Search Results

The Xpert® MTB/RIF Assay, performed on the GeneXpert® Instrument Systems, is a qualitative, nested realtime polymerase chain reaction (PCR) in vitro diagnostic test for the detection of Mycobacterium tuberculosis complex DNA in raw sputum or concentrated sputum sediment prepared from induced or expectorated sputum. In specimens where Mycobacterium tuberculosis complex) is detected, the Xpert MTB/RIF Assay also detects the rifampin-resistance associated mutations of the rpoB gene.

The Xpert MTB/RIF Assay is intended for use with specimens for whom there is clinical suspicion of tuberculosis (TB) and who have received no antituberculosis therapy, or less than three days of therapy. This test is intended as an aid in the diagnosis of pulmonary tuberculosis when used in conjunction with clinical and other laboratory findings.

An Xpert MTB/RIF Assay result of "MTB NOT DETECTED" from either one or two sputum specimens is highly predictive of the absence of M. tuberculosis complex bacilli on serial fluorescent acid-fast sputum smears from patients with suspected active pulmonary tuberculosis and can be used as an aid in the decision of whether continued airborne infection isolation (AII) is warranted in patients with suspected pulmonary tuberculosis. The determination of whether testing of either one or two sputum specimens for decisions regarding removal from AII should be based on specific clinical circumstances and institutional guidelines. Clinical decisions regarding the need for continued AII should always occur in conjunction with other clinical and laboratory evaluations and Xpert MTB/RIF Assay results should not be the sole basis for infection control practices.

The Xpert MTB/RIF Assay must always be used in conjunction with mycobacterial culture to address the risk of false negative results and to recover organisms when MTB-complex is present for further characterization and drug susceptibility testing. However, decisions regarding the removal of patients from AII need not wait for culture results. Sputum specimens for TB culture, AFB smear microscopy, and Xpert MTB/RIF Assay testing should follow CDC recommendations with regard to collection methods and time frame between specimen collection.

The Xpert MTB/RIF Assay does not provide confirmation of rifampin susceptibility since mechanisms of rifampin resistance other than those detected by this device may exist that may be associated with a lack of clinical response to treatment.

Specimens that have both MTB-complex DNA and rifampin-resistance associated mutations of the rpoB gene detected by the Xpert MTB/RIF Assay must have results confirmed by a reference laboratory. If the presence of rifampin-resistance associated mutations of the rpoB gene is confirmed, specimens should also be tested for the presence of genetic mutations associated with resistance to other drugs.

The Xpert MTB/RIF Assay should only be performed in laboratories that follow safety practices in accordance with the CDC/NIH Biosafety in Microbiological and Biomedical Laboratories publication and applicable state or local regulations.

Device Description

The Xpert MTB/RIF Assay is a qualitative, automated, in vitro diagnostic test for qualitative detection of Mycobacterium tuberculosis (MTB) complex DNA in raw sputum samples or in concentrated sputum sediments prepared from induced or expectorated sputa that are either acid-fast bacilli (AFB) smear positive or negative. The assay is performed on the Cepheid GeneXpert Instrument Systems. The Xpert MTB/RIF Assay on the GeneXpert Instrument System automates and integrates sample preparation, nucleic acid amplification, and detection of the target sequences in simple or complex samples using real-time PCR. The system consists of an instrument, personal computer, and preloaded software for running the tests and viewing the results. The system requires the use of single-use disposable cartridges that hold the PCR reagents and host the PCR process. Because the cartridges are self-contained, cross-contamination between samples is minimized.

The Xpert MTB/RIF Assay includes reagents for the detection of MTB and Rifampin (RIF) resistance from raw sputum samples and in prepared sputum sediments. A Sample Processing Control (SPC) and a Probe Check Control (PCC) are also included in the cartridge. The SPC is present to control for adequate processing of the target bacteria and to monitor for the presence of inhibitors in the PCR reaction. The Probe Check Control (PCC) verifies reagent rehydration, PCR tube filling in the cartridge, probe integrity, and dve stability.

The GeneXpert Instrument Systems, comprised of the GeneXpert Dx Systems, the GeneXpert Infinity-48 System, the GeneXpert Infinity-48s, and the GeneXpert Infinity-80 System, have 1 to 80 randomly accessible modules, depending upon the instrument, that are each capable of performing separate sample preparation and real-time PCR tests. Each module contains a syringe drive for dispensing fluids (i.e., the syringe drive activates the plunger that works in concert with the rotary valve in the cartridge to move fluids between chambers), an ultrasonic horn for lysing cells or spores, and a proprietary I-CORE® thermocycler for performing real-time PCR and detection.

AI/ML Overview

The provided document describes the Xpert® MTB/RIF Assay, an in vitro diagnostic test for the detection of Mycobacterium tuberculosis complex DNA and rifampin-resistance associated mutations. A clinical performance study (Study 2) was conducted to evaluate the device's effectiveness.

Here's a breakdown of the acceptance criteria and study information:

1. A table of acceptance criteria and the reported device performance

The document does not explicitly state pre-defined acceptance criteria in terms of specific performance metrics (e.g., "Sensitivity must be >= X%"). Instead, it presents the device's performance in a multi-center study and implicitly suggests that these results fulfill the requirements for substantial equivalence.

Based on the provided tables, we can extrapolate the reported device performance metrics:

Metric	Acceptance Criteria (Implied)	Reported Device Performance (Xpert MTB/RIF Assay)
For Smear Positive Cases:	(Not explicitly stated, but high sensitivity/specificity desirable)	Sensitivity: 98.5% (129/131); 95% CI: 94.6%, 99.6%
		Specificity: 94.4% (17/18); 95% CI: 74.2%, 99.0%
For Smear Negative Cases:	(Not explicitly stated, but good sensitivity/specificity desirable)	Sensitivity: 54.8% (46/84); 95% CI: 44.1%, 65.0%
		Specificity: 98.8% (718/727); 95% CI: 97.7%, 99.4%
Overall (One Xpert MTB/RIF Assay vs. MTB Culture):	(Not explicitly stated, but good sensitivity/specificity desirable)	Sensitivity: 81.4% (95% CI: 75.7%, 86.0%)
		Specificity: 98.7% (95% CI: 97.5%, 99.3%)
Negative Predictive Value (NPV) for Single Test (U.S. Subjects, for AII removal):	(High NPV for ruling out AFB smear-positive TB for AII removal)	NPV: 97.6% (95% CI: 95.9%, 98.6%)
Difference in NPV (Xpert vs. Two AFB Smears, U.S. Subjects):	(Improvement over AFB smears indicated)	Difference in NPV: 2.6% (95% CI: 1.2%, 4.2%), indicating better performance for Xpert for U.S. subjects compared to two AFB smears

2. Sample size used for the test set and the data provenance

Sample Size for Clinical Performance Study (Study 2): 960 subjects.
Data Provenance: The study was a prospective multi-center study conducted at multiple sites in the United States, as well as South Africa and Brazil.

3. Number of experts used to establish the ground truth for the test set and the qualifications of those experts

The document does not specify the number of experts or their qualifications for establishing the ground truth. It states that the ground truth for MTB detection was based on mycobacterial culture (stated in Table 5.5). For rifampin resistance, it is stated that "Specimens that have both MTB-complex DNA and rifampin-resistance associated mutations of the rpoB gene detected by the Xpert MTB/RIF Assay must have results confirmed by a reference laboratory."

4. Adjudication method for the test set

The document does not describe a formal adjudication method (like 2+1 or 3+1) for resolving discrepancies in the ground truth. The ground truth for MTB detection was based on mycobacterial culture, an objective laboratory method. For rifampin resistance, a reference laboratory confirmation is required for positive results.

5. 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

This is not a MRMC study involving human readers or AI assistance in the traditional sense of image interpretation. The device is an automated, in vitro diagnostic test. However, the study does compare the performance of the Xpert MTB/RIF Assay (the device) against AFB smear microscopy which involves human interpretation (laboratory technicians reading smears).

Comparison between device and AFB smears: The document notes: "One Xpert MTB/RIF Assay was associated with a sensitivity of 81.4% (95% CI: 75.7%, 86.0%) for identifying MTB culture-positive subjects compared to a sensitivity of 60.9% (95% CI: 54.3%, 67.2%) for two AFB smears."
Effect Size (implicitly an improvement over human-read smears): For U.S. subjects, the NPV for one Xpert MTB/RIF Assay result was 97.6% compared to 95.0% for two AFB smears results. The difference in NPVs was 2.6% (95% CI: 1.2%, 4.2%). This suggests that the device performs better than two AFB smears in ruling out active pulmonary TB for decisions on airborne infection isolation in the U.S. context.

6. If a standalone (i.e. algorithm only without human-in-the loop performance) was done

Yes, the primary clinical performance study (Study 2) evaluated the standalone performance of the Xpert MTB/RIF Assay. The assay is an automated in vitro diagnostic test that processes the sample and provides a result without direct human interpretation of the PCR data. Human involvement is primarily in sample collection, preparation, loading the cartridge, and interpreting the output from the instrument system.

7. The type of ground truth used

The primary ground truth for MTB detection was mycobacterial culture. For cases where MTB complex DNA and rifampin-resistance associated mutations were detected, confirmation by a reference laboratory served as the ground truth for rifampin resistance.

8. The sample size for the training set

The document does not explicitly mention the size of a "training set" for the device's development. This is typical for a diagnostic assay where algorithms are often based on well-established molecular biology principles and analytical validation rather than machine learning on a large training set in the same way an imaging AI algorithm would. Previous FDA-cleared 510(k) K131706 references are made for analytical sensitivity, reactivity, specificity, interfering substances, carry-over contamination, RIF resistance study, reproducibility, and instrument system precision. These earlier studies would have involved extensive analytical testing to establish the device's performance characteristics.

9. How the ground truth for the training set was established

As there isn't a clearly defined "training set" in the context of machine learning, the establishment of ground truth would refer to the methods used during the development and analytical validation of the assay. This would involve:

Known positive and negative samples: Using characterized bacterial strains (e.g., M. tuberculosis complex, non-tuberculous mycobacteria) and samples with known resistance profiles.
Sequencing and other molecular methods: To confirm the presence of M. tuberculosis DNA and specific rpoB gene mutations associated with rifampin resistance.
Culture results: As the gold standard for viability and identification of M. tuberculosis.

Ask a Question

Ask a specific question about this device

K Number

DEN130032

Device Name

XPERT MTB/RIF ASSAY

Manufacturer

CEPHEID

Date Cleared

2013-07-25

(44 days)

Product Code

Regulation Number

Type

Panel

Reference & Predicate Devices

N/A

Predicate For

N/A

Intended Use

The Xpert® MTB/RIF Assay, performed on the GeneXpert® Instrument Systems, is a qualitative, nested real-time polymerase chain reaction (PCR) in vitro diagnostic test for the detection of Mycobacterium tuberculosis complex DNA in raw sputum or concentrated sediments prepared from induced or expectorated sputum. In specimens where Mycobacterium tuberculosis complex (MTB-complex) is detected, the Xpert MTB/RIF Assay also detects the rifampin-resistance associated mutations of the rpoB gene.

The Xpert MTB/RIF Assay is intended for use with specimens from patients for whom there is clinical suspicion of tuberculosis (TB) and who have received no antituberculosis therapy, or less than 3 days of therapy. This test is intended as an aid in the diagnosis of pulmonary tuberculosis when used in conjunction with clinical and other laboratory findings.

The Xpert MTB/RIF Assay must be used in conjunction with mycobacterial culture to address the risk of false negative results and to recover the organisms for further characterization and drug susceptibility testing.

Device Description

The Xpert® MTB/RIF Assay is an automated in vitro diagnostic test for the qualitative detection of MTB-complex DNA and the genetic mutations associated with rifampin (Rif) resistance in raw sputum samples or concentrated sputum sediments from patients for whom there is clinical suspicion of TB and who have received no antituberculosis therapy, or less than 3 days of therapy. The primers in this test amplify a portion of the rpoB gene containing the 81 base pair core region. The probes are designed to differentiate between the conserved wild-type sequence and mutations in the core region that are associated with Rif resistance. The assay is performed on Cepheid GeneXpert® Instrument Systems.

The Xpert® MTB/RIF Assay includes single-use disposable cartridges and sample reagent for sample preparation. The Xpert® MTB/RIF Assay cartridges contain reagents for the detection of MTB-complex DNA and Rif resistance associated mutations. A Sample Processing Control (SPC) and a Probe Check Control (PCC) are also included in the cartridge. The SPC is present to control for adequate processing of the target microorganism and to monitor the presence of inhibitors in the PCR reaction. The Probe Check Control (PCC) verifies reagent rehydration, PCR tube filling in the cartridge, probe integrity and dye stability.

Sputum specimens are collected according to the institution's standard procedures and transported to the GeneXpert® Instrument System area. For raw sputum, Sample Reagent is added to the sample (2:1, v:v). Sample Reagent is added to the resuspended sputum sediment (1.5 mL Sample Reagent to 0.5 mL suspension or 3:1, v:v, for larger volumes of sediment suspension). For both specimen types, the solution is shaken vigorously to mix, and then incubated at 20-30℃ for 15 minutes. Using the transfer pipette provided, the specimen is transferred to the open port of the Xpert® MTB/RIF Assay cartridge.

The user initiates a test from the system user interface, the Xpert® MTB/RIF Assay cartridge is loaded onto the GeneXpert® Instrument System platform, which performs hands-off, automated sample processing, and real-time PCR for detection of DNA, Summary and detailed test results are obtained in approximately 2 hours and are displayed in tabular and graphic formats.

The Xpert® MTB/RIF Assay simultaneously detects MTB-complex and the genetic mutations associated with rifampin resistance by amplifying a MTB-complex specific sequence of the rpoB gene, which is probed with five molecular beacons (Probes A - E) for mutations within the rifampin-resistance determining region (RRDR). Each molecular beacon is labeled with a different fluorophore.

The valid maximum cycle threshold (Ct) of 39.0 for Probes A, B and C and 36.0 for Probes D and E are set for data analysis.

"MTB DETECTED", is reported when at least two probes result in Ct values within the valid range and a delta Ct min (the smallest Ct difference between any pair of probes) of less than 2.0.
"Rif Resistance NOT DETECTED" is reported if the delta Ct max (the Ct difference between the earliest and latest probe) is ≤4.0.
"Rif Resistance DETECTED" is reported if the delta Ct max is >4.0.
"Rif Resistance INDETERMINATE" is reported when the following two conditions are met:
- 1. the Ct value of any probe exceeds the valid maximum Ct (or is zero, i.e. no threshold crossing); and
- 1. the earliest rpoB Ct value is greater than [(Valid maximum Ct of probe in condition1) - (delta Ct max cut-off of 4.0)]
"MTB NOT DETECTED" is reported when there is only one or no positive probe.

All assay settings are included as automatic calculations in the Xpert® MTB/RIF Assay protocol and cannot be modified by the user.

AI/ML Overview

Here's an analysis of the acceptance criteria and the study proving the device meets them, based on the provided text:

Acceptance Criteria and Device Performance for Xpert® MTB/RIF Assay

1. Table of Acceptance Criteria and Reported Device Performance

The acceptance criteria are generally demonstrated through the sensitivity and specificity values against a reference method (culture and DST).

Acceptance Criteria Category	Specific Criteria (Derived from text)	Reported Device Performance
MTB-Complex Detection (vs. Culture)
- Overall Sensitivity	High sensitivity and specificity for MTB-complex detection (implicitly required by regulatory guidelines for diagnostic tests aid in diagnosis).	Overall Sensitivity (US prospective study): 93.8%
- Overall Specificity		Overall Specificity: 98.7% (95% CI: 97.5% - 99.4%)
- Sensitivity (AFB Smear-Positive)	High sensitivity for smear-positive specimens.	99.7% (350/351) with 95% CI: 98.4% - 99.9% (Table 19)
- Specificity (AFB Smear-Positive)	High specificity for smear-positive specimens.	98.5% (65/66) with 95% CI: 91.9% – 99.7% (Table 19)
- Sensitivity (AFB Smear-Negative)	Acceptable sensitivity for smear-negative specimens.	76.1% (89/117) with 95% CI: 67.6% - 82.9% (Table 20)
- Specificity (AFB Smear-Negative)	High specificity for smear-negative specimens.	98.8% (555/562) with 95% CI: 97.5% – 99.4% (Table 20)
Rifampin Resistance Detection (vs. DST)
- Sensitivity (Rif Res vs. DST)	High sensitivity for detecting rifampin resistance.	94.7% (18/19) with 95% CI: 75.4% - 99.1% (Table 26)
- Specificity (Rif Susc vs. DST)	High specificity for detecting rifampin susceptibility.	99.0% (404/408) with 95% CI: 97.5% - 99.6% (Table 26)
LoD for MTB-Complex	LoD must be determined in the most challenging matrix (e.g., sputum) claimed for use with the device, using both antibiotic susceptible and resistant strains. This is a general regulatory requirement (Q. Risks to Health and Mitigations, P. Proposed Labeling, S. Conclusion).	Demonstrated via probit analysis (Table 8) with claimed LoD values for different M.bovis BCG and M. tuberculosis strains in sputum and sediment, ranging from 525 to 4000 CFU/mL.
Reproducibility (Overall Agreement %)	Reproducibility studies must include at least one antibiotic resistant and one antibiotic susceptible strain of Mycobacterium tuberculosis complex (Q. Risks to Health and Mitigations, P. Proposed Labeling, S. Conclusion). Implicitly looking for high percentage agreement.	External Reproducibility (Table 1):
Cross-Reactivity (Analytical Exclusivity)	Absence of cross-reactivity with common respiratory pathogens and non-tuberculosis mycobacteria. Identification of any potential for cross-reactivity through in silico analysis. Implicitly required to prevent false positives.	Wet-tested microorganisms: No cross-reactivity observed with 131/132 microorganisms. One replicate of M. scrofulaceum showed cross-reactivity at 10^0 CFU/mL but not at 10^7 CFU/mL (presumably a false positive at lower concentration). (Table 14) In silico analysis: 13/18 predicted non-cross-reactive. 5/18 (M. kumamontonense, M. leprae, M. mucogenicum, Tsukamurella spp., Nocardia otitidiscaviarum) showed slight potential for cross-reactivity (Table 15, 16).
Analytical Reactivity (Inclusivity)	Ability of the device to detect genetic mutations associated with antibiotic resistance in a diversity of Mycobacterium tuberculosis complex strains; isolates used must be well characterized. (Q. Risks to Health and Mitigations, P. Proposed Labeling, S. Conclusion).	Wet-tested strains: 62 well-characterized M. tuberculosis strains (26 Rif-S, 36 Rif-R) tested at near LoD. Rif-S detection: 87% (67/77) accurate. 3 DST-susceptible strains showed Rif-resistance mutations by sequence analysis and Xpert. 1 replicate of one Rif-S strain showed "INDETERMINATE." Rif-R detection: 100% (107/107) accurate. (Table 11) In silico analysis: High likelihood of amplification and detection for M. africanum, M. bovis, M. canettii, M. caprae, and M. microti based on sequence alignment (Table 13).
Interfering Substances	Performance should not be significantly impacted by common interfering substances found in sputum. Implicitly required.	Inhibition observed for Lidocaine (30%), mucin (5% and 2.5%), Ethambutol (50, 25, 10 ug/mL), Guaifenesin (5 mg/mL), Phenylephrine (100% and 50%), and tea tree oil (0.5% to 0.015%), leading to false negative or indeterminate results (Table 17).
Carry-over Contamination	No significant carry-over/cross-contamination in single-use cartridges. Implicitly required for device safety and effectiveness.	100% correct results for both high positive samples (20/20 "MTB DETECTED; Rif Resistance NOT DETECTED") and negative samples (22/22 "MTB NOT DETECTED") when negative samples followed high positive samples. No carry-over or cross-contamination observed.
Internal Controls	Device must include internal controls (Sample Processing Control, Probe Check Control) to ensure adequate processing, monitor inhibition, and verify reagent/probe integrity. (Q. Risks to Health and Mitigations, P. Proposed Labeling, S. Conclusion).	SPC: Present in each cartridge, verifies adequate processing and monitors inhibitors. Passes if valid Ct in negative sample. PCC: Verifies reagent rehydration, PCR tube filling, probe integrity, and dye stability. Passes if fluorescence meets acceptance criteria.
External Controls	Device must include an external positive assay control (and implicitly a negative control) as appropriate. (Q. Risks to Health and Mitigations, P. Proposed Labeling, S. Conclusion).	Commercially available INTROL TB Controls evaluated; showed 100% expected results for negative, wild type, and two mutant controls (Tables 4-7). During clinical trial, 97% (453/467) of external controls gave expected results on first attempt, with 8/14 failed controls giving expected results on retest.
Specimen Stability	Claims for transport and storage of sputum and sputum sediment must be supported. Implicitly required.	Sputum: Storage at 3-5°C for up to 3 days, then 2-8°C for additional 7 days supported. Sputum Sediment: Storage at 2-8°C for up to 7 days supported.
Cartridge Hold Time	Claims for cartridge hold time (from adding sample to start of processing) must be supported. Implicitly required.	Sputum & Sputum Sediment: 4-hour hold time supported.
Shelf-Life	Product shelf-life must be demonstrated. Implicitly required.	18 months at 2-28°C (unopened). 6 weeks at 2-45°C (open package).
Failure Mode Testing	Evaluation of effect of operator errors, manufacturing errors, and instrument malfunction. Implicitly required to ensure device robustness.	Operator Errors: Reduced sample volumes (70-90%) and "no sample added" all resulted in correct positive/negative results, except for "no sample" resulting in negative. (Table 29) Manufacturing Errors: Most simulated errors (missing reagents, improper filter loading) resulted in "Error" or "Invalid" results, indicating the system detects internal failures and prevents erroneous results. Some (e.g., missing pre-filter) did not impede correct results. (Table 30)

2. Sample Sizes Used for the Test Set and Data Provenance

Clinical Study (MTB-Complex & Rifampin Resistance):
- Eligible Subjects: 1,126
- Analyzed Subjects (MTB-Complex Detection): 1,096 (after excluding non-determinate Xpert results and culture contamination)
- Analyzed Subjects (Rifampin Resistance Detection): 1,082 (after further exclusions for indeterminate Rif resistance and missing DST results)
- Data Provenance: Multi-center study, includes both US (49% from California, New York, Florida) and non-US (51% from Vietnam, Peru, South Africa, Mexico, Bangladesh) populations. Data included both prospectively collected specimens (450 US, 23 non-US) and archived specimens (92 US, 531 non-US).
Reproducibility Studies:
- External Reproducibility (Study 154A): 5 contrived panel samples (1 negative, 4 positive at low/moderate concentrations) x 3 sites x 2 operators x 3 panel runs/day x 5 days = 90 tests per sample type over 5 types of samples (450 total tests).
- Instrument Precision: 5 contrived panel samples (1 negative, 4 positive at low/moderate concentrations) x 2 instruments x 2 operators x 4 panel runs/day x 12 days = 192 tests per sample type over 5 types of samples (960 total tests). One sample type (MTB/Rif-R ~2-3X LoD) was 190 total tests.
Rifampin Resistance Study (Analytical): 185 isolates (87 Rif-R, 98 Rif-S) spiked into sputum.
Analytical Reactivity (Inclusivity): 62 M. tuberculosis strains (26 Rif-S, 36 Rif-R) tested in triplicate (186 total tests).
Exclusivity/Cross-Reactivity: 132 microorganisms (24 NTM, 87 bacteria, 7 fungi, 14 viruses) tested in triplicate (396 total tests for wet testing).

3. Number of Experts Used to Establish the Ground Truth for the Test Set and the Qualifications of Those Experts

The document does not explicitly state the number or qualifications of experts used to establish the ground truth for the clinical study's test set.

Ground truth for MTB-complex detection: Established by "culture (solid and/or liquid)". This implies standard microbiological laboratory procedures performed by qualified laboratory personnel, but no specific number or qualification of "experts" is given beyond this.
Ground truth for Rifampin resistance: Established by "phenotypic drug susceptibility testing (DST) using agar proportions methods with Middlebrook or Lowenstein-Jensen media or the BD BACTECTM MGITTM 960 SIRE assay." This also implies standard microbiological laboratory procedures.
Discordant analysis: "Bi-directional sequencing of the rpoB region of the MTB genome" was used to resolve discordant results between Xpert and culture/DST. This is a molecular method, likely performed by molecular biologists/scientists, but again, no specific number or detailed qualifications are provided.

4. Adjudication Method (e.g., 2+1, 3+1, none) for the Test Set

The document does not explicitly describe a formal adjudication method like "2+1" or "3+1" for the clinical study test set's primary ground truth (culture/DST).

However, it does describe a process for handling discordant results:

For MTB-complex detection, "Discordant results for MTB culture positive and Xpert MTB/RIF Assay "MTB NOT DETECTED" were further evaluated using bi-directional sequencing of the rpoB region of the MTB genome."
For Rifampin resistance, "Discordant results were further evaluated using bi-directional sequencing of the rpoB region of the MTB genome."
This implies that sequencing served as an adjudicator for specific discrepancies, effectively acting as an additional, confirmatory ground truth method in those cases. The primary ground truth for clinical performance remained culture and DST.

5. Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study

No, a Multi-Reader Multi-Case (MRMC) comparative effectiveness study focusing on human readers' improvement with AI vs. without AI assistance was not done.

This device is an in vitro diagnostic test (a PCR-based assay), not an AI-powered image analysis tool for human readers. Therefore, the concept of "human readers improve with AI vs. without AI assistance" does not apply to this type of device.

6. Standalone (Algorithm Only Without Human-in-the-Loop Performance) Study

Yes, this entire study is a standalone (algorithm only) performance assessment.

The Xpert® MTB/RIF Assay is an automated PCR system that processes samples and provides quantitative results (Ct values) and qualitative interpretations ("MTB DETECTED", "Rif Resistance DETECTED/NOT DETECTED/INDETERMINATE") directly. The results are "displayed in tabular and graphic formats." There is no "human-in-the-loop" interpretation step where a human reviews the algorithm's output and modifies it, or where a human's performance is compared before and after using the device's output. The performance metrics (sensitivity, specificity, reproducibility, etc.) directly reflect the device's algorithmic output compared to the established ground truth.

7. The Type of Ground Truth Used

For Mycobacterium tuberculosis complex DNA detection:
- Primary Ground Truth: Mycobacterial culture (solid and/or liquid).
- Confirmatory (for discordant results): Bi-directional sequencing of the rpoB region of the MTB genome.
For Rifampin-resistance associated mutations:
- Primary Ground Truth: Phenotypic drug susceptibility testing (DST) using agar proportions methods or BD BACTECTM MGITTM 960 SIRE assay.
- Confirmatory (for discordant results): Bi-directional sequencing of the rpoB region of the MTB genome.

8. The Sample Size for the Training Set

The document does not explicitly describe a separate "training set" in the context of machine learning or AI algorithm development. This is expected, as the Xpert® MTB/RIF Assay is a PCR-based diagnostic, not an AI/ML algorithm that is "trained" in the typical sense. Its development would involve analytical verification and validation based on known strains and biological principles, rather than iterative learning from a large dataset.

The term "training set" is generally not applicable to traditional in vitro diagnostic assays like this one. If an analogy were to be drawn, the internal "assay settings" (like valid Ct values) and the design of primers/probes were established during product development using various analytical studies and possibly earlier R&D experiments, which could be considered akin to "training" or "calibration" data, but not a formally delineated machine learning training set.

9. How the Ground Truth for the Training Set Was Established

As noted above, a formal "training set" in the AI/ML sense is not described. Therefore, the method for establishing ground truth for such a set is also not applicable here.

The analytical performance characteristics (like LoD, inclusivity, exclusivity) were established using well-characterized strains (e.g., ATCC Strain H37Rv, specific mutant strains) and clinical isolates, with ground truth determined by:

Prior MTB identification
Phenotypic drug susceptibility testing (DST)
Bi-directional sequencing

These methods serve to define the characteristics of these samples, which are then used to challenge and validate the assay's performance.

Ask a Question

Ask a specific question about this device

Page 1 of 1