This review concluded that photodynamic diagnosis had higher sensitivity but lower specificity than white light cystoscopy for detecting bladder cancer. Strategies that involved biomarkers and/or photodynamic diagnosis provided additional benefits at a cost that society might be willing to pay. The authors’ detailed and cautious conclusions are likely to be reliable.

To assess the test accuracy and clinical effectiveness of photodynamic diagnosis compared with white light cystoscopy and urine biomarkers (FISH, ImmunoCyt, NMP22) and cytology for detection and follow-up of bladder cancer.

Ten databases including MEDLINE, BIOSIS Previews, EMBASE and Science Citation Index were searched for published and ongoing studies, supplemented by screening of reference lists. Further databases were searched to identify relevant systematic reviews and other background information. Search terms were reported.

Randomised controlled trials (RCTs), non-randomised comparative studies and diagnostic cross-sectional studies that reported the absolute numbers of true and false positives and negatives were eligible for the evaluation of test performance. Only RCTs were eligible for the evaluation of effectiveness. Participants with symptoms suspicious for bladder cancer or previously diagnosed with non-muscle-invasive disease were eligible for inclusion. Studies were eligible for inclusion if they evaluated the following test comparisons: photodynamic diagnosis compared with white light cystoscopy; or FISH, ImmunoCyt, NMP22 or cytology, with a reference standard of histopathological examination of biopsied tissue.

Where reported in test accuracy studies, 41% of participants were suspected for bladder cancer and 59% had previously diagnosed bladder cancer. Where reported, 76% of participants were male and the median of mean ages was 67 years (range 52 to 72). The photosensitising agent used was 5-aminolaevulnic acid in 75% of cases, hexaminolaevulinate in 17% of cases, hypericin in 3% and either 5-aminolaevulnic acid or hexaminolaevulinate in 5%. In RCTs that reported recurrence/progression, mean age was 67 years, 73% of participants were male, 5-aminolaevulnic acid was used in all analysed cases and length of follow-up ranged from less than two weeks to eight years. Adverse event data were also reported.

Two reviewers independently assessed studies for inclusion, with disagreements resolved by a third reviewer.

Quality of included test accuracy studies was assessed with a modified version of the QUADAS tool. RCTs were assessed in the same manner, using a published 14-item checklist.

Two reviewers independently assessed the quality, with any disagreements resolved by a third reviewer

Key characteristics were extracted from included studies by one reviewer and checked by a second, with any disagreements resolved by a third. Sensitivity, specificity, predictive values, likelihood ratios and diagnostic odds ratios were calculated for test accuracy studies. Relative risks (RRs) were calculated for effectiveness outcomes.

Summary receiver operating characteristic (SROC) curves were presented for different levels of analysis, and pooled median sensitivity, specificity, likelihood ratio and DOR values with 95% confidence intervals (CI) were reported for each model. Meta-analysis models were fitted using hierarchical summary receiver operating characteristic (HSROC) curves. For effectiveness outcomes following transurethral resection of bladder tumour, pooled relative risk values with 95% CIs were calculated using a Mantel-Haenszel fixed-effect model in the absence of statistical heterogeneity. Χ² and Ι² statistics were used to evaluate statistical heterogeneity. Where studies could not be combined statistically, a narrative synthesis was provided.

Twenty-seven studies (2,949 patients) evaluated photodynamic diagnosis test accuracy. Direct evidence (patient-level analysis) suggested photodynamic diagnosis had higher sensitivity than white light cystoscopy (92%, 95% CI 80 to 100, versus 71%, 95% CI 49 to 93) but lower specificity (57%, 95% CI 36 to 79, versus 72%, 95% CI 47 to 96). For biopsy-level analysis, photodynamic diagnosis also had higher sensitivity than white light cystoscopy (93%, 95% CI 90 to 96, versus 65%, 95% CI 55 to 74) but lower specificity (60%, 95% CI 49 to 71, versus 81%, 95% CI 73 to 90).

Median sensitivities of photodynamic diagnosis and white light cystoscopy for detecting lower risk, less aggressive tumours were broadly similar for patient-level detection (92%, 20 to 95%, versus 95%, 8 to 100%) but sensitivity was higher for photodynamic diagnosis than for white light cystoscopy for biopsy-level detection (96%, 88 to 100%, versus 88%, 74 to 100%). For the detection of more aggressive, higher risk tumours the median sensitivity of photodynamic diagnosis for both patient-level (89%, 6 to 100%) and biopsy-level (99%, 54 to 100%) detection was higher than those of white light cystoscopy (56%, 0 to 100%, and 67%, 0 to 100%). For carcinoma in situ alone, photodynamic diagnosis was more sensitive at both patient-level (83%, 41 to 100%, versus 32%, 0 to 83%) and biopsy-level (86%, 54 to 100%, versus 50%, 0 to 68%).

Four RCTs (709 patients) reported effectiveness outcomes. Compared with white light cystoscopy, the use of photodynamic diagnosis at transurethral resection of bladder tumour resulted in fewer residual tumours at check cystoscopy (RR 0.37, 95% CI 0.20 to 0.69) and longer recurrence-free survival (RR 1.37, 95% CI 1.18 to 1.59). Longer-term outcomes for these measures were less clear.

Seventy-one studies evaluated biomarkers and cytology in detecting bladder cancer. Fourteen studies (3,321 participants) evaluated FISH, 10 studies (4,199 participants) evaluated ImmunoCyt, 41 studies (13,885 participants) evaluated NMP22, and 56 studies (22,260 participants) evaluated cytology. Based on indirect evidence, FISH (76%, 95% CI 65 to 84), ImmunoCyt (84%, 95% CI 77 to 91) and NMP22 (68%, 95% CI 62 to 74) all had higher sensitivity than cytology (44%, 95% CI 38 to 51). However, cytology had higher specificity (96%, 95% CI 94 to 98) than FISH (85%, 95% CI 78 to 92), ImmunoCyt (75%, 95% CI 68 to 83) and NMP22 (79%, 95% CI 74 to 84).

The most effective strategy was flexible cystoscopy and ImmunoCyt followed by photodynamic diagnosis in initial diagnosis and flexible cystoscopy followed by white light cystoscopy in follow-up; this had an incremental cost per life-year of over £270,000. The least effective strategy was cytology followed by white light cystoscopy in initial diagnosis, which was most likely to be considered cost-effective when society’s willingness to pay was less than £20,000 per life-year. No strategy was cost-effective more than 50% of the time, but four of the eight strategies in the probabilistic sensitivity analysis (three involving a biomarker or photodynamic diagnosis) were each associated with a 20% chance of being considered cost-effective.

Photodynamic diagnosis’s higher sensitivity in detecting bladder cancer must be weighed against its disadvantage of having a higher false positive rate. Taking into account the assumptions made in the model, strategies that involved biomarkers and/or photodynamic diagnosis provided additional benefits at a cost that society might be willing to pay. Strategies that replaced white light cystoscopy with photodynamic diagnosis provided more life-years but it was unclear whether they were worth the extra cost.

This review was based on a clearly defined research question that was supported by relevant inclusion criteria. Efforts were made to identify all relevant publications and to minimise the potential for errors and bias throughout the review process. Appropriate methods were used to synthesise the results of individual studies and key aspects of study quality were taken into account. The authors' cautious conclusions reflected the evidence presented and were likely to be reliable.

Practice: The authors stated that, given the evidence presented, a judgement needed to be made as to whether the "standard" strategies for diagnosis and follow-up of bladder cancer should be altered. They provided extensive discussion of the available evidence and remaining uncertainties.

Research: The authors made several recommendations for future research, such as the need for RCTs that compared photodynamic diagnosis with rigid white light cystoscopy plus adjuvant intravesical therapy at transurethral resection of bladder tumour in patients presumed to have non-muscle invasive bladder cancer. Diagnostic cross-sectional studies that compared FISH with ImmunoCyt, NMP22 BladderChek point of care test and voided urine cytology (plus combinations of these tests) against a reference standard of cystoscopy with histological assessment of biopsied tissue in the same patient population were needed.

NIHR Health Technology Assessment Programme.

Mowatt G, N'Dow J, Vale L, Nabi G, Boachie C, Cook JA, Fraser C, Griffiths TR, Aberdeen Technology Assessment Review (TAR) Group. Photodynamic diagnosis of bladder cancer compared with white light cystoscopy: systematic review and meta-analysis. International Journal of Technology Assessment in Health Care 2011; 27(1): 3-10.

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.