Diagnostic accuracy of photodynamic diagnosis with 5-aminolevulinic acid, hexaminolevulinate and narrow band imaging for non-muscle invasive bladder cancer.

To assess the diagnostic test accuracy (DTA) of photodynamic diagnosis with 5-aminolaevulinic acid (5-ALA), hexylaminolevulinate (HAL) and narrow band imaging (NBI) for non-muscle-invasive bladder cancer (NMIBC), with white light-guided cystoscopy (WLC) as reference standard. A systematic review and narrative synthesis was performed in accordance with PRISMA. Major electronic databases were searched until 20 May 2019. All studies assessing the DTA of 5-ALA, HAL and NBI compared with WLC at patient and lesion-level were included. Relevant sensitivity analyses and risk of bias (RoB) assessment were undertaken. 26 studies recruiting 3979 patients were eligible for inclusion. For patient-level analysis, NBI appeared to be the best (median sensitivity (SSY) 100%, median specificity (SPY) 68.45%, median positive predictive value (PPV) 90.75%, median negative predictive value (NPV) 100% and median false positive rate (FPR) 31.55%), showing better DTA outcomes than either HAL or 5-ALA. For lesion-level analysis, median SSY across NBI, HAL and 5-ALA were 93.08% (IQR 87.04-98.81%), 93.16% (IQR 91.48-97.04%) and 94.42% (IQR 82.37-95.73%) respectively. As for FPR, median values for NBI, HAL and 5-ALA were 20.40% (IQR 13.68-27.36%), 17.43% (IQR 12.79-22.40%) and 28.12% (IQR 22.08-42.39%), respectively. Sensitivity analyses based on studies with low to moderate RoB and studies with n>100 patients show similar findings. NBI appears to outperform 5-ALA and HAL in terms of diagnostic accuracy. All three modalities present high FPR, hence indicating the ability to detect additional cases and lesions beyond WLC.