11 studies were selected for BTA stat and 11 studies for the test NMP 22 were chosen. Tables 1 and 2 give an overview of number of patients, BC history/ no BC history, sensitivity, specificity and country of performance.
Table 1
Studies performed with the BTA stat
Study & Publication year
|
n=
|
No BC history
|
Sensitivity
|
Specificity
|
PPV
|
NPV
|
Country
|
Toma et al. 2004[3]
|
126
|
78
|
66.6%
|
78.2%
|
71.4%
|
75.4%
|
Germany
|
Miyake et al. 2012 [4]
|
126
|
62
|
72%
|
53%
|
NA
|
NA
|
USA
|
Raitanen et al. 2008 [8]
|
490
|
0
|
56.6%
|
76.4%
|
NA
|
NA
|
Finland
|
Poulakis et al. 2001 [9]
|
739
|
353
|
70%
|
92%
|
13%
|
88%
|
USA
|
García-Velandria et al. 2014 [10]
|
237
|
31
|
63.2%
|
82.9%
|
NA
|
92.2%
|
Spain
|
Raitanen et al. 1999[13]
|
100
|
100
|
100%
|
98%
|
2%
|
98%
|
USA
|
Nasuti et al. 1998 [14]
|
100
|
100
|
100%
|
84%
|
16%
|
84%
|
USA
|
Wiener et al. 1998 [15]
|
291
|
190
|
57%
|
68%
|
45%
|
78%
|
Austria
|
Giannopoulos et al. 2000 [16]
|
147
|
101
|
71.7%
|
56.5%
|
70.3%
|
58.2%
|
Greece
|
Boman et al. 2002 [17]
|
289
|
66
|
75%
|
52%
|
62%
|
58%
|
Sweden
|
Sun et al. 2006 [18]
|
251
|
100
|
76.8%
|
87.0%
|
89.9%
|
71.3%
|
China
|
Table 2
Studies performed with the NMP 22
Study & Publication year
|
n=
|
No BC history
|
Sensitivity
|
Specificity
|
PPV
|
NPV
|
Country
|
Recommend U/mL
|
Toma et al. 2004[3]
|
126
|
78
|
68.5%
|
65.2%
|
53.6%
|
77.9%
|
Germany
|
|
Poulakis et al. 2001 [9]
|
739
|
353
|
85%
|
94%
|
18%
|
91%
|
USA
|
8.25
|
Wiener et al. 1998 [15]
|
291
|
190
|
48%
|
69%
|
41%
|
74%
|
Austria
|
10
|
Giannopoulos et al. 2000 [16]
|
147
|
101
|
62.6%
|
73.9%
|
72.9%
|
61.4%
|
Greece
|
8
|
Boman et al.2002 [17]
|
299
|
66
|
65%
|
75%
|
NA
|
NA
|
Sweden
|
4
|
Sun et al.2006 [18]
|
251
|
100
|
77.5%
|
81.0%
|
86.0%
|
64.8%
|
China
|
10
|
Stampfer et al. 1996 [20]
|
231
|
0
|
31%
|
NA
|
NA
|
NA
|
USA
|
6.4
|
Mian et al. 2000 [21]
|
240
|
81
|
55.5%
|
79%
|
45%
|
86%
|
Austria
|
10
|
Todenhöfer et al. 2012 [22]
|
1386
|
1386
|
79.2%
|
34.5%
|
86.5%
|
76.1%
|
Germany
|
10
|
Lee et al. 2000 [23]
|
137
|
101
|
69%
|
72%
|
NA
|
97%
|
Taiwan
|
3.75–18.95
|
Bangma et al. 201 [30]
|
1747
|
1747
|
25%
|
96.6%
|
7.1%
|
99.2%
|
The Netherlands
|
10
|
NA: not answered |
What are some problems associated with the use of BTA stat?
Most studies criticized the correlation between hematuria, the leading symptom of BC, and BTA stat. In a cohort of 126 patients, including 64 patients with BC, the BTA showed a sensitivity of 72% and a specificity of 53%. The association between hemoglobin and BTA was 0.73 [4]. Oge et al. demonstrated that in a healthy group of 25 volunteers, an increased number of false-positive rates were observed according to dilution. A specificity of 24% and a false-positive rate of 76% in a dilution of 1/200 was observed [5].
What can we summarize for BTA stat from a cohort with a history of or suspicion of BC?
The latest review by Budman et al. compared the largest studies and gave a clear disapproval for the use of BTA stat or NMP 22 in either the detection or surveillance of BC. Instead, they recommended keeping cytology and cystoscopy as the gold standard [6]. However, a meta-analysis of 3,462 patients published by Guo et al. found that the sensitivity and specificity of BTA stat in comparison to urine cytology in patients with BC history showed a higher sensitivity (0.67 [95% confidence interval (CI): 0.64–0.69] vs 0.43 [95% CI: 0.40–0.46], respectively) [7]. In addition, a study by Raitanen et al. showed that the percentage of patients (n = 445) with an apparent false-positive BTA stat showed a recurrence of BC in cystoscopy [8]. The study with the highest number of patients suspected of BC (n = 739) was published by Poulakis et al., and a follow-up at 27 months showed 406 patients with BC. Looking closely at the results, the BTA stat resulted in a sensitivity of 90% for histological grade 3, but only a sensitivity of 68% for histological grade 2 and 38% for histological grade 1. The specificity of BTA stat was 67%. The specificity varied according to patients with and without an UTI. A specificity of 92% with no UTI was found versus a specificity of 52% with UTI diagnosis. False-positive results did not correlate with future recurrences at follow-up [9]. A study by García-Velandria et al. compared 237 patients (87% of which were surveillance cases) using cytology and BTA stat. Interestingly, cytology and BTA stat achieved almost the same results: a sensitivity of 57.9% and 63.2%, respectively and a specificity of 84.4% and 82.9%, respectively. Therefore, this study showed a higher sensitivity of 5,3 % in BTA stat than cytology. Also the negative predicted value (NPV) in the surveillance patients with low-grade tumors was equal: cytology had an NPV of 95.7% and BTA stat of 95.0%. Their recommendation was to use BTA stat instead of cystoscopy in surveillance cases with previous low-grade tumors for cost reasons [10].
The study by Raitanen et al. and the study by van der Poel et al., gave no recommendation for BTA stat use in place of cystoscopy for diagnostic purposes. However, those studies did suggest replacing cytology with BTA stat in low-grade diseases for patients with no UTI and BC history [11, 12].
The sensitivity of BTA stat ranged from 57–100%, and the specificity ranged from 52–98% (Table 1).
What are the results for a general screening in a healthy population?
A study by Raitanen et al. investigated 100 healthy subjects, 87% of which were female. The results showed a specificity of 98% in a general screening assessment; no other study was identified with healthy subjects, a comparable number of patients, and a similarly high specificity. However, no follow-up was undertaken and the study did not use an RCT design, which would have strengthened the high-specificity finding [13]. In comparison, the study by Nasuti et al. has to be taken in consideration. One hundred healthy subjects with no BC history, but who showed symptoms of dysuria, incontinence, and gross hematuria or microhematuria, were tested using BTA stat. The sensitivity, specificity, and positive predicted value (PPV) were 100%, 84%, and 16%, respectively. The authors did not give a recommendation for the use of BTA stat [14].
What problems are associated with the use of NMP 22?
Most of the studies that we reviewed criticized the correlation between hematuria and inflammation, the leading symptoms of BC, and NMP 22 tests. Todenhöfer et al. described a false-positive rate for NMP 22 of 85.3% in patients with inflammation compared to a rate of 61.4% when cytology, immunocytology, and Fluorescent in situ hybridization (FISH) were used [22]. Lee et al. established different cut-off values in NMP 22 for patients with hematuria. Patients with hematuria received a cut-off value of 6.39 U/ml, while patients with no hematuria got a cut-off value of 18.95 U/ml. Patients with confirmed urothelial carcinoma and a cut-off level > 6 U/ml showed a sensitivity of 81% and the specificity of 50%. Comparing the sensitivity and the specificity now to the same urothelial carcinoma patients with a cut-off level > 10 U/ml the sensitivity and the specificity changed to 69% and 72% [23].
What can we summarize for NMP22 from a cohort with a history of BC or suspicion of BC?
As already discussed in the BTA stat section, the review by Budman et al. and Toma et al. gave no recommendation for NMP 22 testing in terms of BC detection or surveillance, and instead recommended the continued use of cytology and cystoscopy as the gold standard [3, 6]. Looking closely at histological grades of transitional cell carcinoma Poulakis et al. showed a sensitivity of 82%, 89%, and 94% for the use of NMP22 for grades 1–3, respectively, with an optimal value of 8.25 U/ml, compared to a sensitivity of 53%, 76%, and 90% for the use of BTA stat for grades 1–3, respectively. NMP 22 achieved even higher results than voided urine cytology (VUC; the examination of cells under a microscope), which showed a sensitivity of 38%, 68%, and 90% for grades 1–3, respectively. However, VUC presented a specificity of 96%, compared to 68% for NMP22 and 67% for BTA stat. Nevertheless, the study showed a significant change in specificity from 94% (NMP22) and 92% (BTA stat) in patients with no apparent genitourinary disease to a specificity of 52% (in both tests) for patients with chronic UTI. To summarize, NMP 22 presented better results than VUC for the detection of superficial and low-grade BC after the exclusion of risk factors such as chronic UTIs, and the overall specificity also increased [9].
In a German study by Todenhöfer et al., NMP 22 was used on a population of 1,386 patients with possible BC. The specificity of NMP 22 was only 34.5%, with a cut-off value of 10 U/ml, so no recommendation was given for NMP 22 [22]. Lotan et al. developed an NMP22-based algorithm for BC in an asymptomatic but high-risk population. Risk was defined by smoking history, age, presence of hematuria, occupation, race, and gender. The general PPV was 20.3% and the NPV was 96.9%, but the PPV raised to 24% when only the male group was considered compared to 13% in the female group. Also a change in the PPV of more than 5% was observed when the group was divided into males under 65 (16.8%) and over 65 (23.5%). However, it was reduced to 77% in the male-only group when gross hematuria and smoking history were included. The NPV in the over 65-year old female group was 100% [25–27].
What are the results for mass screening in a healthy population?
The largest general screening study was performed by Bangma et al. in the Netherlands. In that study, 1,747 male patients were screened at home with NMP 22, in addition to other molecular-marker tests. In the study population, 23.4% were positive for hematuria (mainly smokers), 17% were current smokers, 58% were past smokers, and 36% had been exposed to occupational risk factors. Seventy-one of 75 men with positive molecular markers underwent cystoscopy. Four BCs and one kidney tumor was found, but in total one BC as well as one kidney tumor were missed. However, it reduced in total number of unneeded cystoscopies. The authors mentioned the possibility of a healthy-subject bias, and they did not give a recommendation for NMP 22 to be used for general screening [28].
Our systematic review of the literature found that the sensitivity of NMP 22 ranged from 25–85% and the specificity ranged from 34.5–96.6% (Table 2).