Detection of Bladder Cancer Using a Pointof- Care Proteomic Assay.
Review of: H.B. Grossman, E. Messing, M. Soloway, K. Tomera, G. Katz, Y. Berger,Y. Shen. JAMA 293, 810-816, 2005.
In this prospective study the urine-based NMP22 point of care dip-stick test was evaluated. The study design was that in 22 participating centres in the US patients with risk factors or symptoms of bladder cancer who were to undergo cystoscopy gave urine samples for evaluation by NMP22 dip stick and standard urine cytology. Altogether, 1,331 patients were examined and among these in 79 cases bladder cancer was found. In 44 of these 79 cases, the NMP22 test was positive (sensitivity 55.7%), while cytology was positive in only 12 of 76 cases (sensitivity 15.8%). Specificity was 85.7% for NMP22 and 99.2% for cytology. The authors conclude that the point of care test can improve diagnosis in conjunction with cystoscopy.
This study was designed, supervised and funded by Matritech Inc. and served as the basis for FDA approval of the new NMP22 dipstick test. However, design and conduct of the study and presentation of the findings deserve some critical comments. The inclusion criteria are ill-defined and vague and suggest that almost anybody undergoing cystoscopy could be included. The urinary cytological methods used were not standardized since each of the 22 centres sent urine for cytology to their regular laboratories. Thus, it is highly likely that urine conservation, cell asservation and staining methods were different, if used at all. 7/79 patients counted as positive for carcinoma had no histological verification at all (of these 4 had a positive NMP22 test but none a positive cytology). Patients with a negative cystoscopy who for unspecified reasons underwent repeat cystoscopy within 3 months and were then found to have tumour (n=9) were counted as positive for the initial evaluation.
In the analysis of their data the authors stress the relatively higher sensitivity of the NMP22 test compared to cytology in their study. They altogether fail to mention the fact that in their data the dip-stick test had a false-positive rate of 80%. Thus, there are serious limitations to this study which are not discussed by the authors. The number of positive cases is far to small to assess test sensitivity adequately let alone differentiate tumour grade-dependent sensitivity. The test used for comparison, urine cytology, was well below average sensitivity reported in the literature. The authors conclusion that the NMP22 dip-stick test with a sensitivity of 56% can improve bladder cancer diagnosis seems rather doubtful when this test in the same study has a false-positive rate of 80% and a positive predictive value of only 19.7%.