The causes of asymptomatic haematuria are numerous. As such the investigating algorithm for haematuria is composed of a number of tests. Approximately 600 patients are accepted each year to the Canterbury District Health Board (DHB) Urology Department for evaluation of haematuria. All referrals are accepted if they have laboratory confirmation of haematuria and the investigations completed.

One of the most common important causes of haematuria is bladder cancer. While a number of these are detected on imaging, the ‘gold standard’ for diagnosing bladder cancer is cystoscopy.1 While generally well-tolerated by patients, flexible cystoscopy is uncomfortable and may have adverse post-procedural consequences.2,3 Anecdotally, it is the test in the haematuria algorithm least cherished by patients and also necessitates patients seeing a specialist urologist. If it were safe not to undertake cystoscopy in a group of patients presenting with haematuria, then their work-up could potentially be completed by clinicians other than a urologist, such as a general practitioner (GP).

The need to improve risk stratification of patients who may require cystoscopy and imaging was emphasised in a recent review of guidelines for assessing microhaematuria.4 Numerous biomarkers have been identified in urine or blood samples that have the potential to detect and monitor bladder cancers.5–9 Current trends have moved towards measuring the expression of microRNAs (miRNAs), small non-coding RNAs that regulate genes involved in cancer development, progression and metastasis.10–12 While incorporating multiparametric assays of miRNAs in algorithms may improve diagnostic accuracy suitable for clinical application,13 there remain concerns regarding the ability of these assays to detect low-grade tumours and their relatively high rate of false positive results.6,14,15 To date, no biomarker or combination of biomarkers has sufficient validation to serve as a reliable alternative to cystoscopy for detecting bladder malignancies.

Cxbladder TriageTM (CxbT) measures the expression of five urinary miRNA biomarkers (CDC2, MDK, IGFBP5, HOXA-13 and CXCR2) using the reverse transcriptase quantitative polymerase chain reaction (RT-qPCR) method.16,17 The expression levels when combined with four phenotypic factors (age, sex, smoking status and haematuria frequency) are used to calculate a segregation index that identifies patients with a low risk of having an invasive transitional cell carcinoma (TCC).17–19 Using a previously validated cut-off value of <4.0,17 the segregation index has been reported to have a sensitivity of 96% and a negative predictive value (NPV) of 97%, with the potential to reduce the number of flexible cystoscopies in patients investigated for haematuria by approximately 40%.20,21 Red blood cells (<105 cells/ml of urine) and haemolysis (ie, visible haematuria) have no effect on the assay, although a marked inflammatory response is known to interfere with the measurements [Personal communication, Pacific Edge Ltd, Dunedin New Zealand].

Prior to April 2016, GPs investigated patients with haematuria according to a clinical pathway on a local clinical guidance website https://www.healthpathwayscommunity.org/Home.aspx. In that pathway, all patients received MSU urinalysis, culture and microscopy, urine cytology, imaging (ultrasound of the renal tract or computed tomography [CT-IVU]), cystoscopy and specialist opinion. Microhaematuria was investigated using ultrasound and macrohaematuria by CT-IVU, except in patients aged <40 years or >85 years who underwent ultrasound regardless of the type of haematuria. This stratification of patients was based on prior local assessment of 2,436 patients treated at a haematuria clinic.

From April 2016 onwards the CxbT test was added to the laboratory assessment section of the clinical pathway. All patients referred by their GP for public or private urological assessment of haematuria from this time until May 2017 were included in a review of the addition of CxbT. Investigations were otherwise identical to those used before the addition of CxbT. The data were then used to design and evaluate a theoretical pathway that would have the potential to safely reduce the number of patients requiring a cystoscopy.

This paper summarises our review and evaluation of adding CxbT to the laboratory testing section of a clinical pathway used to investigate patients with haematuria.

Patients and methods

Study design

The work was carried out in the Canterbury region of New Zealand. The addition of CxbT to the routine investigation of haematuria was carried out following consensus between local clinicians. No other change to clinical practice was made and there was no randomisation or control group.The Health and Disability Ethics Committee, Ministry of Health, New Zealand advised that the review did not require ethical approval, as it constituted monitoring and improvement of usual patient care carried out by the Canterbury DHB.

The prevalence of bladder cancer in patients with microhaematuria has been reported to be 2–4%22 and in those with macrohaematuria between 8–14%.18,23 Based on these findings and assuming a 10% prevalence of bladder cancer, inclusion of at least 500 patients in the review would provide sufficient statistical power to allow precise estimates of both the sensitivity and negative predictive value (NPV) for CxbT alone and in combination with the clinical pathway. The exact 95% confidence interval (CI) for sensitivity would be 83.5–98.7%, and for 98% NPV the exact 95% CI would be 93.8–99.3%. The review was therefore carried out over a 13-month period to obtain the required number of patients.

Patients

The clinical outcomes of 571 patients investigated for haematuria by their GP were reviewed. Four hundred and seventy-eight patients were referred for urological assessment, 73 were managed solely by their GP and did not proceed to cystoscopy, while 20 were excluded from the final analysis, 10 because of sampling or assay problems with CxbT (high levels of inflammatory markers n=4, inadequate sample volume n=4, expired sample tube n=1 and excessive blood in sample n=1), and 10 for other reasons (non-attendance at urologist appointment n=6, and incomplete clinical data n=4).

Collection of the urine samples for the Cxbladder TriageTM assay

A single mid-stream urine sample for the CxbT assay was collected from each patient, preferably from the second void of the day. A 4.5ml aliquot of this sample was transferred immediately to a stabilisation liquid via vacuum-driven aspiration, followed by storage at 4oC until assayed as described previously.18

Data collection

Data was collected in a non-blinded manner and included clinical and demographic characteristics of the patients and the findings of laboratory tests, imaging, cystoscopies, histology and specialist assessment. A cystoscopy and histology were required for diagnosis of bladder cancer. Patients not referred for specialist urological assessment (n=73) were followed-up by review of their medical records for at least two years to ensure that no bladder malignancy had been missed.

Statistical analysis

The diagnostic accuracy of CxbT was evaluated by calculation of sensitivity, specificity, NPV and likelihood ratio. Continuous data in the referred and non-referred patients were compared using Welch’s unpaired t test and the distribution of categorical data in the two groups compared using Chi square and Fisher’s exact tests where appropriate. A p value ≤0.05 was considered statistically significant.

Design and evaluation of theoretical pathway

Patients were only included in the analysis and subsequent modeling of a theoretical pathway if they had been referred to a urologist. A pathway was developed that did not include cytology in the laboratory tests, and only referred patients for secondary urological assessment and cystoscopy if the CxbT index or imaging was positive. The sensitivity, specificity and NPV of this theoretical pathway to detect bladder cancer using a CxbT segregation index cut-off value of <4.017 was calculated as the test alone, and within the context of the pathway.

Results

The clinical and demographic characteristics, results of the laboratory and imaging investigations and diagnoses of the 478 patients referred for urological assessment and the 73 patients managed solely by their GP are summarised in Table 1. All the patients lived in the funded area of the Canterbury District Health Board, and were predominantly middle-aged or older, with two-thirds being male. Approximately 50% of the referred patients were classified as having an increased risk for bladder cancer because of their smoking history (n=230) or having previously received radiation therapy of the pelvis (n=8). In comparison, only 26% of the non-referred patients were at increased risk, solely because of their smoking history. The proportion of patients with macrohaematuria was significantly higher in the referred patients than in those who were not referred (70% vs 44%). The mean CxbT score in the two groups was not significantly different, although the results indicated 69% of the referred patients required further investigation compared to 33% of the non-referred patients.

Table 1: Clinical and demographic characteristics, diagnosis, and results of laboratory tests of the 478 referred patients and 73 non-referred patients.

‍

aFisher’s exact test.
bWelch’s unpaired t test.
cChi square test.

Detection of bladder cancers

The number of bladder cancers detected by the clinical pathway with imaging and CxbT, CxbTalone, or urine cytology alone in the referred patients is shown in Table 2. Forty-four patients were diagnosed with a TCC, with two of these lesions being missed by CxbT, giving the test a sensitivity of 95.5% (95% CI 84.5–99.4) and NPV of 98.6% (95%CI 95.3–99.8%). The high proportion of false positive tests however resulted in a low specificity of 34.3% (29.9–39.0). Expressed as likelihood ratios, for a positive test CxbT had a ratio of 1.45, indicating the test result was moderately associated with bladder cancer, and for a negative test, a ratio of 0.07, indicating the result was strongly associated with absence of disease.

Table 2: The number of bladder cancers detected in the 478 referred patients with haematuria by the model pathway with imaging and Cxbladder TriageTM, Cxbladder TriageTM alone, and urine cytology alone. The bottom panel shows the stage and grading of the cancers grouped according to the type of haematuria.

‍

The first of the missed lesions was a papillary pTaG1 tumour smaller than 1cm, which was seen on ultrasound. The second missed lesion was 2mm in size and papillary in appearance, although no histology was obtained as the specimen was destroyed in the resection process. The operating surgeon judged the appearance to be likely either a pTaG1 tumour or a papilloma. When combined with imaging, assessment only missed the 2mm papillary lesion, giving the pathway a sensitivity of 97.7% (95% CI 88.0–99.9) and NPV of 99.8% (95% CI 98.7–99.9). Six patients with a positive CxbT and bladder cancer had negative urine cytology and imaging results.

For data stratified according to haematuria type the diagnostic accuracy of CxbTwas; macrohaematuria, sensitivity 95.1% (95% CI 83.5–99.4), specificity 32.8% (95% CI 27.4–38.5) and NPV of 98.0% (95%CI 92.8–99.8%); microhaematuria, sensitivity 100.0%% (95% CI 29.2–100.0), specificity 42.6% (95% CI 34.3–51.5) and NPV of 100.0% (95%CI 94.0–100.0%).

In comparison, urine cytology only detected 22 of the 44 TCCs giving a sensitivity of 50% (95% CI 26.3–56.8), while the positive predictive value was also low at 72% (95% CI 53.3–90.2). In nine of the 22 cases with negative urine cytology, imaging also failed to detect the bladder lesion, although in eight of these cases CxbT indicated referral to a urologist was warranted. Further, none of the one abnormal or eight atypical cytology reports with a CxbTindex <4.0 had a malignancy.

Of the 73 non-referred patients, none subsequently presented with bladder cancer, and the GPs of all these patients were reminded to check appropriate care had been provided.

Causes of haematuria

The causes of the haematuria are outlined in Table 3, along with the likelihood of being CxbT positive and the percentage of each cause of haematuria that would have been detected by the modified pathway.

Table 3: Causes of haematuria in the review cohort.

‍

Development of the theoretical pathway

The findings were used to construct a new clinical pathway that involved a GP requesting MSU urinalysis, culture and sensitivity and CxbTas the initial laboratory screening tests for bladder cancer in combination with appropriate imaging based on the age and type of haematuria of the patient. In this pathway, urine cytology would only be requested in patients with a CxbT>4.0, who were referred for specialist assessment and cystoscopy.

Figure 1 shows the flow of patients if they had gone through the theoretical clinical pathway. In 60 (42%) of patients with microscopic haematuria and 91 (27%) with macroscopic haematuria the Cxbladder Triage and imaging results indicated no further urological assessment or a cystoscopy was required (total 151, 32% of patients). No invasive or high grade TCC would have been missed by the new pathway.

Figure 1: Flow of patients through the hypothetical clinical pathway for investigation of patients with haematuria.

Discussion

Urinary biomarkers of bladder cancer are not currently recommended in the laboratory investigation of haematuria because of their low specificity and limited accuracy in low-stage and low-grade tumours.15,24 The review described in this paper was novel as it collected prospective data following the inclusion of CxbT in a haematuria assessment algorithm, with the patients included solely on the basis that their GP had requested a CxbT test. This approach was observation of usual care and provided real-time data on clinical utilisation of the test and would potentially minimise selection bias. The data were then used to inform further development of the haematuria clinical pathway. The pathway developed leveraged the high NPV of the CxbT segregation index to identify patients who do not require a primary care urinary cytological evaluation or a secondary care appointment and cystoscopy.

The risk of avoiding a cystoscopy in the evaluation of haematuria is principally the risk of missing a significant bladder cancer. Our results showed a false negative CxbT result was obtained in only two of 44 patients diagnosed with bladder cancer, with both being low-grade superficial urothelial lesions. The parameters of diagnostic accuracy we measured for CxbT are similar to those reported previously,16–18 with a sensitivity of 95% and NPV >98%.

In our theoretical pathway, CxbT is not used in isolation, but as part of an algorithm. The performance of this pathway would have missed a single very low-risk bladder lesion. In other words, 150 cystoscopies in CxbT negative patients were performed to detect a single very low-risk lesion. This is consistent with published studies of other Cxbladder tumour markers where a negative test excludes invasive and high-grade cancers.17–20 The risk of missing a significant cancer from the adoption of the theoretical pathway appears very low and clinically acceptable. Similar results and the ability to predict patients with a high risk of bladder cancer have been reported for other urinary markers such as immunocytology8 and methylation products combined with standard urine cytology.9 However, further large prospective cohort studies are necessary to prove the true clinical value of inclusion of these biomarkers in investigative pathways.

There can be other lower urinary tract causes of asymptomatic haematuria. These are highlighted in Table 2. While the numbers were small in this study, no malignancies, bladder stones or benign lesions of the bladder were missed by the clinical pathway. Two cases thought to have had an inflammatory cause (one with normal cystoscopy and vulval inflammation, and the other with asymptomatic trigonal squamous metaplasia), were not detected. The risk of missing other significant lower tract pathology also appears to be negligible.

The model pathway we constructed and evaluated involves CxbT being the sole urine test to screen for bladder cancers in patients with confirmed haematuria, with urine cytology only being requested for patients referred for specialist urological assessment. It is consistent with several reviews that concluded urine cytology has insufficient sensitivity to rule out malignancy or exclude patients with haematuria from further investigations. 25,26 The continued use of urinary cytology in those CxbTpositive patients presenting for secondary care and cystoscopy reflects local unit practice of doing upper tract endoscopy where there is positive cytology and no lower tract cause found.

The majority of patients treated in secondary care within the New Zealand health system are seen though publicly funded DHBs, with access to services tightly managed to capacity using strict criteria and careful review of referrals. The secondary care system is supported by a robust and competent primary care sector through GPs. As such, in our health system the avoidance of cystoscopy in the haematuria algorithm would allow the assessment of haematuria in patients with a negative CxbT to be undertaken in primary care with oversight of specialist urologists.

The theoretical pathway has the potential to save considerable amounts of these resources without severely compromising clinical safety. Although we acknowledge this pathway differs from current international guidelines,27 it is pragmatic in our setting and is the best utilisation of resources in the financially capped New Zealand public health system. Our review showed approximately one-third of patients can be assessed without the need for a cystoscopy, and in our health system, we judge that these patients can remain in primary care without being referred to secondary care for specialist review. In our unit this would free up an estimated 200 patient consultations a year and allow management of conditions that might not otherwise reach the thresholds for referral. Importantly, the patient with haematuria would also safely avoid the social disruption and discomfort of a secondary care visit for cystoscopy.

The cost of seeing all patients with haematuria in secondary care, even if not performing a cystoscopy, will vary by health system. It is anticipated that, in the future, new-generation multiparametric assays will have greater specificity and potentially cost less, thereby improving further the financial benefits of using these assays in the investigation of haematuria.

A third of patients will be CxbT negative and therefore not need cystoscopies. While there may be other savings with less urine cytology tests and freed-up urologist time, the relevant workforce will be used in other areas of pathological and clinical endeavour, thus our assumption that the saving of a third of the price for cystoscopies should be around the cost neutral price for the CxbT test. This is simplistic, but there will clearly be more productivity in other areas from cytologists and urologists, allowing a rationed health system to dig deeper into diseases in the community.

Conclusions

This study adds to the increasing evidence that urinary mRNA biomarkers have a place in the assessment of haematuria. When clinicians are provided with CxbT results in combination with imaging they are able to reliably identify patients in whom cystoscopy can be avoided with negligible risk. In our health system we judge that these CxbT-negative patients can also be assessed in primary care without the need for secondary care referral.

This new haematuria assessment algorithm was adopted into the Canterbury Community HealthPathways in February 2018 and continuous audit has been carried out over the last year to ensure patient safety. Data on approximately 890 patients managed using the new pathway will be published when follow-up is complete. The new pathway should be applicable in any health system with effective general practice or primary care and the ability to inform GPs of locally recommended assessment and management of haematuria. Health systems with less constrained urological specialist services might continue to choose to recommend specialist referral for investigation of all patients with haematuria.

‍