Interobserver agreement of confocal laser endomicroscopy for bladder cancer.

Emerging optical imaging technologies such as confocal laser endomicroscopy (CLE) hold promise in improving bladder cancer diagnosis. The purpose of this study was to determine the interobserver agreement of image interpretation using CLE for bladder cancer. Experienced CLE urologists (n=2), novice CLE urologists (n=6), pathologists (n=4), and nonclinical researchers (n=5) were recruited to participate in a 2-hour computer-based training consisting of a teaching and validation set of intraoperative white light cystoscopy (WLC) and CLE video sequences from patients undergoing transurethral resection of bladder tumor. Interobserver agreement was determined using the κ statistic. Of the 31 bladder regions analyzed, 19 were cancer and 12 were benign. For cancer diagnosis, experienced CLE urologists had substantial agreement for both CLE and WLC+CLE (90%, κ 0.80) compared with moderate agreement for WLC alone (74%, κ 0.46), while novice CLE urologists had moderate agreement for CLE (77%, κ 0.55), WLC (78%, κ 0.54), and WLC+CLE (80%, κ 0.59). Pathologists had substantial agreement for CLE (81%, κ 0.61), and nonclinical researchers had moderate agreement (77%, κ 0.49) in cancer diagnosis. For cancer grading, experienced CLE urologists had fair to moderate agreement for CLE (68%, κ 0.64), WLC (74%, κ 0.67), and WLC+CLE (53%, κ 0.33), as did novice CLE urologists for CLE (53%, κ 0.39), WLC (66%, κ 0.50), and WLC+CLE (61%, κ 0.49). Pathologists (65%, κ 0.55) and nonclinical researchers (61%, κ 0.56) both had moderate agreement for CLE in cancer grading. CLE is an adoptable technology for cancer diagnosis in novice CLE observers after a short training with moderate interobserver agreement and diagnostic accuracy similar to WLC alone. Experienced CLE observers may be capable of achieving substantial levels of agreement for cancer diagnosis that is higher than with WLC alone.