Search Results

The BOND MMR Antibody Panel is intended to be used for the qualitative identification by light microscopy of human mismatch repair (MMR) proteins MLH1, MSH2, MSH6 and PMS2 in formalin-fixed, paraffin-embedded (FFPE) colorectal cancer (CRC) tissue sections by immunohistochemical staining. The BOND MMR Antibody Panel includes BOND Ready-to-Use Primary Antibody MLH1 (Mismatch Repair Protein) (ES05), BOND Ready-to-Use Primary Antibody MSH2 (Mismatch Repair Protein) (79H11), BOND Ready-to-Use Primary Antibody MSH6 (Mismatch Repair Protein) (EP49) and BOND Ready-to-Use Primary Antibody PMS2 (Mismatch Repair Protein) (EP51). The BOND MMR Antibody Panel is intended for use on the BOND-III or BOND-MAX fully automated systems with BOND Polymer Refine Detection.

The BOND MMR Antibody Panel is indicated for the detection of MMR protein deficiency as an aid in the identification of potential hereditary nonpolyposis colorectal cancer (HNPCC)/Lynch Syndrome in patients diagnosed with CRC. Patients with "MMR Loss" results should receive additional diagnostic testing consistent with clinical practice guidelines for diagnosis of Lynch syndrome. The BOND MMR Antibody Panel is not intended for use in indications other than CRC. This test should not be used for diagnosis of CRC.

The clinical interpretation of any staining or its absence when using the BOND MMR Antibody Panel should be complemented by morphological studies and proper controls and should be evaluated within the context of the patient's clinical history and other diagnostic tests by a qualified pathologist.

The clinical performance of this device to guide treatment of MMR deficient patients has not been established.

Device Description

The BOND MMR Antibody Panel [subject device] consists of the following BOND Ready-to-Use (RTU) Primary Antibody (PA) products:

MLH1 (Mismatch Repair Protein) (ES05) (PA0988-U) .
MSH2 (Mismatch Repair Protein) (79H11) (PA0989-U) .
. MSH6 (Mismatch Repair Protein) (EP49) (PA0990-U)
. PMS2 (Mismatch Repair Protein) (EP51) (PA0991-U)

The BOND MMR Antibody Panel is intended for use on the BOND-III or BOND-MAX fully automated systems with BOND Polymer Refine Detection (DS9800). The BOND MMR Antibody Panel is indicated for the detection of mismatch repair protein deficiency as an aid in the identification of potential Hereditary Non-Polyposis Colorectal Cancer (HPNCC)/Lynch Syndrome in patients diagnosed with CRC.

MLH1 (Mismatch Repair Protein) (ES05) is a mouse anti-human monoclonal antibody produced as a tissue culture supernatant, and supplied in Tris buffered saline with carrier protein, containing 0.35 % ProClin™ 950 as a preservative and in a total volume of 7 mL. The antibody is optimally diluted for use on the automated BOND-MAX or BOND-III instrument staining platforms in combination with BOND Polymer Refine Detection (DS9800).

MSH2 (Mismatch Repair Protein) (79H11) is a mouse anti-human monoclonal antibody produced as a tissue culture supernatant, and supplied in Tris buffered saline with carrier protein, containing 0.35 % ProClin™ 950 as a preservative and in a total volume of 7ml. The antibody is optimally diluted for use on the automated BOND-MAX or BOND-III instrument staining platforms in combination with BOND Polymer Refine Detection (DS9800).

MSH6 (Mismatch Repair Protein) (EP49) is a rabbit anti-human monoclonal antibody produced as an affinity-purified tissue culture supernatant, and supplied in Tris buffered saline with carrier protein, containing 0.35 % ProClin™ 950 as a preservative and in a total volume of 7ml. The antibody is optimally diluted for use on the automated BOND-MAX or BOND-III instrument staining platforms in combination with BOND Polymer Refine Detection (DS9800).

PMS2 (Mismatch Repair Protein) (EP51) is a rabbit anti-human monoclonal antibody produced as an affinity purified tissue culture supernatant, and supplied in Tris buffered saline with carrier protein, containing 0.35 % ProClin™ 950 as a preservative and in a total volume of 7ml. The antibody is optimally diluted for use on the automated BOND-MAX or BOND-III instrument staining platforms in combination with BOND Polymer Refine Detection (DS9800).

Instrument and Software: The BOND-MAX and BOND-III instruments are fully automated slide stainers that perform automated deparaffinization (dewaxing), antigen retrieval, immunohistochemistry (IHC) staining/in situ hybridization (ISH) staining, and counterstaining. The major components of the BOND staining platforms are the processing module, computer (BOND controller), handheld ID scanner, and slide label printer. The BOND staining platforms are composed of a number of discrete software components including the BOND application software, BOND instrument/processing module software, BOND service software, and Laboratory interface system - integration package (LIS-IP).

AI/ML Overview

Here's a breakdown of the acceptance criteria and the study that proves the device meets them, based on the provided text:

Device: BOND MMR Antibody Panel

Intended Use: Qualitative identification by light microscopy of human mismatch repair (MMR) proteins (MLH1, MSH2, MSH6, PMS2) in formalin-fixed, paraffin-embedded (FFPE) colorectal cancer (CRC) tissue sections by immunohistochemical staining. It aids in identifying potential hereditary nonpolyposis colorectal cancer (HNPCC)/Lynch Syndrome in CRC patients.

1. Table of Acceptance Criteria and Reported Device Performance

The document doesn't explicitly state "acceptance criteria" in a single, consolidated table. However, the precision studies reference "pre-specified acceptance criteria of ≥85% lower bound confidence interval." The clinical performance section states the point estimates of agreement, which implicitly serve as performance benchmarks.

Criterion Type	Specific Criterion (Implicit Acceptance Threshold)	Reported Device Performance (with 95% Confidence Interval)
Precision (Repeatability)	Lower bound 95% CI ≥ 85% for OPA, PPA, NPA	Intra-run:
		Anti-MLH1: OPA 100% [94.0%-100%] (BOND-III & BOND-MAX)
		Anti-MSH2: OPA 98.1% [90.2%-99.7%] (BOND-III), OPA 100% [93.4%-100%] (BOND-MAX)
		Anti-MSH6: OPA 100% [94.0%-100%] (BOND-III & BOND-MAX)
		Anti-PMS2: OPA 100% [94.0%-100%] (BOND-III), OPA 98.3% [91.1%-99.7%] (BOND-MAX)
		Between-day:
		All antibodies on both instruments generally met 100% OPA, with some slight variations (e.g., Anti-MSH2 on BOND-III: OPA 98.8% [95.6%-99.7%]; Anti-MSH6 on BOND-III: OPA 99.4% [96.9%-99.9%]) - all met ≥85% lower bound CI.
		Between-lot:
		Similar to Between-day, all results generally met 100% OPA, with some slight variations (e.g., Anti-MSH2 on BOND-III: OPA 98.8% [95.6%-99.7%]; Anti-PMS2 on BOND-III: OPA 98.9% [96.0%-99.7%]) - all met ≥85% lower bound CI.
Reproducibility (Pathologist & Laboratory)	Lower bound 95% CI ≥ 85% for OPA, PPA, NPA	Intra-pathologist: OPA 99.3% - 100%, PPA 99.2% - 100%
		Inter-pathologist: OPA 99.3% - 100%, PPA 99.2% - 100%
		Inter-instrument: OPA 98.9% - 100%, PPA 98.8% - 100%, NPA 93.3% - 100%
		Inter-laboratory: OPA 98.9% - 100%, PPA 98.8% - 100%
		Inter-day and Inter-site (BOND-III only): Overall OPA 94.4% to 100%, PPA 91.7% to 100%, NPA 97.8% to 100%. One exception for MSH6 PPA (91.7% [81.9%-96.4%]) was within specification due to a specific challenging case. All other lower bounds of 95% CI were ≥ 85%.
Clinical Performance (Agreement with DNA Sequencing Panel)	(Implicitly, high agreement is expected given the predicate device comparison)	Combined Cohort: PPA 93.3% [84.1%-97.4%], NPA 95.9% [88.6%-98.6%], OPA 94.7% [89.5%-97.4%]
		Sequential Cohort: PPA 83.3% [60.8%-94.2%], NPA 98.5% [92.0%-99.7%], OPA 95.3% [88.5%-98.2%]
		Enrichment Cohort: PPA 97.6% [87.7%-99.6%], NPA 66.7% [30.0%-90.3%], OPA 93.8% [83.2%-97.9%]
		Individual Protein Agreement (Combined Cohort):
		Anti-MLH1: PPA 89.2%, NPA 99.0%, OPA 96.2%
		Anti-MSH2: PPA 92.3%, NPA 99.2%, OPA 98.5%
		Anti-MSH6: PPA 65.0%, NPA 99.1%, OPA 94.0%
		Anti-PMS2: PPA 97.5%, NPA 95.7%, OPA 96.2%

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

Test Set Sample Size:

Precision Studies (Intra-run, Between-day, Between-lot):
- 40 FFPE CRC tissue cases (10 per MMR protein: 5 protein deficient, 5 intact) - used for intra-run, between-day, between-lot precision.
- One MSH2 case was excluded in some analyses due to insufficient tumor.
Reproducibility Studies (Pathologist & Laboratory):
- 30 FFPE CRC tissue cases (specific breakdown for each MMR protein provided in Table 4, e.g., MLH1: 25 intact, 5 loss).
Inter-day and Inter-site Reproducibility (BOND-III):
- 24 FFPE CRC tissue cases (3 intact and 3 loss cases for each of the 4 MMR proteins).
Clinical Performance Study:
- Initially, 155 cases procured.
- 143 cases were eligible and tested by both methods (BOND MMR Antibody Panel and DNA sequencing panel).
- 133 cases had valid results by both methods and were evaluable for agreement analysis. These comprised:
  - Sequential cohort: 94 cases (unknown MMR status, sequentially obtained from a single US site).
  - Enrichment cohort: 49 cases (known MMR protein deficiencies from multiple sites).

Data Provenance:

Clinical Performance Study: Specimens for the sequential cohort were obtained from a single US site. The enrichment cohort specimens were from multiple sites. The text indicates "Eligible remnant FFPE CRC tissues ("cases") were procured." This suggests the data are retrospective, using banked FFPE tissue samples.

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

Precision & Reproducibility Studies:

Intra-run, Between-day, Between-lot Precision: A single pathologist read and scored all stained slides. No specific qualifications are provided for this pathologist, other than being "a single pathologist."
Intra- and Inter-Pathologist Reproducibility: 3 pathologists read and scored the 30 cases. No specific qualifications beyond "pathologist" are provided.
Inter-Instrument Reproducibility: One pathologist evaluated the cases. No specific qualifications are provided.
Inter-Laboratory Reproducibility: One pathologist at each of the 3 sites independently evaluated each case. No specific qualifications beyond "pathologist" are provided.
Inter-day and Inter-site Reproducibility (BOND-III): One pathologist at each of the 3 sites read and scored the stained slides. No specific qualifications beyond "pathologist" are provided.

Clinical Performance Study:

"One pathologist at the testing site read and scored BOND MMR Antibody Panel stained slides in accordance with the scoring guidance." No specific qualifications are provided for this pathologist.
For the ground truth (DNA sequencing panel results): The ground truth was established by assessing "pathogenic mutation(s) likely to affect MMR protein expression in CRC" using a DNA sequencing panel. This is an objective molecular test, not dependent on expert visual interpretation.

4. Adjudication Method for the Test Set

Precision & Reproducibility Studies:

For precision studies where a single pathologist scored, "majority score" was used as the reference where multiple replicates were performed. For reproducibility studies involving multiple pathologists or sites, concordance between pathologists/sites was evaluated. No explicit adjudication process like "2+1" or "3+1" is described for resolving discrepancies to establish a single ground truth from pathologist reads for performance metrics. Instead, the studies assess agreement between readers and sites. The "majority score" used in the precision studies implied that if there were discordant reads, the majority would determine the "true" result for that specific replicate.

Clinical Performance Study:

The ground truth for the clinical performance study was the DNA sequencing panel result. The pathologist's interpretation of the BOND MMR Antibody Panel staining was compared against this molecular ground truth. Therefore, no pathologist adjudication of the test device results was used to establish the ground truth for this comparison.

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

Was an MRMC study done? Yes, aspects of MRMC designs are present in the reproducibility studies, particularly the "Intra- and Inter-Pathologist Reproducibility" and "Inter-Laboratory Reproducibility" sections. These studies involved multiple pathologists reading multiple cases to assess the reproducibility of the device's output.
Effect size of human reader improvement with AI vs. without AI assistance: Not applicable. This device is an immunohistochemistry (IHC) panel, interpreted by pathologists using light microscopy. It is not an AI-assisted diagnostic device, nor does the study evaluate human reader performance with or without AI assistance. The study focuses on the reproducibility and clinical validity of the IHC panel itself.

6. Standalone Performance Study (Algorithm Only)

This question is not applicable as the device is an Immunohistochemistry (IHC) panel, not an algorithm or AI system. Its performance is intrinsically tied to human interpretation (by a pathologist). The studies evaluate the performance of the IHC panel as interpreted by pathologists.

7. Type of Ground Truth Used

Precision/Reproducibility Studies: The "ground truth" for evaluating these studies was largely based on the expected protein expression status of the selected CRC tissue cases (e.g., "5 cases being protein deficient and 5 cases expressing intact protein"). For calculating agreements over repeat measurements, often a "majority score" from multiple reads or comparison to a baseline read was used as the reference.
Clinical Performance Study: The ground truth was established by a molecular test: a DNA sequencing panel validated for detecting pathogenic mutations likely to affect MMR protein expression in CRC. This provides an objective measure of MMR gene status, which is then correlated with the protein expression detected by the IHC panel.

8. Sample Size for the Training Set

The document describes pre-market testing and performance characterization, not the development or training of an AI algorithm. Therefore, there is no mention of a "training set" sample size for an algorithm. The "Immunoreactivity" section (Table 15-17) shows the testing of the antibodies across a wide variety of normal, neoplastic, and colorectal cancer tissues (dozens to hundreds of cases across various tissue types) to characterize their expected staining patterns, which could be considered part of the development/characterization phase, but not an algorithmic "training set."

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

As there is no training set for an AI algorithm mentioned in the document (the device is an IHC panel), this question is not applicable. The "Immunoreactivity" section's characterization of normal and tumor tissue staining patterns served as the basis for understanding expected reactivity, likely through known biological knowledge and expert pathological evaluation.

Ask a Question

Ask a specific question about this device

Page 1 of 1