(86 days)
The UroVysion Bladder Cancer Recurrence Kit (UroVysion Kit) is designed to detect aneuploidy for chromosomes 3, 7, 17, and loss of the 9p21 locus via fluorescence in situ hybridization (FISH) in urine specimens from subjects with transitional cell carcinoma of the bladder. Results from the UroVysion Kit are intended for use as a noninvasive method for monitoring for turnor recurrence in conjunction with cystoscopy in patients previously diagnosed with bladder cancer.
The UroVysion Kit is based upon fluorescence in situ hybridization (FISH) DNA probe technology. The UroVysion probes are fluorescently labeled nucleic acid probes for use in in situ hybridization assays on urine specimens fixed on slides. . The UroVysion Kit consists of a 4-color, four-probe mixture of DNA probe sequences homologous to specific regions on chromosomes 3, 7, 9, and 17. The UroVysion probe mixture consists of Chromosome Enumeration Probe (CEP®) 3 SpectrumRed™, CEP 7 SpectrumGreen™, CEP 17 SpectrumAqua™, and Locus Specific Identifier (LSI®) 9p21 SpectrumGold TM .
The UroVysion Bladder Cancer Recurrence Kit is intended for use as a noninvasive method for monitoring for tumor recurrence in conjunction with cystoscopy in patients previously diagnosed with bladder cancer. The study demonstrated that the device is substantially equivalent to the predicate device, the Bard® (Bion) BTAstat™ Test, and meets the acceptance criteria for specificity and performance compared to the standard of care (cystoscopy/histology).
Acceptance Criteria and Reported Device Performance
| Acceptance Criteria | Reported Device Performance |
|---|---|
| Specificity | Overall Specificity: 93.0% (332/357), with 95% CI not provided, but based on a multi-center, prospective study of healthy volunteers and urology patients without a history of bladder cancer. |
| Specificity for unique patients: 94.5% (260/275). | |
| Performance vs. Standard of Care | Overall Agreement with Cystoscopy/Histology: |
| Agreement of (+) results: 71.0% (95% CI = 58.1% - 81.8%) | |
| Agreement of (-) results: 65.8% (95% CI = 56.3% - 74.4%) | |
| Overall Agreement: 67.6% (95% CI = 60.2% - 74.5%) | |
| (+) Predictive Value: 53.0% (95% CI = 41.7% - 64.1%) | |
| (-) Predictive Value: 80.6% (95% CI = 71.1% - 88.1%) | |
| Performance in patients on BCG therapy within 3 months: | |
| Agreement of (+) results: 92.3% (95% CI = 64.0% - 99.8%) | |
| Agreement of (-) results: 61.5% (95% CI = 40.6% - 79.8%) | |
| Substantial Equivalence (vs. BTAstat) | The 95% CIs for UroVysion's Agreement (+), Agreement (-), and Overall Agreement were all greater than the BTAstat scores minus 15%. |
| UroVysion Agreement (+): 58.1% (lower 95% CI) vs. BTAstat - 15%: 35.0% | |
| UroVysion Agreement (-): 56.3% (lower 95% CI) vs. BTAstat - 15%: 54.3% | |
| UroVysion Overall Agreement: 60.2% (lower 95% CI) vs. BTAstat - 15%: 47.5% | |
| For patients on BCG therapy within 3 months: | |
| UroVysion Agreement (+): 64.0% (lower 95% CI) vs. BTAstat - 15%: 54.2% | |
| UroVysion Agreement (-): 40.6% (lower 95% CI) vs. BTAstat - 15%: 27.3% | |
| UroVysion Overall Agreement: 55.1% (lower 95% CI) vs. BTAstat - 15%: 36.3% | |
| Hybridization Efficiency | ≥87% in conditions simulating clinical practice (observed: 87% in clinical study, 92.7% in specificity study using patient urine). |
| Analytical Specificity | No cross-hybridization to other chromosome loci observed (limited to intended target regions). |
| Interference | No interference detected from any of the numerous substances tested at high concentrations. |
| Reproducibility | Mean number of signals for each probe varied within a narrow range with acceptable %CVs (e.g., for CEP 3: 6.79% in Specimen 1, 2.49% in Specimen 2). No false negative results in bladder carcinoma cell line study (48/48 classified positive). One false positive out of 16 normal specimen evaluations. Informative results in 95.0% (76/80) of specimens on the first attempt using cell lines. |
| Longitudinal Study | Statistical difference in recurrence rates (p=0.014) between FISH+/cysto:histo- group (41.67% recurrence) and FISH-/cysto:histo- group (19.12% recurrence). |
Study Information
-
Sample size used for the test set and the data provenance:
- Specificity Study: 309 usable office visits (resulting in 357 data points due to patients having multiple conditions) from healthy volunteers and urology patients without a prior history or clinical evidence of bladder cancer. This was a multi-center, prospective study. The country of origin is not explicitly stated but implies US clinical sites due to FDA submission.
- Performance vs. Standard of Care Study: 251 usable office visits (representing 176 unique patients) from patients with a history of bladder cancer. This was a multi-center, prospective, longitudinal study conducted at 21 investigation sites. The country of origin is not explicitly stated but implies US clinical sites.
- Reproducibility (Cell Lines): 80 specimens prepared from human bladder carcinoma cell lines.
- Interference Study: Three voided urine pools from normal healthy volunteers, spiked with 29 different substances.
- HYBrite/VP 2000 Validation: Three human urine pools from normal donors.
-
Number of experts used to establish the ground truth for the test set and the qualifications of those experts:
- The document implies that the "standard of care" (cystoscopy with histology confirmation) was the ground truth. While no specific number of experts is given for the histological analysis, it is implicitly performed by qualified pathologists, which is standard practice for histology confirmation. No specific qualifications (e.g., 10 years of experience) are provided for these pathologists.
-
Adjudication method for the test set:
- For the "Performance vs. Standard of Care" study, the comparative reference for all percent agreement calculations was cystoscopy with histology confirmation for positive or suspicious cystoscopies.
- If a patient had a positive cystoscopy but histology was absent (e.g., lesion fulgurated), the specimen was considered positive for bladder cancer.
- If a test had a suspicious cystoscopy but histology was absent, the case was omitted from analysis.
- This indicates a hierarchical adjudication or ground truth definition rather than a consensus among multiple readers of the test device's results.
-
If a multi-reader multi-case (MRMC) comparative effectiveness study was done, If so, what was the effect size of how much human readers improve with AI vs without AI assistance:
- No MRMC comparative effectiveness study involving human readers and AI assistance was conducted. The device is a diagnostic kit (FISH probes), and its interpretation relies on visual recognition by an analyst. The studies compare the UroVysion kit's performance against the standard of care (cystoscopy/histology) and the predicate device (BTAstat), and there is no mention of AI.
-
If a standalone (i.e. algorithm only without human-in-the-loop performance) was done:
- The UroVysion Kit requires visual analysis by an analyst to recognize fluorescent signals on chromosomes. Therefore, this is not a standalone algorithm-only device. The "analyst visually recognizes chromosomes" (page 1) indicating human-in-the-loop performance.
-
The type of ground truth used (expert consensus, pathology, outcomes data, etc.):
- Pathology/Outcomes Data: The primary ground truth for the "Performance vs. Standard of Care" study was cystoscopy with histology confirmation. For cases where biopsy was not performed but cystoscopy was positive, it was still considered positive. The longitudinal study also used recurrence confirmed by cystoscopy/histology as an outcome.
-
The sample size for the training set:
- The document does not explicitly describe a "training set" in the context of machine learning. The studies described are for validation/testing of the UroVysion Kit itself. The device is a FISH-based diagnostic kit, not an AI/ML algorithm that undergoes a distinct training phase with a labeled dataset. Its "training" would align more with its development and optimization, rather than a quantifiable dataset used for algorithm training.
-
How the ground truth for the training set was established:
- As there is no explicit "training set" for an AI/ML algorithm described, this question is not directly applicable. The performance of the UroVysion Kit itself (hybridization efficiency, specificity, etc.) was established through laboratory tests and clinical studies, where outcomes like histology served as the reference standard.
§ 866.6010 Tumor-associated antigen immunological test system.
(a)
Identification. A tumor-associated antigen immunological test system is a device that consists of reagents used to qualitatively or quantitatively measure, by immunochemical techniques, tumor-associated antigens in serum, plasma, urine, or other body fluids. This device is intended as an aid in monitoring patients for disease progress or response to therapy or for the detection of recurrent or residual disease.(b)
Classification. Class II (special controls). Tumor markers must comply with the following special controls: (1) A guidance document entitled “Guidance Document for the Submission of Tumor Associated Antigen Premarket Notifications (510(k)s) to FDA,” and (2) voluntary assay performance standards issued by the National Committee on Clinical Laboratory Standards.