Impact of clinical factors, including a point-of-care nuclear matrix protein-22 assay and cytology, on bladder cancer detection.

OBJECTIVE

To determine whether the nuclear matrix protein-22 (NMP22) assay can improve the accuracy of discriminating between high-risk patients with and without bladder cancer.

PATIENTS AND METHODS

Age, gender, race, smoking status, haematuria and its extent, and the NMP22 and urinary cytology results, were available for 1272 patients. The data of 670 (52.7%) from four study sites were used to develop a logistic regression model-based nomogram to predict the presence of bladder cancer. The remaining data from 602 (47.3%) patients from nine study sites were used to externally validate the nomogram. A separate nomogram was developed for urinary cytology, and for the combination of NMP22 and urinary cytology findings.

RESULTS

Of 1272 patients, 76 (6.0%) had bladder cancer, 217 (17.1%) were NMP22-positive and 17 (1.3%) had malignant cells on urinary cytology. NMP22 and urinary cytology results were independent predictors of bladder cancer (P = 0.005 and 0.007, respectively). In external validation, the area under the curve (AUC) for NMP22 was 76.0% vs 56.2% for cytology. External validation of the multivariable NMP22-based bladder cancer nomogram gave an AUC of 82.4% vs 74.7% for the multivariable cytology-based nomogram (gain 7.7%; P = 0.006) vs 82.6% for the multivariable nomogram combining NMP22 and cytology results (gain 0.2%; P = 0.1).

CONCLUSIONS

The ability of the NMP22 test to predict bladder cancer in high-risk patients significantly exceeds that of urinary cytology. The NMP22-based nomogram can help to identify individuals at risk of bladder cancer.