. was evaluated in a testing set by means of AUROC using PCR primers were designed using the PyroMark Assay pROC R-package [19]. Student’s t test was used to evalu- Design software v2.0 (Qiagen). PCR was performed in a ate statistical differences in DNA methylation. Statistical volume of 25 μl with 2 μl of converted genomic DNA, significance was established at p value of 0.05. 0.6 U Ampli Taq Gold 360 DNA polymerase (Thermo- R-software and SPSS v23.0 were used for calculations. fisher), 0.8 μl of a mix of Primer-F and biotinylated Primer-R at 10 μM, 2 μl MgCl 25 mM, and 0.5 μl Results dNTPs 10 mM. Amplification was performed according Training set to the following thermocycling conditions: denaturation DNA methylation of all seven selected genes was signifi- at 95 °C for 10 min, followed by 45 cycles of 95 °C for cantly increased in urine sediments from BC patients 30 s, the optimal Tm for 30 s, and 72 °C for 1 min; and compared to controls (Fig. 1). To determine the combin- a final extension at 72 °C for 7 min. The formation of ation of markers capable of detecting BC in urine PCR products with accurate size was confirmed by sediments with the highest accuracy, we built a model of resolving PCR samples (1 μl) by 2% agarose gel elec- multiple markers by logistic regression. The best pos- trophoresis, with visualization by ethidium bromide sible biomarker combination based on AUC was pro- staining. vided by the combination of CFTR, SALL3, and TWIST1. This three-gene methylation classifier achieved Pyrosequencing for quantitative methylation an AUC = 0.874 (Fig. 2a); at a fixed overall SN of 85%, Biotin-labeled single-stranded amplicons were isolated the classifier provides a SP of 68%. Moreover, the SN of from 20 μl of the PCR product according to the protocol the three-gene methylation classifier increases through using the Pyromark Q96 Work Station and pyrose- the BC risk groups and grading (Table 3). quenced with 0.3 μM sequence primer using PSQ96MD System (Biotage AB). Additional file 2: Table S1 shows Testing set the sequences of the primer sets used for bisulfite se- To examine whether the three-gene methylation BC quencing. The percent methylation for each of the CpGs diagnostic classifier was able to identify recurrences in a was calculated using PyroQ CpG Software (Qiagen). The clinical setting, the classifier was validated in an inde- differences in the percentage of methylation were calcu- pendent multicenter international series of 458 urine lated between BC vs. control and recurrent vs. sediments from patients in follow-up for bladder cancer non-recurrent PFBC (R-PFBC and NR-PFBC, respect- (PFBC), of whom 173 had a recurrence. Recurrent PFBC ively) samples. A high correlation in the methylation (R-PFBC; N = 173) displays higher percentages of DNA percentages of the CpG dinucleotides in the same is- methylation compared with non-recurrent PFBC land was observed (Additional files 3 and 4:Figures (NR-PFBC; N = 285) (Fig. 3). SN of the three-gene S2 and S3). For this reason, hypermethylation was an- methylation BC classifier increased in the validation alyzed in all genes at the first CpG dinucleotide series (SN = 90%), while SP drops (SP = 31%), as evi- present (Additional file 5:Table S2). denced by the ROC curves and AUC value (AUC = 0.741) (Table 3 and Fig. 2a). Figure 4a depicts the risk Data analysis probabilities derived from the three-gene methylation Univariable and multivariable logistic regression analyses classifier in R-PFBC and NR-PFBC. were used to examine the associations between BC and DNA methylation status of urinary sediments. A forward Comparison of test performance with urine cytology stepwise logistic regression was performed to determine A total of 399 urine cytologies (91 from the training and the best classifier between BC and control samples. The 308 from the testing set) were performed. In both van der Heijden et al. Clinical Epigenetics (2018) 10:71 Page 6 of 10 Fig. 1 Percentage of DNA methylation in bladder cancer and control urine sediments for the seven selected genes analyzed in the discovery phase. The number of samples in each group is given in brackets. Abbreviations: BC, bladder cancer; C, control training and testing set, SN of the three-gene methyla- three-gene methylation classifier than that of the urine tion classifier (86 and 93%, respectively) was higher than cytology in training (40 and 29%, respectively) and that of the urine cytology (54 and 46%, respectively) testing set (82 and 69%, respectively). Contrary, posi- in this subset of samples (Additional files 6 and 7: tive predictive value (PPV) is higher for the urine cy- Table S3 and Figure S4). In the testing set, this means tology than for the methylation classifier in training that 55% of the recurrences (68 out of 124) were (98 and 89%, respectively) and testing set (85 and detected by the three-gene classifier but were missed by 50%, respectively). Cytology had a SP of 93 and 94% urine cytology. On the other hand, 12 recurrences were while the three-gene methylation classifier achieved a missed by the three-gene classifier of which half was SP, in this subset of samples, of 47 and 27%, in the training detected by cytology (Additional file 8: Figure S5). and testing set, respectively. In the testing set, 120 Negative predictive value (NPV) is also higher for the NR-PFBC samples were positive by the three-gene Fig. 2 ROC curves in the training and testing series for (a) the three-gene BC methylation classifier and (b) the combined three-gene methylation/cytology classifier van der Heijden et al. Clinical Epigenetics (2018) 10:71 Page 7 of 10 Table 3 Diagnostic performance of the three-gene methylation methylation classifier. Nine of them also had positive classifier in the training and testing set of samples (at fixed urine cytology (Additional file 8:Figure S5). After 1year, sensitivity of 85% in the training set) two NR-PFBC with a positive test (who had a negative cy- Training set Testing set tology) had a tumor recurrence. Overall N samples 111 BC/57 C 173 R-PFBC/285 NR-PFBC Combination of the three-gene methylation classifier with AUC 0.874 0.741 urine cytology Combining the three-gene methylation classifier and SN (%) 84.68 89.6 urine cytology results showed an improved diagnostic SP (%) 68.42 30.53 performance in the training (AUC 0.858) and as well as PPV (%) 83.93 43.91 in the testing set (AUC 0.86) (Fig. 2b). A SN of 96% and NPV (%) 69.64 82.86 a NPV of 92% are achieved in the testing set. Of note, in Non-high-risk NMIBC HG tumors, a 100% SN and NPV are achieved N samples 26 BC/57 C 61 R-PFBC/285 NR-PFBC (Additional file 6: Table S3). The risk probabilities derived from the combined classifier in R-PFBC and SN (%) 73.08 88.52 NR-PFBC are shown in Fig. 4b. SP (%) 68.42 30.53 PPV (%) 51.35 21.43 Reducing the number of follow-up cystoscopies by using NPV (%) 84.78 92.55 the three-gene methylation/cytology combined classifier High-risk NMIBC In our study, we oversampled patients with BC. We N samples 60 BC/57 C 112 R-PFBC/285 NR-PFBC therefore cannot directly calculate predictive values SN (%) 86.67 90.18 from the test results. In the hospitals participating in SP (%) 68.42 30.53 thestudy,recurrenceisdetectedinapproximately 10% of all follow-up cystoscopies performed (90% of PPV (%) 74.29 33.78 patients previously diagnosed with NMIBC are NPV (%) 82.98 88.78 without recurrence at the time of follow-up cystoscopy). MIBC In order to calculate the predictive values that reflect N samples 25 BC/57C – values in real clinical practice, we assumed the distribution SN (%) 92 – of recurrent vs. non-recurrent to be 10 vs. 90%. For this, SP (%) 68.42 – we multiplied the NR-PFBC samples by 7. Using the SN and SP that we found in the study, the PPV and NPV in PPV (%) 56.1 – the validation phase become 15 and 99%, respectively. If NPV (%) 95.12 – patients with a negative classifier will not undergo a cyst- Low-grade oscopy, this means that more than a third (~ 36%) of all N samples 47 BC/57 C 96 R-PFBC/285 NR-PFBC cystoscopies can be prevented at the cost of 4% of recur- SN (%) 76.6 90.62 rences remaining undiagnosed which all were LG tumors. SP (%) 68.42 30.53 PPV (%) 66.67 30.53 Discussion NPV (%) 78 90.62 In the present study, a set of DNA methylation markers to predict the presence of bladder cancer (BC) in urine High grade samples has been selected. The best possible marker N samples 64 BC/57 C 77 R-PFBC/285 NR-PFBC combination to discriminate BC from controls was the SN (%) 90.62 88.31 combination CFTR, SALL3, and TWIST1. We confirmed SP (%) 68.42 30.53 that these genes (and specifically CpG dinucleotides PPV (%) 76.32 25.56 analyzed here) are hypermethylated in the bladder can- NPV (%) 86.67 90.62 cer tissue using methylation data from the TCGA Research Network [20] (Additional file 9: Figure S6). LG low-grade, HG high-grade, AUC area under the curve, MIBC muscle-invasive bladder cancer, NMIBC non-muscle invasive bladder cancer, NPV negative This supports their use as diagnostic markers in urine predictive value, PPV positive predictive value, SN sensitivity, SP specificity, BC samples. In the training set, the three-gene methylation bladder cancer, C control, R-PFBC recurrent patients in follow-up for bladder cancer, NR-PFBC non-recurrent patients in follow-up for bladder cancer. classifier achieved an AUC 0.874 while in the testing set, an AUC 0.741 was achieved to discriminate recurrent from non-recurrent patients in follow-up for bladder cancer (PFBC). These results improved significantly in van der Heijden et al. Clinical Epigenetics (2018) 10:71 Page 8 of 10 Fig. 3 Percentage of DNA methylation in recurrent and non-recurrent PFBC urine sediments for the three genes of the classifier in the validation phase. The number of samples in each group is given in brackets. Abbreviations: NR-PFBC, non-recurrent patients in follow-up for bladder cancer; R-PFBC, recurrent patients in follow-up for bladder cancer the testing set when cytology results were included in [21]. However, the results were significantly better in the analysis (AUC 0.86). thesubgroupof activesmokers.Unfortunately,we did TWIST1 hypermethylation in BC was first described not collect information about tobacco smoking, and by Renard and co-workers [16]. In a case-control study therefore, we cannot perform a subset analysis for (n = 145 cases/321 controls), they detected TWIST1 and smoking behavior. NID2 hypermethylation in urine sediments of BC pa- Yu and co-workers previously found in a case-control tients using methylation-specific PCR. This two-gene study that CFTR and SALL3, out of 59 genes that were panel achieved a SN of 90%, SP of 93%, PPV of 86%, screened, were the most frequently methylated genes to and NPV of 95%. Nevertheless, the group of Fantony predict the presence of BC in urine [15]. However, in published conflicting results. They found only a SN of this study, no PFBC were included. In daily clinical 67% and SP of 69% for this two-gene urine panel practice, this group is especially of interest. It is not very ab Fig. 4 Box plots showing the individual risk probabilities derived from (a) the three-gene methylation classifier and (b) the combined three-gene methylation/cytology classifier for recurrent and non-recurrent PFBC in the cross-sectional study. Dots above the cutoff value (dashed line) denote positive samples, whereas those below signify negatives scores. Cutoff values for the three-gene methylation classifier and the combined three-gene methylation/cytology classifier are ≥ 0.464 and ≥ 0.617, respectively. Abbreviations: NR-PFBC, non-recurrent patients in follow-up for bladder cancer; R-PFBC, recurrent patients in follow-up for bladder cancer van der Heijden et al. Clinical Epigenetics (2018) 10:71 Page 9 of 10 likely that biomarkers will replace ureterocystoscopy in validation series, and we had to make an estimation to patients with macroscopic hematuria referred to the ur- calculate the number of cystoscopies that could be ologist. But in PFBC, urinary biomarkers with a high SN skipped. Secondly, 13% of the samples had to be ex- and high NPV could make a difference, i.e., lower the cluded due to technical failures. Thirdly, intravesical number of follow-up cystoscopies. Our combined three-gene treatments in NMIBC patients may have influenced methylation/cytology classifier achieves a high SN and NPV, methylation patterns. Finally, we have evaluated only a for both high-risk and non-high-risk NMIBC. Consequently, limited number of hypermethylated genes with diagnos- more than a third of all cystoscopies could be prevented. tic value previously described in the literature. The SP of the combined classifier in the testing set, using PFBC samples, is expected to be lower than the SP Conclusions of the training set, using BC and control samples. In conclusion, this combined three-gene methylation/cy- Possible reasons for methylation observed in PFBC sam- tology classifier can reduce the number of follow-up ples are small tumors not yet detected by cystoscopy, re- cystoscopies in PFBC. This approach may improve the sidual tumor cells at the resection site, or epigenetically patients’ quality of life. For a definitive conclusion, repli- changed urothelial cells at the resection site or else in cation of the classifier in another series of patients and the bladder also known as epigenetic field defect [22]. In cost-effectiveness studies are needed. patients with persistent hypermethylation, which does not recur within 18 months, the presence of an Additional files epigenetic field defect is most likely. Wolff and co-workers suggested that the aberrant methylation is Additional file 1: Figure S1. Flowchart of the entire study. A total of caused by a generalized epigenetic alteration in the seven hypermethylated genes, differentially expressed between BC patients whole bladder urothelium, and this widespread methyl- and controls (n = 168), were determined in the discovery phase. With these results, a three-gene methylation classifier was developed. This three-gene ated urothelium may be the cause of the high recurrence classifier was tested in a cross-sectional study (validation phase; n =458). rate in NMIBC [22]. Samples with available cytology results in each phase are indicated. Abbrevia- The discovery of highly sensitive methylation markers tions: BC, bladder cancer; C, control; R-PFBC, recurrent patients in follow-up for bladder cancer; NR-PFBC, non-recurrent patients in follow-up for bladder allows us to lower the number of follow-up cystoscopies cancer. (PPTX 75 kb) in more than a third of all follow-up patients. In the Additional file 2: Table S1. Primer sequences used in PCR and clinical situation, patients supply a urine sample before pyrosequencing. (DOCX 13 kb) cystoscopy to perform the test. If the combined test is Additional file 3: Figure S2. Pearson correlation coefficient heat map positive, patients will undergo a cystoscopy. In our of the percentage of methylation for every CpG dinucleotide in the seven genes. Every CpG sites are correlated (via the Pearson correlation) validation series, 60% of NR-PFBC had a positive com- with all others. Correlations are scaled by the color of the corresponding bined test and should undergo a cystoscopy. This is not cell. Parameters are represented in the same order on the x- and y-axes. a major problem since in normal daily practice, (PPTX 1232 kb) follow-up patients would have undergone a cystoscopy Additional file 4: Figure S3. Representative pyrograms showing gene methylation patterns in DNA urine samples from a bladder cancer anyway. If the combined test is negative, cystoscopy patient. Percentage of methylation is indicated above peaks (gray could be skipped. columns) corresponding to the CpG sites in this region. (PPTX 183 kb) However, a methylation test has also financial and lo- Additional file 5: Table S2. Percentage of methylation for each CpG gistic implications, which means that a cost-effectiveness dinucleotide in the seven selected genes in control and bladder cancer urine samples. Underlined in grey the CpG site used for methylation analysis is necessary. Using our three-gene methylation/ analysis. Abbreviations: SDV; Standard Deviation. (DOCX 21 kb) cytology classifier, 4% of PFBC are wrongly diagnosed as Additional file 6: Table S3. Diagnostic performance of the three-gene not having a recurrence; all of them had LG NMIBC. Of methylation classifier, cytology, and the combined methylation/cytology note, cystoscopy, our gold standard, misses up to 15% of classifier in the training and testing subset of samples with cytology available. Abbreviations: LG, Low Grade; HG, High Grade; AUC, area under the papillary and up to 30% of the flat lesions [4]. curve; MIBC, muscle invasive bladder cancer; NMIBC, Non-Muscle Invasive The strengths of this study lie in the fact that we have Bladder Cancer; NPV, Negative Predictive Value; PPV, Positive Predictive chosen for a two-stage approach using PFBC in the val- Value; SN, Sensitivity; SP, Specificity; BC, Bladder Cancer; C, Control; R- PFBC, Recurrent Patients in Follow up for Bladder Cancer; NR-PFBC, Non idation phase. Furthermore, the use of voided urine Recurrent Patients in Follow up for Bladder Cancer. (DOCX 20 kb) samples to analyze the DNA methylation status allows Additional file 7: Figure S4. Sensitivity, negative and positive predictive the development of a non-invasive BC diagnostic tool values of urine cytology, the three-gene methylation classifier, and the with an easy translation into clinical practice. However, combined three-gene methylation/cytology classifier in the testing set (N = 308). Overall specificity was 94% for urine cytology, 27% for the three-gene some limitations should be mentioned. To avoid ineffi- methylation classifier, and 40% for the combined three-gene methylation/ ciency, patients with a recurrence were oversampled by cytology classifier. Abbreviations: LG, low-grade; HG, high-grade; NMIBC also recruiting patients who were scheduled for a nHR, non-muscle-invasive bladder cancer non-high risk; NMIBC HR, non-- muscle-invasive bladder cancer high risk; NPV, negative predictive value; TURBT of a proven bladder tumor. Consequently, the PPV, positive predictive value. (PPTX 189 kb) number of NR-PFBC was misrepresented in the van der Heijden et al. Clinical Epigenetics (2018) 10:71 Page 10 of 10 Received: 1 December 2017 Accepted: 3 May 2018 Additional file 8: Figure S5. Flow diagram of participants in the cross-sectional study according a) to the three-gene methylation classifier and cytology results and b) to the combined three-gene methylation/cytology References classifier. Abbreviations: R-PFBC, recurrent patients in follow-up for bladder 1. Oosterlinck W, Lobel B, Jakse G, et al. Guidelines on bladder cancer. Eur cancer; NR-PFBC, non-recurrent patients in follow-up for bladder cancer; Cytol, Urol. 2002;41:105–12. cytology; NA, non-available; Test, combined three-gene methylation/cytology 2. Fernandez-Gomez J, Madero R, Solsona E, et al. Predicting nonmuscle invasive classifier. (PPTX 85 kb) bladder cancer recurrence and progression in patients treated with bacillus Additional file 9: Figure S6. DNA methylation profiles for bladder Calmette-Guerin: the CUETO scoring model. J Urol. 2009;182:2195–203. cancer and control tissue samples for the three-gene classifier. Data obtained 3. Babjuk M, Bohle A, Burger M, et al. EAU Guidelines on Non-Muscle-invasive from Wanderer Web page: http://maplab.imppc.org/wanderer/.The Urothelial Carcinoma of the Bladder: update 2016. Eur Urol. 2016;71:447–61. red arrow indicates the CpG dinucleotide analyzed in each of the 4. Grossman HB, Gomella L, Fradet Y, et al. A phase III, multicenter comparison three genes. (PPTX 203 kb) of hexaminolevulinate fluorescence cystoscopy and white light cystoscopy for the detection of superficial papillary lesions in patients with bladder cancer. J Urol. 2007;178:62–7. Abbreviations 5. Fradet Y, Grossman HB, Gomella L, et al. A comparison of AUC: Area under the curve; BC: Bladder cancer; BPH: Benign prostatic hexaminolevulinate fluorescence cystoscopy and white light cystoscopy for hyperplasia; C: Control; CIS/Tis: Carcinoma in situ; HG: High-grade; HR: High-risk; the detection of carcinoma in situ in patients with bladder cancer: a phase LG: Low-grade; MIBC: Muscle-invasive bladder cancer; NA: Not available; III, multicenter study. J Urol. 2007;178:68–73. nHR: Non-high risk; NMIBC HR: Non-muscle-invasive bladder cancer high risk; 6. Brown FM. Urine cytology. It is still the gold standard for screening? Urol NMIBC nHR: Non-muscle-invasive bladder cancer non-high risk; NMIBC: Non- Clin North Am. 2000;27:25–37. muscle-invasive bladder cancer; NPV: Negative predictive value; NR-PFBC: Non- 7. Sherman AB, Koss LG, Adams SE. Interobserver and intraobserver differences recurrent patients in follow-up for bladder cancer; PPV: Positive predictive value; in the diagnosis of urothelial cells. Comparison with classification by R-PFBC: Recurrent patients in follow-up for bladder cancer; SN: Sensitivity; computer. Anal Quant Cytol. 1984;6:112–20. SP: Specificity; TURBT: Transurethral resection of the bladder tumor 8. Parker J, Spiess PE. Current and emerging bladder cancer urinary biomarkers. ScientificWorldJournal. 2011;11:1103–12. Acknowledgements 9. Esteller M. Epigenetics in cancer. N Engl J Med. 2008;358:1148–59. We thank the technical support from the staff of the Servei Veterinari de 10. Saxonov S, Berg P, Brutlag DL. A genome-wide analysis of CpG Genètica Molecular, Facultat de Veterinària, Universitat Autònoma de Barcelona. dinucleotides in the human genome distinguishes two distinct classes of Part of the work was developed at the building Centre de Recerca Biomèdica promoters. Proc Natl Acad Sci U S A. 2006;103:1412–7. Cellex, Barcelona. Furthermore, funding from CERCA Programme/Generalitat de 11. Shames DS, Minna JD, Gazdar AF. DNA methylation in health, disease, and Catalunya is gratefully acknowledged. cancer. Curr Mol Med. 2007;7:85–102. 12. Su SF, Castro Abreu AL, Chihara Y, et al. A panel of three markers hyper- and hypomethylated in urine sediments accurately predicts bladder cancer Funding recurrence. Clin Cancer Res. 2014;20:1978–89. This work was supported by grants from the Dutch Cancer Society. 13. Sobin LH, Wittekind CH. TNM classification of malignant tumours. International union against cancer. 6th ed. New York: Wiley; 2002. Availability of data and materials 14. Lopez-Beltran A, Sauter G, Gasser T, et al. Tumours of the urinary system. In: Please contact the corresponding author for data requests. Eble JN, Sauter G, Epstein JI, Sesterhenn IA, editors. Pathology and genetics of tumours of the urinary system and male genital organs. World Health Authors’ contributions Organization classification of tumours. Lyon: IARC Press; 2004. p. 89–157. AGvdH, LM, LALMK, MJR, JAW, and AA contributed to the conception and 15. Yu J, Zhu T, Wang Z, et al. A novel set of DNA methylation markers in urine design. AGvdH, LM, MI-T, CCMvR-vdW, MB, LALMK, MJR, JAW, and AA sediments for sensitive/specific detection of bladder cancer. Clin Cancer contributed to the methodology, collection, and assembly of data. All Res. 2007;13:7296–304. authors contributed to the data analysis and interpretation and manuscript 16. Renard I, Joniau S, van Cleynenbreugel B, et al. Identification and validation writing and reviewing. All authors approved the final manuscript. of the methylated TWIST1 and NID2 genes through real-time methylation- specific polymerase chain reaction assays for the noninvasive detection of primary bladder cancer in urine samples. Eur Urol. 2010;58:96–104. Ethics approval and consent to participate 17. Brait M, Begum S, Carvalho AL, et al. Aberrant promoter methylation of The study was approved by the Institutional Review Boards of the Hospital multiple genes during pathogenesis of bladder cancer. Cancer Epidemiol Clínic of Barcelona (Spain); Radboud University Medical Center in Nijmegen Biomark Prev. 2008;17:2786–94. (The Netherlands), St. John Emergency Hospital, Bucharest (Romania); and 18. Costa VL, Henrique R, Danielsen SA, et al. Three epigenetic biomarkers, MD Anderson Cancer Center, Houston, Texas, (USA). Patients’ informed GDF15, TMEFF2, and VIM, accurately predict bladder cancer from DNA- consents were obtained before the sample collection. based analyses of urine samples. Clin Cancer Res. 2010;16:5842–51. 19. Robin X, Turck N, Hainard A, et al. pROC: an open-source package for R and Competing interests S+ to analyze and compare ROC curves. BMC Bioinformatics. 2011;12:77. The authors declare that they have no competing interests. 20. Diez-Villanueva A, Mallona I, Peinado MA. Wanderer, an interactive viewer to explore DNA methylation and gene expression data in human cancer. Epigenetics Chromatin. 2015;8:22. Publisher’sNote 21. Fantony JJ, Abern MR, Gopalakrishna A, et al. Multi-institutional external Springer Nature remains neutral with regard to jurisdictional claims in validation of urinary TWIST1 and NID2 methylation as a diagnostic test for published maps and institutional affiliations. bladder cancer. Urol Oncol. 2015;33:387–6. 22. Wolff EM, Chihara Y, Pan F, et al. Unique DNA methylation patterns Author details distinguish noninvasive and invasive urothelial cancers and establish an Department of Urology Radboud University Medical Center, Nijmegen, The epigenetic field defect in premalignant tissue. Cancer Res. 2010;70:8169–78. Netherlands. Laboratory and Department of Urology, Hospital Clinic of Barcelona, IDIBAPS, University of Barcelona, Barcelona, Spain. CIBERehd, Plataforma de Bioinformática, Centro de Investigación Biomédica en red de Enfermedades Hepáticas y Digestivas, Barcelona, Spain. Saint John Emergency Clinical Hospital, Bucharest, Romania. MD Anderson Cancer Center, Houston, Texas, USA. Hospital Clínic de Barcelona, Centre de Recerca Biomèdica CELLEX, office B22, C/Casanova, 143, 08036 Barcelona, Spain.

Journal

Clinical EpigeneticsSpringer Journals

Published: May 30, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off