Mixed‐methods approach to exploring patients’ perspectives on the acceptability of a urinary biomarker test in replacing cystoscopy for bladder cancer surveillance

Objectives To determine the minimal accepted sensitivity (MAS) of a urine biomarker that patients are willing to accept to replace cystoscopy and to assess qualitatively their views and reasons. Patients and Methods Patients were part of a prospective multicentre observational study recruiting people with bladder cancer for a urine biomarker study (DETECT II; ClinicalTrials.gov: NCT02781428). A mixed‐methods approach comprising (1) a questionnaire to assess patients’ experience with cystoscopy and patients’ preference for cystoscopy vs urinary biomarker, and (2) semi‐structured interviews to understand patient views, choice and reasons for their preference. Results A urine biomarker with an MAS of 90% would be accepted by 75.8% of patients. This was despite a high self‐reported prevalence of haematuria (51.0%), dysuria/lower urinary tract symptoms (69.1%) and urinary tract infection requiring antibiotics (25.8%). There was no association between MAS with patient demographics, adverse events experienced, cancer characteristics or distance of patients’ home to hospital. The qualitative analysis suggested that patients acknowledge that cystoscopy is invasive, embarrassing and associated with adverse events but are willing to tolerate the procedure because of its high sensitivity. Patients have confidence in cystoscopy and appreciate the visual diagnosis of cancer. Both low‐ and high‐risk patients would consider a biomarker with a reported sensitivity similar to that of cystoscopy. Conclusion Patients value the high sensitivity of cystoscopy despite the reported discomfort and adverse events experienced after it. The sensitivity of a urinary biomarker must be close to cystoscopy to gain patients’ acceptance.


Introduction
Bladder cancer is diagnosed in 8.0% of patients presenting with haematuria and, at initial diagnosis, 75% of cases are non-muscle-invasive bladder cancer (NMIBC) [1,2]. Patients with NMIBC have a 28-50% risk of recurrence and a 5-20% risk of progression at 5 years [3]. The risk of recurrence necessitates regular surveillance cystoscopy, and guidelines recommend a risk-adapted approach which can be as frequent as 3-monthly cystoscopy, with lifelong follow-up for high-risk disease [4]. The requirement for vigilant surveillance strategies is responsible for the high cost of healthcare associated with bladder cancer [5]. Cystoscopy has a sensitivity of 98% for detection of bladder cancer although it remains an invasive procedure associated with patient discomfort and a 5% risk of UTI [6,7].
The potential for urinary biomarkers for the detection of bladder cancer is an area of active research, and several biomarkers have been approved for use. Commercially available tests are licensed only as companion tests as they do not have the required diagnostic performance to replace cystoscopy (sensitivities of 57-82%) [8]. Novel urinary biomarkers have reported high sensitivities but often lack robust prospective validation [9,10]. Regardless of the performance of current or future urinary-based biomarkers, it is essential to understand patient perception and willingness to forgo cystoscopy for non-invasive testing.
To understand patient views relating to cystoscopy and the potential to integrate urinary-based biomarkers into a surveillance programme, we conducted a multicentre prospective observational study in which patients with NMIBC completed a questionnaire to determine the minimal accepted sensitivity (MAS) of a urine biomarker and we explored patients' willingness to accept a biomarker rather than cystoscopy. We also report patient experience and adverse events during or after flexible cystoscopy. Reasons for preference were assessed through a qualitative analysis using semi-structured interviews.

Study Design
Between September 2016 and April 2017, a total of 370 patients with histologically confirmed NMIBC and a minimum of 6 months' (two surveillance cystoscopies and two urine collection for biomarker testing) follow-up were recruited from 52 UK hospitals. Patients were sent questionnaires by post and 213 patients (57.6%) returned completed questionnaires. A total of 20 English-speaking patients from this cohort consented to participate in a semistructured telephone interview. Of these patients, four (20%) had low-grade cancers to ensure adequate representation of NMIBC. The full study protocol has been previously described [11]. The reported study represents a secondary endpoint for the DETECT II study (clinicaltrials.gov: NCT02781428).
Patient demographics, education level attained, and history of bladder cancer were recorded. Cancer stage was assessed using TNM WHO cancer classification [12]. Cancer risk was assessed using the European Association of Urology (EAU) risk classification [1]. Distance from patients' home to local hospital by private transport was calculated using www.ma ps.google.com.

Patient Questionnaire
We constructed a patient experience questionnaire after consultation with the UCL Health Behaviour Research Centre (author M.H.) as no validated questionnaire exists. The questionnaire domains assessed overall experience of cystoscopy, anxiety preceding cystoscopy and pain experienced, using a five-point Likert-scale. Preference for cystoscopy or urinary biomarker was assessed using the standard gamble method [13]. Cystoscopy was defined as having a sensitivity of 98% [6]. The MAS for a urinary biomarker was defined as the sensitivity at which patients expressed either a preference for a urine biomarker or were neutral about accepting either the biomarker or cystoscopy.

Qualitative Semi-Structured Interview
Each qualitative semi-structured interview lasted between 20 and 40 min. All interviews were carried out by one interviewer and patients were approached after a minimum of 6 months' follow-up from initial cancer diagnosis. The interview was designed to explore the following: experience of cystoscopy; perceived advantages and disadvantages of cystoscopy; perceived advantages and disadvantages of a urine biomarker; reasons for the preference for cystoscopy or a urine biomarker; perceived acceptable sensitivity of a urine biomarker for detection of bladder cancer; and preference for a surveillance pathway combining a urine biomarker interspaced with cystoscopy.

Statistical Methods and Data Analysis
Continuous data were reported using descriptive statistics such as mean, median, interquartile range and 95% CI. Categorical variables were compared using the chi-squared test. The t-test and ANOVA were used to compare the mean of continuous variables. SPSS v22 (IBM Corp, Armonk, New York, NY, USA) was used for statistical analysis and statistical significance was set at P < 0.05. Patient interviews were performed using Skype (Microsoft, Redmond, WA, USA) and recorded using an Evaer Skype recorder. Interview recordings were transcribed, and data were managed using Nvivo 11 (QSR International, Melbourne, Vic., Australia). A thematic analysis was used, open coding was performed by two researchers (W.S.T. and C.H.T.) on the first two transcripts and differences were resolved by discussion. Codes were assigned to sentences/paragraphs of transcripts based on the study objective. Axial coding was performed, and existing codes combined to create larger themes. One researcher (W.S.T.) continued to code remaining transcripts and any new emerging codes were discussed. Comparisons were made throughout the analysis to form the final framework. Background notes throughout all study phases were reviewed to avoid potential bias in reporting.

Ethics and Informed Consent
The corresponding author certifies that, when applicable, a statement(s) has been included in the manuscript documenting institutional review board, ethics committee or ethical review board study approval; principles of Helsinki Declaration were followed in lieu of formal ethics committee approval; institutional animal care and use committee approval; all human subjects provided written informed consent was obtained from all patients. DETECT II study protocol received Health Research Authority: London-Stanmore Research Ethics Committee approval on 30 August 2016 (IRAS project ID: 203022, REC reference: 16/LO/1044). This trial is registered on clinicaltrials.gov: NCT02781428.

Patient Demographics
A patient cohort flow diagram is shown in Fig. S1. Baseline characteristics and clinical-pathological variables for 213 patients are shown in Table 1. The median patient age was 74.0 years and 167 patients (78.4%) were men. Patients with a primary diagnosis of NMIBC in the preceding 6 months accounted for 62.9% of the cohort (n = 134). A total of 74.1% of patients (n = 158) had ≤5 previous cystoscopies. High-risk NMIBC according to EAU risk classification was confirmed in 40.3% of patients (n = 83), but only 17.8% of patients (n = 38) perceived their cancer as high risk.

Patient-Reported Adverse Events
The majority of patients experienced an adverse event after cystoscopy, with 77.5% (n = 165) reporting ≥1 adverse event ( Table 2). The self-reported prevalence of haematuria, dysuria/LUTS and UTI requiring antibiotics after cystoscopy was 46.9% (n = 100), 67.1% (n = 143) and 23.1% (n = 51), respectively. The significance of symptoms after cystoscopy is highlighted by the fact that 13.4% of patients had an unpleasant experience after the procedure. Moderate to significant pain was reported by 11.5% of patients and 10.0% of patients reported moderate to significant anxiety preceding cystoscopy (Fig. 1).

Minimum Acceptable Sensitivity for a Urine Biomarker to Replace Cystoscopy
A total of 78.6% (n = 163) of patients would be happy to use a urine biomarker if it was as sensitive as cystoscopy (Table 3), with 24.2% of patients prepared to accept a urine biomarker even if sensitivity was as low as 85%. There was no relation between patient experience of cystoscopy and MAS. Most patients (74.2%, n = 158) would accept the option to interspace cystoscopy with a urine biomarker to extend the time interval between surveillance cystoscopies. There was no difference in MAS according to patient demographics, previous adverse events or experience during or after cystoscopy, cancer characteristics or distance to hospital from home (Table 4). There was a trend towards significance between a lower MAS for men (P = 0.052) and patients without a recurrence in the previous 6 months (P = 0.078; Table 4). Patient confidence in the accuracy of cystoscopy was the top reason for choosing cystoscopy (70.5%), followed by reassurance by a clinician (51.9%) and a preference for a diagnostic test performed in hospital (51.4%; Table S1). The top reasons for choosing a urine biomarker was previous discomfort after cystoscopy (30.5%), avoiding a hospital visit (28.6%) and the non-invasive nature of the test (28.1%).

Qualitative Analysis
The demographics and tumour characteristics of the 20 patients interviewed are shown in Table S2. Main themes that emerged are shown in Table 5.

Views and Experience of Cystoscopy
Patients appreciate the fact that cystoscopy provides a visual diagnosis of cancer and attributed this to a presumed nearperfect sensitivity for the detection of bladder cancer. While patients did not like cystoscopy, they were prepared to tolerate it because of its good diagnostic ability. Patients also valued the fact that cystoscopy provides an instant diagnosis and appreciated that a healthcare professional performs the cystoscopy.
Patient perception of passing a cystoscope along the urethra was that it can be disturbing, although they recognize the requirement to visualize the bladder. Some patients described the procedure as embarrassing and felt violated after cystoscopy. Patients also appreciate that cystoscopy performed by an experienced urologist would reduce adverse events and patient discomfort.

Views and Experience of Urine Testing
Patients valued the convenience of a urine biomarker, reducing the need to attend hospital and be subjected to a procedure. Furthermore, patients appreciated that a urine biomarker was free of adverse events, unlike cystoscopy, and believed that a urine biomarker would allow earlier testing and subsequently allow prompt commencement of treatment.

Active Comparison Between Cystoscopy and Urine Test
When comparing between cystoscopy and the urine biomarker, patients were pragmatic and understood that no test is 100% accurate. Patients prioritized the test with the highest sensitivity and most would only accept a urine test with a similar sensitivity to cystoscopy. Missing bladder cancer during surveillance was a significant worry to patients, and patients with high-grade bladder cancer felt particularly concerned about missing recurrence and prioritized the high sensitivity of cystoscopy. Some patients' familiarity with cystoscopy and the fact that they had a positive experience with cystoscopic detection of cancer reinforced their preference for cystoscopy over a urine test. An overarching theme was that patients were not confident in the ability of a urine test to identify bladder cancer with a high sensitivity as they perceived it to be 'experimental' when compared to cystoscopy, the current 'gold standard'.
Patients who had previously experienced embarrassment related to cystoscopy were willing to accept a lower diagnostic sensitivity for a biomarker. All patients were open to interspacing cystoscopy with a urine biomarker to increase the interval between cystoscopies, although most reinforced the requirement for comparable sensitivity. Some patients expressed the opinion that a molecular urine test may potentially identify cancers before they are diagnosed visually. Further, some patients were skeptical about the ability of a biomarker which would be able to match cystoscopy.

Discussion
This represents the first study to explore qualitatively the views and decision-making of patients when considering between cystoscopy and a urine biomarker for the detection of bladder cancer recurrence. The majority of patients (75.8%) recognized biomarker performance as important and would discount any biomarker with MAS <90%; however, 63.3% of patients would accept a biomarker with a MAS of ≥95%, suggesting that patients' willingness to accept a biomarker is linked to its performance characteristics. Nevertheless, 21.3% of patients would prefer cystoscopy regardless of urine biomarker performance because of factors such as immediate readout and clinician interaction despite having to travel to a hospital.
Patient acceptance of cystoscopy was independent of experience of adverse events relating to the test, suggesting that the high sensitivity of cystoscopy is of paramount importance. This is similar to data from colonoscopy, where patients consider a high sensitivity to be of paramount importance with risk of adverse effects a secondary concern [14]. Our data suggest that the prevalence of complications after cystoscopy are not negligible, with 50% of patients' selfreporting haematuria and urinary symptoms after the procedure and 24% developing UTI requiring antibiotics. This is considerably higher than previous reports, although it represents a cumulative experience of patients who have had multiple cystoscopies and not an incidence rate [7].
We did not observe an association between patient demographics, education level and clinical-pathological variables with a lower MAS. In addition, higher disease stage, grade, actual risk or patient-perceived risk classification was not associated with a higher acceptable sensitivity. The data indicate that even patients with a lower risk of recurrence or progression place a high emphasis on accurate cancer detection. Patients have a perceived benefit for early detection of recurrence which is clear in high-risk bladder cancer, although limited data exist to support this in low-risk cancers. Observational reports suggest that active surveillance of patients with G1 pTa NMIBC does not increase   [16,17] report that patients who were older (>67 years) and those who experienced a higher pain intensity after cystoscopy, respectively, were significantly more likely to accept a lower sensitivity.
The present study differs from these two studies. Besides reporting the MAS of a urine biomarker that patients were prepared to accept, we also investigated the reasons for patients' preference for cystoscopy or urine biomarker using qualitative and quantitative methods. We did not find any variables associated with a lower MAS. A reason for this may be the fact that patients in the present study completed the questionnaire 6 months following a cancer diagnosis, suggesting a shorter time interval compared with the other two studies (minimum of 12 months [16] and median follow-up of 50 months [17]). The fear of cancer recurrence may be prioritized over pain attributed to cystoscopy [16,17]. Patients in both studies also did not experience the use of a urine biomarker and assumed that cystoscopy was 100% sensitive, which it is not in clinical practice.
Patients valued the visual element of cystoscopy as this was something they could relate to, and the absence of a visual diagnosis of cancer provides significant reassurance. Patients are aware that cystoscopy is the gold standard diagnostic test for bladder cancer surveillance and perceived a urine test as experimental, which may have introduced bias; however, if the sensitivity of a urine test were proven to be close to cystoscopy, 78.7% of patients would be happy to accept a urine test at intervals.
Limitations of the present study should be acknowledged. Complications such as UTI requiring antibiotics after cystoscopy were self-reported by patients and not confirmed by urine culture. In addition, the present study was not designed to report the incidence rate of UTI per cystoscopy performed, but to identify correlation between patient-reported and perceived adverse events and acceptability of a urine biomarker test. Further, all patients had at least one general anaesthetic for cystoscopy or transurethral resection of bladder cancer and, while the questionnaire was intended to gauge patient's perspectives on surveillance flexible cystoscopy, the patient's overall experience with any cystoscopy may have had an influence in shaping their opinion. It was assumed that patients would comprehend the questionnaire when completing it, as no medical terminology was used. Although all patients provided a urine sample for biomarker testing, they were not provided with the results of the biomarker. The knowledge of the biomarker results combined with a patients' experience with cystoscopy may affect the MAS. Finally, cystoscopies were nearly universally performed by urology trainees/residents and in more experienced hands, adverse events and patient's acceptability may be improved.
In conclusion, we report that patients with NMIBC valued the high sensitivity that cystoscopy accords despite the reported patient discomfort and adverse events experienced. Patients considered any diagnosis of cancer to be significant and were not willing to compromise on the diagnostic ability of a test. Hence, a diagnostic sensitivity of any urine biomarker must be close to that of cystoscopy before patients are prepared to accept it over cystoscopy.
the Medical Research Council. Additional funding was from the UCLH Biomedical Research Centre, the Urology Foundation and the Mason Medical Research Trust. It was supported both in the delivery and development by the UCL Surgical Interventional Trials Unit. NHS-associated cost supported all standard-of-care investigations and procedures.