Using a novel experimental model, we evaluated a number of factors that may influence the NMP-22 urine tests. Although the model does not mimic the actual physiological situation exactly, the aim was to test whether the NMP-22 analyte could originate from blood (serum or whole blood) or benign cells of urothelial origin. Previous studies have investigated the influence of blood in the NMP-22 assay. Atsu et al.
reported that when whole blood was added to a urine sample, positive NMP-22 results increased in parallel with the increase in the amount of red blood cells in the sediment. In addition, the leukocyte count in the urine sediment also had a significant impact, and the investigators concluded that pyuria and hematuria significantly affected urinary NMP-22 [11
]. In our study, the influence of blood was confirmed, but we also showed that spiking with intact, benign or cancerous cells caused a positive NMP-22. As expected, cellular lysate from those cells had a greater effect.
NMP-22 (nuclear matrix protein-22) is also known as nuclear mitotic apparatus protein (NUMA or NUMA-1) [12
]. The function of NUMA-1 is to provide structural support for the nucleus and to ensure the correct separation of genetic material during mitosis into the respective daughter cells through mitotic spindle stabilization [14
]. To see how NUMA-1 is distributed in healthy and diseased tissues we queried the Human protein Atlas (http://www.proteinatlas.org
). The Atlas provides immunohistochemistry data, and expression is confirmed in the majority of cases using 2 or more antibodies for the same antigen [15
]. NUMA-1 expression data is available on a wide range of tissues, including both normal and diseased urinary bladder tissue. An overview of the staining patterns across tissues confirms that NUMA-1 is expressed in all tissues with an epithelial component, but also in glandular cells, cells of the lymph node and macrophages. Notably, it is apparent that the normal urothelial lining of the bladder expresses NUMA-1 at a high level (Figure ). As reported previously, NUMA-1 is localized to the nucleus primarily, but there is granular cytoplasmic staining of NUMA-1 protein also. Carcinoma tissues typically have NUMA-1 expression across the tumor, with some expected heterogeneity, but intensity of staining is often less than that seen in the normal urothelia (Figure ), and the cytoplasmic component is less pronounced. The imunohistochemistry data suggest that at the interface between urine and bladder lining, the cellular expression level will be at least as strong in normal tissue as in tumor tissue.
Figure 4 Immunostaining of NMP-22 (NUMA1) in normal urothelia and bladder cancer tissue. Images originate from the Human Protein Atlas web portal. Top panels show magnified insets of tissue microarray images directly below. A, normal urinary bladder tissue from (more ...)
We also queried public databases for transcriptome profile data. NUMA-1 expression is reported to be increased in tumors of a variety of tissues, but given the epithelial expression profile of NUMA-1 this can be misleading unless the details of how the cellular composition of solid tissue specimens was corrected for in specific studies are known. However, of note, the profile databases do report that NUMA-1 mRNA is high in whole blood. This is of interest because blood incursion into the bladder can create difficulties in interpretation of urine tests for bladder cancer. We showed in our experimental model that spiking of NMP-22 negative urine with whole blood can cause a false-positive result.
If the NMP-22 test is measuring a nuclear protein that is present in the majority of cell types throughout the body, and one that is particularly prevalent in the normal bladder lining, how does the test perform reasonably well for the detection of BCa. The test could be detecting NMP-22 that is introduced into the urine by hematuria, and since hematuria is the most associated symptom of bladder cancer, then the test would perform no better than a test for hemoglobin as long as the threshold and sensitivity of the two assays were comparable. Conversely, measuring NMP-22 introduced by bleeding would impact the sensitivity of the test, calling falsely positive cases that had hematuria through non-malignant causes such as infection or trauma. Alternatively, perhaps the NMP-22 test is a measure of the release and lysis of urothelial cells into the urine. Bladder tumors are often friable and shed large numbers of tumor cells into the urine relative to the normal turnover of healthy urothelia. The total number of urothelia shed per unit of time will be greater if the patient has a bladder lesion. NMP-22 is a nuclear scaffold protein and is not known to be secreted, thus, cell lysis must occur for the NMP-22 test to detect soluble protein. If cell lysis occurs at a constant rate, then the increased number of cells released into the urine in the presence of a lesion may suffice for a test, but it is also possible that tumor cells lyse in urine more readily than normal urothelia, thereby amplifying the test to some extent. A likely scenario is that a combination of these factors (more cells released and lysed, and the presence of blood) contribute to the NMP-22 test result. As long as the test outperforms VUC, one can argue that it has value, but identifying what any bladder cancer test is actually measuring would seem to be pertinent and should enable improvement of a given test, or provide insight into how best to combine multiple markers.
Other investigators have associated false-positive NMP-22 tests with instrumentation of the genitourinary tract and inflammation [16
] and grade of urinary hematuria [17
]. Specifically, Huber et al.
reported from a large prospective screening study that NMP-22 outcomes are affected by gross hematuria, urinary tract infections and concentrated urine (creatinine
]. Furthermore, impaired renal function (i.e., reduced glomerular filtration rate) has been linked to false-positive NMP-22 results, while proteinuria associated with impaired renal function is associated with reduction in urinary cytology specificity [19
]. Similar to other urine-based assays, NMP-22 has a disappointingly low sensitivity for small, low-grade tumors [20
]. Based on our results we believe the reduced sensitivity is related to low tumor burden and consequent low cell turnover. Perhaps if NMP-22 urinary protein was normalized to total urinary proteins, or to urinary creatinine, then a NMP-22 assay may be better able to diagnose these small, low-grade tumors, which are the most common, yet least threatening of the tumors.