Clinical Validation of the Proenkephalin (PENK) Methylation Urine Test for Monitoring Recurrence of Non–muscle-invasive Bladder Cancer

Take Home Message The mePENK test showed good performance in detecting recurrence of non–muscle-invasive bladder cancer and outperformed cytology and the NMP22 test in diagnostic accuracy. The mePENK test offers promising clinical utility for NMIBC management and monitoring, particularly for patients with high-risk disease.


Introduction
Bladder cancer (BC) is the ninth most common cancer worldwide, with approximately 549 000 new cases reported annually [1].Some 75% of these cases are nonmuscle-invasive BC (NMIBC), which has a high recurrence rate after transurethral resection of bladder tumor (TURBT) and potential for progression to muscle-invasive disease [2].Cystoscopy in conjunction with urine cytology is the standard for NMIBC surveillance.Although urine cytology has high specificity, its sensitivity is limited, especially for low-grade tumors [3], which can lead to overdiagnosis and unnecessary invasive procedures.Accurate and precise surveillance methods are essential for prompt detection and treatment of recurrent NMIBC [4].
Urine-based biomarker tests for BC surveillance, such as the BTA, NMP22, and UroVysion tests, have been approved by the US Food and Drug Administration.Tests such as CxBladder test and Xpert Bladder Cancer Monitor, which detect multiple mRNA biomarkers in urine samples, have been also developed for BC recurrence surveillance [5][6][7][8].
Aberrant DNA methylation, a major epigenetic modification, often results in inactivation of tumor suppressor and other cancer-related genes [9,10].Specific aberrant methylation sites on relevant genes are potential biomarkers for early detection of BC [11,12].The EpiCheck test uses a panel of 15 DNA methylation sites as biomarkers in urine and is a promising BC detection approach [13].
We previously identified proenkephalin (PENK) methylation as a specific diagnostic biomarker for BC and demonstrated its clinical validity for BC detection by measuring PENK methylation in urine via quantitative methylationspecific polymerase chain reaction (PCR) qMSP) following linear target enrichment (LTE) (mePENK-LTE/qMSP test) [14,15].In the present study we further optimized this urine-based mePENK test using a closed single-tube system.The method involves two consecutive rounds of PCR: (1) a linear amplification step to enrich methylated PENK target DNA from a bisulfite-treated samples; and (2) qMSP for amplification of the mePENK target region and the control gene.Our aim was to evaluate the clinical effectiveness of the mePENK test for BC recurrence surveillance after TURBT and to compare its performance with the NMP22 test and cytology.

Study design and patients
We performed a retrospective case-control analysis to determine the optimal cutoff value for the mePENK test for detection of BC.The study population comprised 54 patients with primary BC and 29 healthy individuals.
In an additional prospective pilot study, 188 patients with a history of BC and scheduled for cystoscopy surveillance after TURBT were enrolled between January and December 2022 (Fig. 1).This study received institutional review board approval (#4-2021-0406) and all patients provided written consent.

Sample collection, DNA isolation, and bisulfite treatment
A sample of voided urine (20 ml) was collected before cystoscopy using a Urine Collection Kit I (Genomictree Inc., Seoul, Republic of Korea).Genomic DNA was extracted from urine sediment and was chemically modified with sodium bisulfite for methylation analysis.

PENK methylation assessment using the mePENK test
The mePENK test was developed to assess PENK methylation in urine-derived DNA.This assay, conducted in a closed tube system, involves a two-round PCR process for accurate and continuous amplification of PENK methylated DNA and control DNA targets (Supplementary Fig. 1).In the first PCR round, a high annealing temperature is applied to facilitate unidirectional DNA synthesis.A specific primer, featuring a unique tag sequence (UTS) at its 5 0 end, was designed to anneal to the PENK methylation site under the elevated temperature (70°C) conditions, preventing other primers with regular melting temperatures from initiating DNA synthesis.The template DNA synthesis occurs in one direction only, replicating in each PCR cycle.
In the second PCR round, the reaction is conducted at a normal temperature (60°C).Two distinct sets of primers and probes are used for amplification of the PENK methylation target and of the control target.The forward primer for the PENK methylation target was designed to bind specifically to the PENK methylation site, and the reverse primer contained a UTS.For the control target, a primer set specific to a DNA region of COL2A1 was used, with annealing at the normal temperature.Probes were designed to bind to the internal sites of each target PCR product; generation of discernible signals indicated PCR product formation.
Genomic DNA isolated from the sediment from 10 ml of urine was used as the input for bisulfite treatment.Each 25ll reaction mixture contained 18 ll of bisulfite-converted DNA, 200 nmol/l PENK methylation-specific forward primer, 40 nmol/l PENK methylation-specific reverse primer with a UTS at the 5 0 end, 160 nmol/l PENK probe, 200 nmol/l COL2A1 forward primer, 200 nmol/l COL2A1 probe, 200 nmol/l COL2A1 reverse primer, 200 nmol/l reverse primer for the UTS, and 5 ll of 5Â Fast qPCR PreMIX TaqMan Probe (Enzynomics Inc., Daejon, Republic of Korea).COL2A1, which lacks the CpG dinucleotide, was used as a control.The primer and probe sequences are provided in Supplementary Table 1.
The entire mePENK test procedure was performed in a single closed tube on an AB7500 FAST Real-Time PCR system (Thermo Fisher Scientific, Waltham, MA, USA).The specific thermal cycling conditions were 95°C for 5 min, followed by 15 cycles at 95°C for 15 s and 70°C for 45 s, then 35 cycles at 95°C for 15 s and 60°C for 45 s.The relative level of PENK methylation in each sample was calculated as 35ÀDC T , where DC T is the difference in cycle threshold between the amplified PENK target and COL2A1 [12].Higher 35ÀDC T values indicated higher levels of PENK methylation.If the PENK C T was undetectable, the value was set to 20, which was the value closest to the lowest 35ÀDC T result for all tests.

NMP22 test
The Alere NMP22 test (Abbott, Abbott Park, IL, USA), an enzyme immunoassay method, was used for NMP22 detection in urine samples.The detection threshold was 10 IU/ml.

Reference standard
Cystoscopy with a pathology result (TURBT under general anesthesia) was the reference standard.Results were classified as recurrence or recurrence-free.Recurrence events were subcategorized as progression to muscle-invasive disease or progression-free.

Survival analysis
Recurrence-free survival (RFS) was defined as the time from diagnosis to the occurrence of recurrence.Patients who were recurrence-free at the time of diagnosis had followup cystoscopy at median follow-up of 223 d.Survival curves were estimated using the Kaplan-Meier method, and differences in survival distributions were tested for significance using the log-rank test.The hazard ratio (HR) and its 95% confidence interval (CI) were estimated using a Cox proportional-hazards model.

Data analysis
Diagnostic accuracy was evaluated using receiver operating characteristic (ROC) analysis and calculation of the area under the ROC curve (AUC).Statistical significance was set at a p value of <0.05.Sensitivity and specificity were calculated in a binary manner for comparison of the clinical performance of the mePENK test to NMP22 and cytology tests using the McNemar test.CIs (95%), negative predictive values (NPVs), and positive predictive values (PPVs) were computed using MedCalc version9.3.2.0 (MedCalc Software Ltd., Ostend, Belgium).The incidence of recurrent BC was estimated on the basis of this study.

Study design and patient demographics
We included 54 patients with primary BC and 29 healthy individuals in a case-control study to determine optimal cutoff values for the mePENK test.A total of 188 participants were recruited for a prospective study, of whom two were excluded because of insufficient data.Of the 186 participants with results that could be fully evaluated, 59 (31.7%) had a recurrent bladder tumor at evaluation.Among these recurrent tumors, nine (15.3%) were stage Ta, 12 (20.3%)were carcinoma in situ (CIS), 30 (50.8%) were stage T1, and eight (13.6%) were stage T2-4.Demographic information for participants who were fully evaluated is presented in .Table 1 3.2.
We assessed the performance of the mePENK test in risk groups stratified according to the National Comprehensive Cancer Network classification [16].In the group with low/ intermediate risk, the mePENK test detected BC recurrence with sensitivity of 41.7% (5/12 ; 95% CI 15.7-72.3%),NPV of 95.4% (95% CI 92.8-97.1%),and PPV of 16.2% (95% CI 8.1-29.9%).For this subgroup, the sensitivity of the mePENK test did not significantly differ from that of the NMP22 test or cytology (p > 0.05).By contrast, for the high-risk group, the mePENK test had sensitivity of 89.7% (35/39) for detection of recurrent BC.The NPV was 96.9% (95% CI 92.5-98.8%)and the PPV was 61.5% (95% CI 51.0-71.0%).Notably, in the high-risk subgroup the sensitivity of the mePENK test was significantly superior to both the NMP22 test (p = 0.002) and cytology (p < 0.001), although its specificity was lower than for cytology (p < 0.001).Table 3 provides detailed results.

3.4.
Prognostic and predictive value of the mePENK test We evaluated the prognostic relevance of the mePENK test for BC surveillance in a group of 71 patients who were recurrence-free at the time of diagnosis and had follow-up cystoscopy at median follow-up of 223 d.Of these, 19/71  patients experienced BC recurrences.Importantly, median RFS was significantly shorter for mePENK-positive patients (n = 14; 245 d) than for mePENK-negative patients (n = 57; 503 d), with a HR of 9.4 (95% CI 2.4-35.9;p < 0.001, log-rank test; Fig. 3A).In detail, of the 14 mePENKpositive patients, 11 (78.5%) had BC recurrence, while only eight of 57 mePENK-negative patients (14.0%) experienced BC recurrence.This result indicates that patients with positive mePENK results had a significantly higher risk of tumor recurrence.In addition, if we assume that these patients had BC recurrence at the time of mePENK testing, the overall sensitivity for detection of BC recurrence is 80.0% (56/70), with specificity of 93.1%.The urinary NMP22 test also had prognostic power for RFS in BC (Fig. 3B).However, regardless of the NMP22 result, RFS was significantly shorter for mePENK-positive than for mePENK-negative patients.This result indicates that the mePENK test had better prognostic power than the NMP22 test (p = 0.0437 for the NMP22 + subgroup; p < 0.0001 for the NMP22 À subgroup, log-rank test; Fig. 3C,D).

Discussion
We previously demonstrated that the mePENK-LTE/qMSP test has promising potential for initial diagnosis of primary BC among patients with hematuria [15].The findings from the present study may add value to the potential clinical applications of the modified mePENK test beyond initial diagnosis, specifically in the context of monitoring for BC recurrence.
Cystoscopy with urine cytology is the standard method used for BC surveillance.Although cystoscopy is effective, it is expensive and invasive.Moreover, its sensitivity varies between 62% and 84%, depending on the tumor type, stage, and grade [17].Urine cytology remains a prevalent approach for BC diagnosis and surveillance, but it has shortcomings in terms of sensitivity [6].The utility of the urinary NMP22 test for NMIBC diagnosis and surveillance has been demonstrated, with sensitivity and specificity values for detection of primary or recurrent BC ranging from 49.5% to 55.7%, and from 85.7% to 87.3%, respectively [18][19][20].
Previous studies directly compared the ability of urine biomarkers, cytology, FISH, uCyt+, and NMP22 tests to predict the risk of NMIBC recurrence and progression during surveillance.Patients with biomarker-positive findings but negative cystoscopy have a higher risk of recurrence and progression; for patients with negative cytology, only NMP22 remains predictive for recurrence [21].Therefore, we chose cytology and the NMP22 test as a reasonable benchmark to evaluate the mePENK test for detection of BC recurrence during surveillance.The overall sensitivity of the mePENK test for detection of BC recurrence was 76.3%, with specificity of 85.0% and an NPV of 88.5%.The test had higher sensitivity for the high-risk group, at 89.7%, with an NPV of 96.9%.Notably, the mePENK test detected CIS with high sensitivity of 91.7%.
In general, when considering the intended use of a test for BC monitoring to exclude recurrence during follow-up, higher NPV is often considered more practical.Our study included 29 healthy individuals in the training set to establish the cutoff value, and only six of these exhibited a signal, ranging from 31.0 to 26.7.It is noteworthy that adjusting the cutoff from the original 32.0 to a lower value may result in more favorable test outcomes.For example, setting the cutoff to 26.7 yields sensitivity for the high-risk group of 100%, with a corresponding NPV of 100%.While this adjustment holds potential for allowing individuals with a negative result to forego invasive cystoscopy examinations, it is important to acknowledge that this improvement comes at the expense of specificity, which decreased to 38.6%.Therefore, a prospective clinical trial with a large-scale longitudinal design is imperative to ascertain the optimal cutoff that balances effectiveness and utility for use in clinical practice.Further research will contribute to refining the test performance and ensuring its reliable application in recurrence monitoring for BC.
The sensitivity of the mePENK test for detection of BC recurrence in the subgroup with low/intermediate risk was 41.7%.The relatively low sensitivity for this subgroup may be attributed to the small size or to the adherent nature of the recurrent tumors; at this stage, the tumors might not actively shed a detectable number of tumor cells into the urine [22].Patients with BC in the high-risk group have a lifelong risk of disease progression and mortality.Patients in the low-risk group have very low risk of progression: the 15-yr progression-free survival rate is 95%, with no cancerspecific mortality [23,24].Thus, in the clinical setting, identification of patients with high-risk BC is more critical for reduction of disease-related morbidity and mortality.In the high-risk setting, the specificity of the mePENK test was similar to that of the NMP22 test, but inferior to the specificity of cytology.
Molecular tests such as Xpert Bladder (five mRNA markers), CxBladder (five mRNA markers), AssureMDx (three gene mutations and three gene methylations), and Bladder EpiChek (15 gene methylations) are available for BC surveillance.These tests have sensitivity ranging from 68% to 82%, and specificity ranging from 80% to 88% [25].Our findings are comparable to results from clinical studies using these tests.Nonetheless, high-level evidence obtained from large-scale clinical studies is needed before biomarkerbased urine tests can be adopted in routine clinical practice.Only a few studies have evaluated the prognostic value of urine markers in follow-up surveillance for NMIBC, and the effectiveness of these markers has not been proven [21].Here we analyzed the prognostic relevance of the mePENK test for BC recurrence.Follow-up cystoscopy surveillance for cystoscopy-negative patients revealed that median RFS was significantly shorter for mePENK-positive than for mePENK-negative patients.This finding indicates that mePENK-positive patients had a significantly higher risk of tumor recurrence.The group with positive mePENK results had significantly shorter median RFS, regardless of NMP22 results.This finding indicates that the urinary mePENK methylation test performed better for prediction of BC recurrence than the NMP22 test.
Among 71 patients with negative cystoscopy results, 19 experienced a subsequent recurrence.Importantly, 11/19 patients (57.9%) had a positive mePENK test at the time of baseline cystoscopy, which confirms that the mePENK result was predictive of recurrence, as validated via follow-up cystoscopy [26].If we consider these 11 patients as having recurrent BC, the overall sensitivity and specificity for detection of BC recurrence improved to 80.0% and 93.1%, respectively.These findings suggest that a positive mePENK result is a useful tool for identification of patients with a high likelihood of BC recurrence.
Our study has some limitations, such as the small sample size, single-site enrollment, and short follow-up.Healthy individuals were used as the control group when determining the cutoff value for the test, and there is an age disparity between this group and the BC patients.Nevertheless, the performance findings provide valuable information and warrant validation of the mePENK test in a large-scale prospective study.

Conclusions
In our study, the urinary mePENK test demonstrated notable sensitivity in detecting recurrent BC, especially for high-risk cases, surpassing the performance of NMP22 and urine cytology.positive mePENK result correlated significantly with shorter RFS times and accurately predicted recurrence following cystoscopy.This finding underscores the potential of the mePENK test as a valuable tool for providing prognostic and predictive information in BC surveillance.Our study suggests that patients with negative mePENK results may benefit from extended intervals between cystoscopies in the surveillance of NMIBC.Use of the urine-based mePENK test with its NPV holds promise for reducing unnecessary cystoscopy procedures during follow-up for patients with negative results.Furthermore, lowering the cutoff value could increase the test sensitivity and maximize its NPV, with potential to exclude BC more effectively.Acknowledging that the optimal cutoff value may vary depending on the clinical context and objectives is crucial.Large-scale longitudinal clinical studies are imperative to validate and integrate these proposed adjustments into routine clinical practice.
Author contributions: Young-Deuk Choi had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
2) BC = bladder cancer.a Recurrence with an increase in tumor stage or grade in comparison to the previous visit.E U R O P E A N U R O L O G Y O P E N S C I E N C E 6 2 ( 2 0 2 4 ) 9 9 -1 0 6

Fig. 2 -
Fig. 2 -Cutoff determination for the mePENK test and clinical validation of the test for detection of recurrent BC in the prospective study.(A) Distribution of PENK methylation values (35-DC T ) for 83 urine samples.Error bars denote the 95% confidence intervals.(B) Receiver operating characteristic curve for detection of all bladder cancer cases.(C) Scatter plot of PENK methylation values (35-DC T ) for the 186 urine samples evaluated.Error bars denote the 95% confidence intervals.AUC = area under the curve; CIS = carcinoma in situ; C T = cycle threshold; H.C = healthy controls.* p < 0.001 (Kruskal-Wallis test).

Fig. 3 -
Fig. 3 -Recurrence-free survival for patients with non-muscle-invasive bladder cancer during follow-up surveillance after negative cystoscopy.(A) All patients stratified by mePENK status.(B) NMP22-positive subgroup stratified by mePENK status.(C) All patients stratified by NMP22 status.(D) NMP22negative subgroup stratified by mePENK status.

Table 1 -
Demographic and clinical characteristics of the study population

Table 2 -
Diagnostic performance of the mePENK test for detection of recurrent bladder cancer

Table 3 -
Clinical performance of the mePENK test in comparison to NMP22 and cytology, stratified by risk group a mePENK test versus NMP22; McNemar's test.b mePENK test versus cytology; McNemar's test.