Oncomine Dx Express Test

P240040 · Life Technologies Corporation · PQP · Jul 2, 2025 · Pathology

Device Facts

• Record ID: P240040
• Device Name: Oncomine Dx Express Test
• Applicant: Life Technologies Corporation
• Product Code: PQP · Pathology
• Decision Date: Jul 2, 2025
• Decision: APPR
• Device Class: Class 3
• Attributes: PCCP

Intended Use

The Oncomine Dx™ Express Test is a qualitative in vitro diagnostic test that uses targeted next-generation sequencing (NGS) technology to detect substitutions, insertions, and deletions in 42 genes from DNA, copy number variants (CNVs) in 10 genes from DNA, and fusions or splice variants in 18 genes from RNA isolated from formalin-fixed, paraffin-embedded (FFPE) tumor samples, using the Genexus Dx™ Integrated Sequencer. The test is intended to be used as a companion diagnostic to identify patients who may benefit from treatment with the targeted therapy listed in Table 1 in accordance with the approved therapeutic product labeling. Additionally, the test is intended to provide tumor mutation profiling information to be used by qualified health care professionals in accordance with professional guidelines in oncology for cancer patients with solid malignant neoplasms. Genomic findings other than those listed in Table 1 are not prescriptive or conclusive for labeled use of any specific therapeutic product.

Device Story

ODxET is a distributed NGS-based IVD test for FFPE tumor samples; utilizes Ion Torrent Genexus Dx System (sequencer, purification, reagents). Input: DNA/RNA from FFPE tumor tissue. Workflow: automated nucleic acid extraction, library prep, template prep, sequencing, and analysis via Genexus Dx Software. Output: report identifying variants (SNVs, indels, CNVs, fusions, splice variants) categorized by clinical significance (Level 1 CDx, Level 2/3 profiling). Used in clinical settings by qualified professionals. Output informs oncology treatment decisions, specifically identifying NSCLC patients for sunvozertinib therapy and providing broader mutation profiling for solid tumors.

Clinical Evidence

Retrospective clinical bridging study (N=280) using samples from WU-KONG1B trial (NCT03974022) and commercial sources. ODxET demonstrated 97.9% PPA and 100% NPA (excluding unknowns) compared to CTAs. Clinical efficacy (ORR) in ODxET-positive population (41.4%) was consistent with CTA-positive population (45.9%). Analytical validation included accuracy, LoD, precision, and stability studies across multiple tumor types.

Technological Characteristics

Targeted NGS panel using Ion AmpliSeq HD primers. Detects DNA variants (42 genes, 10 CNV genes) and RNA variants (18 genes). Platform: Ion Torrent Genexus Dx Integrated Sequencer. Software: Genexus Dx Software with Assay Definition File (ADF). Connectivity: Thermo Fisher Connect (TFC) for documentation. Sterilization: N/A (reagents).

Indications for Use

Indicated for patients with solid malignant neoplasms, including non-small cell lung cancer (NSCLC) for EGFR exon 20 insertion detection to identify eligibility for ZEGFROVY (sunvozertinib) therapy. No contraindications.

Regulatory Classification

Identification

A next generation sequencing (NGS) oncology panel is a device used for the qualitative detection of germline or somatic variants in one or more cancer-related genes. The device is intended to be used on DNA or RNA isolated from human clinical specimens.

Reference Devices

• Oncomine Dx Target Test (P160045/S027)

Submission Summary (Full Text)

{0}

# SUMMARY OF SAFETY AND EFFECTIVENESS DATA (SSED)

## I. GENERAL INFORMATION

Device Generic Name: Next generation sequencing oncology panel, somatic or germline variant detection system

Device Trade Name: Oncomine™ Dx Express Test

Device Procode: PQP

Applicant's Name and Address: Life Technologies Corporation 7305 Executive Way Frederick, MD 21704

Date(s) of Panel Recommendation: None

Premarket Approval Application (PMA) Number: P240040

Date of FDA Notice of Approval: July 2, 2025

## II. INDICATIONS FOR USE

The Oncomine Dx™ Express Test is a qualitative *in vitro* diagnostic test that uses targeted next-generation sequencing (NGS) technology to detect substitutions, insertions, and deletions in 42 genes from DNA, copy number variants (CNVs) in 10 genes from DNA, and fusions or splice variants in 18 genes from RNA isolated from formalin-fixed, paraffin-embedded (FFPE) tumor samples, using the Genexus Dx™ Integrated Sequencer.

The test is intended to be used as a companion diagnostic to identify patients who may benefit from treatment with the targeted therapy listed in Table 1 in accordance with the approved therapeutic product labeling.

Table 1. Companion Diagnostic Indications

| Tissue Type | Gene | Variant | Targeted Therapy |
| --- | --- | --- | --- |
| Non-small cell lung cancer (NSCLC) | EGFR | EGFR exon 20 insertions | ZEGFROVY™ (sunvozertinib) |

Additionally, the test is intended to provide tumor mutation profiling information to be used by qualified health care professionals in accordance with professional guidelines in oncology for cancer patients with solid malignant neoplasms. Genomic findings other than those listed in Table 1 are not prescriptive or conclusive for labeled use of any specific therapeutic product.

PMA P240040: FDA Summary of Safety and Effectiveness Data

{1}

PMA P240040: FDA Summary of Safety and Effectiveness Data
2 of 85

# III. CONTRAINDICATIONS

There are no known contraindications.

# IV. WARNINGS AND PRECAUTIONS

The warnings and precautions can be found in the Oncomine Dx Express Test labeling.

# V. DEVICE DESCRIPTION

The Oncomine Dx Express Test (ODxET) is a distributed *in vitro* diagnostic test that provides primer panels, assay controls and interpretative software (an Assay Definition File [ADF]) designed for use with the Ion Torrent™ Genexus Dx System (including Ion Torrent™ Genexus Dx Integrated Sequencer, Ion Torrent™ Genexus Dx Purification System, and Reagents & Consumables) for detection of alterations in deoxyribonucleic acid (DNA) and ribonucleic acid (RNA) isolated from FFPE tumor specimens. The test is intended to detect substitutions, insertions, and deletions in 42 genes; CNVs in 10 genes; and fusions or splice variants in 18 genes, as outlined in Table 2, across solid tumors, along with EGFR exon 20 insertions in NSCLC.

ODxET Panel provides a set of Ion AmpliSeq HD primers that target key regions of 46 cancer-related genes. The ODxET reports targeted variants from the following genes:

Table 2. ODxET Reportable Gene List

| Alteration Type | Genes |
| --- | --- |
| Deletions, insertions, substitutions from DNA (42 genes) | AKT1, AKT2, AKT3, ALK, AR, ARAF, BRAF, CDK4, CHEK2, CTNNB1, EGFR, ERBB2, ERBB3, ERBB4, ESR1, FGFR1, FGFR2, FGFR3, FGFR4, FLT3, GNAS, HRAS, IDH1, IDH2, KEAP1, KIT, KRAS, MAP2K1, MAP2K2, MET, NRAS, NTRK1, NTRK2, NTRK3, PDGFRA, PIK3CA, PTEN, RAF1, RET, ROS1, STK11, TP53 |
| Copy number variants from DNA (10 genes) | AR, EGFR, ERBB2, ERBB3, FGFR1, FGFR2, FGFR3, KRAS, MET, PIK3CA |
| Gene fusions and splice variants from RNA (18 genes) | ALK, AR, BRAF, EGFR, ESR1, FGFR1, FGFR2, FGFR3, MET, NRG1, NTRK1, NTRK2, NTRK3, NUTM1, RET, ROS1, RSPO2, RSPO3 |

# A. Test Output

The test report includes variants reported in the following levels:

- Level 1: Companion Diagnostics (CDx) biomarkers noted in Table 1 of the Intended Use
- Level 2: Cancer mutations with evidence of clinical significance
- Level 3: Cancer mutations with potential clinical significance

{2}

Genomic findings (i.e., Levels 2 and 3) other than those listed in Table 1 of the companion diagnostic indications table of the intended use statement are not prescriptive or conclusive for labeled use of any specific therapeutic product.

## B. Test Kit Contents and Instruments

The ODxET consists of the following:

- Oncomine™ Dx Express Test and Controls Kit:
- Oncomine™ Dx Express Test Panel
- Oncomine™ Dx Express Test FFPE DNA Control
- Oncomine™ Dx Express Test FFPE RNA Control
- Oncomine™ Dx Express Test No Template Control

- Genexus™ Dx FFPE DNA/RNA Purification Combo Kit:
- Genexus™ Dx FFPE DNA/RNA Purification Combo Kit Ion Torrent™ Dx cDNA Synthesis Kit
- Genexus™ Dx Nucleic Acid Quantitation
- Genexus™ Dx Purification Supplies 2

- Genexus™ Dx Integrated Sequencer Reagents and Consumables:
- Genexus™ Dx Library Strips 1 and 2-HD
- Genexus™ Dx Templating Strips 3-GX5 and 4
- Genexus™ Dx Sequencing Kit
- Genexus™ Dx Barcodes 1-32 HD
- Genexus™ Dx GX5 Chip and Genexus Coupler
- Genexus™ Dx Pipette Tips
- Genexus™ Dx Filter
- Genexus™ Dx Conical Bottles

- Instrumentation and Software:
- Genexus™ Dx Purification System
- Genexus™ Dx Integrated Sequencer
- Genexus™ Dx Software

The ODxET User Guides and ADF are provided as electronic files delivered to users through Thermo Fisher Connect (TFC). Users receive the specific ODxET product SKUs they have ordered. The packaging label includes URLs for further information: www.thermofisher.com/GXDxDocs and www.thermofisher.com/ODxETDocs, which direct users to an online PDF with detailed instructions on accessing the User Guides and ADF.

PMA P240040: FDA Summary of Safety and Effectiveness Data
3 of 85

{3}

C. Test Workflow

a. Sample and Assay Run Setup

Information about each sample is entered into Genexus™ Dx Software, which directs the progress of the sample from creation of a run plan through automated sample preparation (as applicable), library preparation, template preparation, sequencing, and analysis. During creation of a run plan, samples are selected and assigned to the run plan. The assay selected in the run plan specifies instrument run and analysis parameters and directs the reporting output of sample results.

b. Nucleic Acid Extraction

Nucleic acid extraction is performed with the Genexus™ Dx Purification System or other IVD-labeled extraction chemistries. When using the Genexus™ Dx Purification System specimens are treated with protease, then loaded into a sample input plate. The Genexus™ Dx Purification System is loaded with the sample plate and consumables specific for the sample and run type. During the automated run, the instrument purifies samples using a magnetic bead-based procedure, quantifies sample concentration, then loads the purified samples in a sequencer-ready output plate.

c. Library Preparation

In the first stage of a Genexus™ Dx Integrated Sequencer run, complementary DNA (cDNA) is prepared from RNA sample, then sample and control amplicon libraries are prepared from DNA or cDNA. Oncomine™ Dx Express Test panel primers, reagents in Genexus™ Dx Strip 1 and Genexus™ Dx Strip 2-HD, and barcode adapters from the Genexus™ Dx Barcodes 1–32 HD plate are used in library preparation. Each library that is generated in a run has a distinguishing nucleic acid sequence barcode that is incorporated into each amplified molecule.

d. Template Preparation

In the template preparation stage of the automated Genexus™ Dx Integrated Sequencer run, library molecules are bound onto Ion Sphere Particles (ISPs) and each nucleic acid sequence is amplified over the ISP surface using reagents in the Genexus™ Dx Strip 3-GX5™ and Genexus™ Dx Strip 4. The templated ISPs are enriched and collected using magnetic beads, then the enriched ISPs are magnetically loaded into wells of the sequencing chip. A second amplification step is then carried out with ISPs loaded on the chip, one ISP per well. After amplification, sequencing primer is annealed to the single-stranded template, and sequencing enzyme is added.

PMA P240040: FDA Summary of Safety and Effectiveness Data
4 of 85

{4}

# e. Sequencing

In the sequencing stage of the run, the sequencer flows one of four nucleotides over the Genexus™ Dx GX5™ Chip surface in a programmed series of flows. A base that is incorporated into the newly synthesized strands of an ISP during a flow, results in a release of protons and a concomitant pH change in the well. The change in pH is detected by sensors at the base of each well on the chip. Electrical signals that correspond to pH changes generated in each well during sequencing are processed and stored in data files on the system. Throughout this procedure, as the sample is processed by the instrument, sample and reagent information is recorded and tracked by Genexus™ Dx Software.

# f. Data Analysis

On the internal server of the Genexus™ Dx Integrated Sequencer, the initial signals are processed and each base in a sequence is called (identified). Base calls are assembled into files representing the reads, which are strings of nucleotide bases in the order found in the original library molecules. The reads are then mapped to the reference files provided with the test. Finally, Genexus™ Dx Software assesses the mapped reads at specific nucleotide locations and detects variation in sequence information in comparison to a human reference sequence.

# D. Assay Validity Criteria—FFPE Samples (DNA and RNA)

Quality control (QC) specifications for the templating control, sample, no-template control (NTC), and purification are provided in Table 3.

Table 3. FFPE Workflow—Quality Control specifications for Templating Control, Sample, No-Template Control, and Purification

| Metric | Criteria |
| --- | --- |
| Templating Control QC - CF-1 (Run Level Control) | |
| Average Reads per Lane | ≥2383 |
| Mean AQ20 Read Length (bp) | ≥82 |
| Sample QC - DNA | |
| Median Absolute Pairwise Difference | ≤0.5 |
| DNA Mapped Reads | ≥100,000 |
| DNA Mean AQ20 Read Length (bp) | ≥64 |
| Sample QC - RNA | |
| Mapped Fusion Reads | ≥6,000 |
| RNA Expression Controls Detected | ≥5 |

PMA P240040: FDA Summary of Safety and Effectiveness Data

{5}

| Metric | Criteria |
| --- | --- |
| NTC QC - DNA | |
| Hotspot Calls | 0 |
| Average Base Coverage Depth | ≤15 |
| NTC QC - RNA | |
| Mapped Reads | ≤2848 |
| Total Fusion Calls | 0 |
| Purification QC | |
| Sample Concentration DNA | 0.5 ng/μL minimum – 512 ng/μL maximum |
| Sample Concentration RNA | 0.5 ng/μL minimum – 512 ng/μL maximum |

## E. Variant Calling Criteria

In addition to the run and clinical sample validity criteria listed above, several variant calling criteria established separately for different biomarkers are used to determine which biomarkers are reportable in a passing sample in a successful run.

Primary parameters evaluated to produce each variant call include number of reads supporting a variant call, allele frequency, chromosomal strand coverage, and ultimately, alignment with known and targeted hotspots by the panel included in the assay definition file (ADF). The variant caller algorithm calls variants using either read-based evidence or molecular family-based evidence.

### a. Variant Calling Criteria for Companion Diagnostic Biomarker

A targeted EGFR exon 20 insertion mutation will be called ‘positive’ if it meets the variant calling criteria specified in the ADF. The ODxET covers codons E762 to R776 and Q787 to L814 of EGFR exon 20. The EGFR exon 20 insertion annotation is applied only to in-frame insertions (or in-frame block substitutions with a net nucleotide gain) affecting codons 762–775.

### b. Variant Calling Criteria for Tumor Profiling Variants

The variant calling approach for tumor profiling incorporates multiple algorithms tailored to each variant class. Single nucleotide variants (SNVs), insertions, and deletions are identified by the variant caller algorithm using criteria such as read depth, quality score, and allele frequency. Copy number variants (CNVs) are evaluated using a separate algorithm that considers sample-specific tumor cellularity. Gene fusions and RNA splice variants are detected based on minimum thresholds for unique molecule counts or reads. In certain cases, variant-specific criteria are also applied.

PMA P240040: FDA Summary of Safety and Effectiveness Data

{6}

PMA P240040: FDA Summary of Safety and Effectiveness Data
7 of 85

# F. Predetermined Changed Control Plan

The ODxET includes a predetermined change control plan (PCCP) approved by the US Food and Drug Administration (FDA). The PCCP describes specific test methods for clinical and analytical validation, which includes analysis methods, acceptance criteria, and documentation requirements to support the implementation of variant updates in the ADF and software modifications without additional regulatory review.

The ADF consists of individual files representing unique variants, such as SNVs, insertions, deletions, CNVs, and fusion isoforms, along with their associated clinical categories. The PCCP includes a standardized verification and validation (V&V) protocol for assessing changes in targeted variants and variant categorization during the aggregation of FDA-approved ADFs into the commercially released ODxET ADF. The protocol outlines specific validation steps and acceptance criteria to support this assessment. Life Technologies Corporation will perform variant category file validation as part of the ADF V&V process to ensure accurate implementation of variant re-categorization. Variant category files are generated with corresponding QC files that document the supporting evidence used for category determination. These QC files will be reviewed to verify category assignments prior to integration into the ADF. For companion diagnostic (CDx) claims, QC files for all associated tumor types will be reviewed. For tumor profiling claims, QC files for a subset of clinically relevant tumor types undergo manual review before ADF integration. The review process includes verification of mutation identifiers, variant types, and consistency of category assignments between the QC and category files. Manual review of supporting clinical evidence, such as FDA-approved therapies, National Comprehensive Cancer Network (NCCN) guidelines, and published global clinical trial data, will also be performed to confirm appropriate categorization. Newly validated variants will be aggregated to generate an updated base ADF. During the aggregation process, only unique content from each contributing ADF is incorporated into the updated base ADF to ensure that all validated and reportable content is preserved and accurately reflected in the software deliverables used for clinical reporting. Regression testing will also be performed during aggregation to confirm that the integration does not negatively impact patient outcomes.

The PCCP also addresses software modifications, including major and minor releases, security updates, data management and restoration procedures, configuration updates, and enhancements related to assay development mode. For each software release, regression testing of all existing test cases, as well as testing of new features and bug fixes, is conducted in accordance with an updated software verification test plan. Additional algorithm regression testing is performed using a subset of data from the clinical and analytical validation studies. For instrument control software, verification includes testing of new features, resolution of known issues, and regression testing of any functionality potentially affected by the implemented changes. If a new software release version is issued due to modifications implemented in accordance with the

{7}

PCCP, Life Technologies Corporation notifies all affected customers to ensure the most current version is installed and in use.

## VI. ALTERNATIVE PRACTICES AND PROCEDURES

There is an FDA approved CDx alternative for the detection of $EGFR$ exon 20 insertions using NSCLC FFPE tissue specimens, as shown in Table 4. A patient should fully discuss these alternatives with his/her physician to select the method that best meets expectations and lifestyle.

Table 4. FDA-Approved CDx Alternative for EGFR exon 20 insertions in NSCLC

| Gene | Variant | Device | Company | Technology | Therapy |
| --- | --- | --- | --- | --- | --- |
| EGFR | EGFR exon 20 insertion | Oncomine Dx Target Test | Thermo Fisher Scientific | Next Generation Sequencing (NGS) | RYBREVANT™ (amivantamab-vmjw) (P160045/S027) |

For additional details, see FDA List of Cleared or Approved Companion Diagnostic Devices at: https://www.fda.gov/medical-devices/in-vitro-diagnostics/list-cleared-or-approved-companion-diagnostic-devices-in-vitro-and-imaging-tools.

## VII. MARKETING HISTORY

The ODxET has not been marketed in the United States. The ODxET is commercially available in 17 countries in Europe (Austria, Denmark, France, Germany, Greece, Hungary, Ireland, Italy, Malta, Netherlands, Norway, Poland, Romania, Slovenia, Spain, Sweden, and United Kingdom), and Malaysia. The ODxET has not been withdrawn from these markets for any reasons related to its safety or effectiveness.

## VIII. POTENTIAL ADVERSE EFFECTS OF THE DEVICE ON HEALTH

Failure of the device to perform as expected or failure to correctly interpret test results may lead to incorrect ODxET results and subsequently improper patient management and treatment decisions for patients with NSCLC. Patients with false positive results may undergo treatment with the therapy listed in the intended use statement without clinical benefit and may experience adverse reactions associated with the therapy. Patients with false negative results may not be considered for treatment with the indicated therapy. There is also a risk of delayed results, which may lead to delay of treatment with the indicated therapy.

No device-related adverse events were reported in connection with the clinical studies used to support this PMA as the studies were performed retrospectively using banked samples.

PMA P240040: FDA Summary of Safety and Effectiveness Data

{8}

For the specific adverse events that occurred in the clinical study, please see the FDA approved package insert for ZEGFROVY™ (sunvozertinib), which is available at Drugs@FDA.

## IX. SUMMARY OF NON-CLINICAL STUDIES

### A. Laboratory Studies

The performance characteristics for the ODxET were established using DNA and RNA derived from a variety of FFPE tumor types. These FFPE tumor samples were analyzed for CNVs and a broad range of representative substitutions, insertions, and deletions in DNA, as well as fusions and splice variants in RNA, including CDx and tumor profiling variants. Several analytical validation studies were conducted to support the CDx and tumor profiling claims described in the intended use statement. All studies were performed using intended use specimens and sample blends and are summarized below.

#### 1. Analytical Accuracy/Concordance– Comparison to an Orthogonal Method

##### a. Analytical Accuracy for CDx Variants

The accuracy of the ODxET for detecting EGFR exon 20 insertions in patients with NSCLC was evaluated by comparing the test results of the ODxET with those of a validated NGS-based orthogonal method. A total of 202 tissue samples were obtained from subjects enrolled in the sunvozertinib clinical trial (WU-KONG1B, NCT03974022). Of these, 200 were EGFR exon 20 insertion-positive and 2 were negative, as determined by the clinical trial assays (CTAs) used for patient enrollment. Additionally, 232 samples were procured from commercial vendors and screened using NGS- and polymerase chain reaction (PCR)-based assays representative of the technologies commonly used for patient enrollment in the WU-KONG1B trial to identify EGFR exon 20 insertion-negative cases. Of these biomarker-negative samples, 143 were confirmed as negative, 3 were positive, and 86 were deemed invalid.

A total of 348 samples including 203 EGFR exon 20 insertion-positive samples and 145 EGFR exon 20 insertion-negative samples were initially submitted for testing with ODxET. Of the 203 biomarker-positive sample, 150 were tested using ODxET. Among these, 144 were confirmed as EGFR exon 20 insertion-positive, 3 were identified as negative, and 3 yielded indeterminate results. All 150 samples tested with ODxET were submitted for testing with the orthogonal method; however, 56 samples had insufficient DNA, and only 94 could be evaluated. Of these, 84 were confirmed as positive, 5 were negative, and 5 produced invalid or no call results by the orthogonal method. Among the 145 biomarker-negative samples, 138 were tested with ODxET. Of these, 134 were confirmed as EGFR exon 20 insertion-negative, while 4 did not yield valid results. All 138 samples were

PMA P240040: FDA Summary of Safety and Effectiveness Data
9 of 85

{9}

submitted for testing with the orthogonal method; however, due to insufficient DNA, only 108 were evaluable. Among the evaluable samples, 106 were confirmed as $EGFR$ exon 20 insertion-negative, and 2 produced invalid results.

The analytical concordance between the results from the ODxET and the orthogonal method for detecting $EGFR$ exon 20 insertions, based on a total of 348 samples (203 $EGFR$ exon 20 insertion-positive samples and 145 $EGFR$ exon 20 insertion-negative samples) initially submitted for testing with ODxET, is summarized in Table 5.

Table 5. Analytical Concordance Between ODxET and Orthogonal Method for EGFR Exon 20 Insertions

| | Orthogonal Method | | | | | Total | |
| --- | --- | --- | --- | --- | --- | --- | --- |
| | | Positive | Negative | \( Invalid^1 \) | No Calls | | \( TNP^2 \) |
| ODxET | Positive | 84 | 5 | 3 | 2 | 50 | 144 |
| | Negative | 0 | 105 | 1 | 0 | 31 | 137 |
| | \( Invalid^1 \) | 0 | 0 | 0 | 0 | 2 | 2 |
| | No Calls | 0 | 1 | 0 | 1 | 3 | 5 |
| | \( TNP^2 \) | 0 | 0 | 0 | 0 | 60 | 60 |
| | Total | 84 | 111 | 4 | 3 | 145 | 348 |
| 1Invalid or no calls were considered unknown or unevaluable.2TNP=Test Not Performed due to DNA concentration fail, tumor content fail, insufficient material, or DNA. Samples with TNP are categorized as unknown. | | | | | | | |

The positive percent agreement (PPA), negative percent agreement (NPA), overall percent agreement (OPA), and their corresponding two-sided $95\%$ Wilson score confidence intervals (CIs), using the orthogonal method as the reference, are provided in Table 6. When excluding ODxET unknowns, the point estimates were $100\%$ (84/84) for PPA, $95.5\%$ (105/110) for NPA, and $97.4\%$ (189/194) for OPA. When including ODxET unknowns, the point estimates were $100\%$ (84/84) for PPA, $94.6\%$ (105/111) for NPA, and $96.9\%$ (189/195) for OPA.

Table 6. Agreement Between ODxET and Orthogonal Method for EGFR Exon 20 Insertion Detection

| | | Excluding ODxET Unknowns | | | Including ODxET Unknowns | | |
| --- | --- | --- | --- | --- | --- | --- | --- |
| Parameter | Agreed (N) | Total (N) | Percent Agreement | 95% CIs | Total (N) | Percent Agreement | 95% CIs |
| PPA | 84 | 84 | 100% | (95.6%, 100%) | 84 | 100% | (95.6%, 100%) |
| NPA | 105 | 110 | 95.5% | (89.8%, 98.0%) | 111 | 94.6% | (88.7%, 97.5%) |
| OPA | 189 | 194 | 97.4% | (94.1%, 98.9%) | 195 | 96.9% | (93.5%, 98.6%) |

In the WU-KONG1B trial, the study population was enriched for cases positive for $EGFR$ exon 20 insertions. Consequently, the analytical concordance between ODxET and the orthogonal method was reevaluated by

PMA P240040: FDA Summary of Safety and Effectiveness Data

{10}

calculating prevalence-adjusted values for PPA and NPA, with corresponding two-sided 95% CIs estimated using a percentile bootstrap method. At a prevalence of 1.8%, the adjusted point estimates for PPA and NPA were 100.0% (95.6%, 100.0%) and 99.9% (99.8%, 100.0%), respectively, demonstrating high analytical concordance between ODxET and the comparator method.

Additionally, the positive predictive value (PPV) and adjusted negative predictive value (NPV) for the comparison between ODxET and the orthogonal method were calculated based on the prevalence of EGFR exon 20 insertions in NSCLC, which is 1.8%. Excluding ODxET unknowns, the PPV was 95.7% (91.9%, 99.1%), while including unknowns, it was 94.9% (90.9%, 98.3%). The NPV was 100% (96.5%, 100%) regardless of whether ODxET unknowns were excluded or included. The 95% CIs for the PPV were estimated using the percentile bootstrap method, whereas the 95% CIs for the NPV were calculated using the Wilson Score method due to the point estimate being 100%.

## b. Analytical Accuracy for Tumor Profiling Variants

To evaluate the analytical accuracy of ODxET for detecting SNVs, insertions and deletions (indels), CNVs, fusions, and RNA splice variants in FFPE clinical tumor samples, ODxET results were compared with those obtained from validated orthogonal methods. The accuracy assessment for tumor profiling variants was conducted across three separate studies. Study I included Level 2 and Level 3 SNVs, insertions, deletions, CNVs, fusions, and RNA splice variants across various tumor types. Study II focused on Level 2 ERBB2 CNVs in breast cancer, ALK fusions in NSCLC, and RET fusions in both NSCLC and thyroid cancer. Study III evaluated Level 2 EGFR SNVs and exon 19 deletions in NSCLC. For all three phases, samples were pre-screened using an orthogonal method similar to ODxET to confirm the presence or absence of target variants when vendor-provided screening information was insufficient or unclear. In the three analytical accuracy studies conducted for the tumor profiling indication, the PPA and NPA estimates were reported without adjustment for the distribution of samples selected by the pre-screening method. The analytical accuracy results from each study are presented separately below.

## Accuracy Study I – Tumor Profiling Variants

The analytical accuracy study I evaluated the performance of ODxET for detecting a range of tumor profiling variants, including SNVs, insertions, deletions, CNVs, fusions, and RNA splice variants, in FFPE clinical tumor samples. The samples were sourced from qualified commercial vendors and were pre-screened for target variants using an orthogonal method similar to ODxET. The study included 397 SNVs/multi nucleotide variants (MNVs), 5 insertions, 23 deletions, 114 CNVs, and 47 fusion or splice samples.

PMA P240040: FDA Summary of Safety and Effectiveness Data
11 of 85

{11}

Additionally, 58 unique wild-type (WT) samples and 42 samples pre-identified as negative for entire variant classes were evaluated. Of these, 36 WT samples and 40 variant class-negative samples were included in the concordance analysis. A total of 462 FFPE tumor samples representing 15 tumor types was tested, as shown in Table 7.

Table 7. Number of Samples per Cancer Type

| Tissue Type | Number of Samples |
| --- | --- |
| Bile Duct | 13 |
| Bladder | 56 |
| Brain | 48 |
| Breast | 61 |
| Colon | 70 |
| Endometrium | 26 |
| Gastrointestinal | 1 |
| Head/Neck | 2 |
| Kidney | 2 |
| Lung | 90 |
| Pancreas | 9 |
| Prostate | 7 |
| Skin | 48 |
| Soft Tissue | 3 |
| Thyroid | 26 |
| TOTAL | 462 |

The study compared ODxET results with those obtained from a validated NGS-based orthogonal method. Concordance analysis was performed at the variant level using results from both the ODxET and the orthogonal test. PPA and NPA estimates, along with their respective two-sided $95\%$ CIs, were calculated using the orthogonal method as the reference. The accuracy results, excluding unknown results, are summarized by variant type and further stratified according to the FDA's Biomarker Class Level for tumor profiling NGS tests, as shown in Table 8.

Table 8. Accuracy Results for ODxET Tumor Profiling Variants by Variant Type and Clinical Significance – Accuracy Study I

| Variant Type | FDA Biomarker Category1 | Total Unique Variants/Genes | ODxET (+) | Comparator (+) | ODxET (+) | Comparator (-) | ODxET (-) | Comparator (+) | ODxET (-) | Comparator (-) | PPA (n/N) [95% CIs]2 | NPA (n/N) [95% CIs]2 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| All Variant Types | All | 2944/46 | 586 | 112 | 11 | 800035 | 98.2% (586/597) [96.7%, 99.0%] | 100.0% (800035/800147) [100.0%, 100.0%] |
| Level 2 | 1544/22 | 190 | 19 | 0 | 116800 | 100.0% (190/190) [98.0%, 100.0%] | 100.0% (116800/116819) [100.0%, 100.0%] |

PMA P240040: FDA Summary of Safety and Effectiveness Data

{12}

| Variant Type | FDA Biomarker Category1 | Total Unique Variants/Genes | ODxET (+) | Comparator (+) | ODxET (-) | Comparator (-) | ODxET (-) | Comparator (-) | ODxET (-) | Comparator (-) | PPA (n/N) [95% CIs]2 | NPA (n/N) [95% CIs]2 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| | Level 3 | 1400/43 | 396 | 93 | 11 | 683235 | 97.3% (396/407) [95.3%, 98.5%] | 100.0% (683235/683328) [100.0%, 100.0%] |
| SNVs | All | 1324/42 | 397 | 47 | 2 | 387643 | 99.5% (397/399) [98.2%, 99.9%] | 100.0% (387643/387690) [100.0%, 100.0%] |
| Level 2 | 265/17 | 137 | 4 | 0 | 10594 | 100.0% (137/137) [97.2%, 100.0%] | 100.0% (10594/10598) [99.9%, 100.0%] |
| Level 3 | 1059/38 | 260 | 43 | 2 | 377049 | 99.2% (260/262) [97.3%, 99.8%] | 100.0% (377049/377092) [100.0%, 100.0%] |
| MNVs + Complex | All | 287/14 | 0 | 0 | 0 | 6466 | N/E | 100.0% (6466/6466) [99.9%, 100.0%] |
| Level 2 | 210/12 | 0 | 0 | 0 | 343 | N/E | 100.0% (343/343) [98.9%, 100.0%] |
| Level 3 | 77/19 | 0 | 0 | 0 | 6123 | N/E | 100.0% (6123/6123) [99.9%, 100.0%] |
| Insertions | All | 75/11 | 5 | 1 | 0 | 61874 | 100.0% (5/5) [56.6%, 100.0%] | 100.0% (61874/61875) [100.0%, 100.0%] |
| Level 2 | 52/3 | 0 | 0 | 0 | 2772 | N/E | 100.0% (2772/2772) [99.9%, 100.0%] |
| Level 3 | 23/9 | 5 | 1 | 0 | 59102 | 100.0% (5/5) [56.6%, 100.0%] | 100.0% (59102/59103) [100.0%, 100.0%] |
| Deletions | All | 258/14 | 23 | 2 | 0 | 78635 | 100.0% (23/23) [85.7%, 100.0%] | 100.0% (78635/78637) [100.0%, 100.0%] |
| Level 2 | 206/5 | 14 | 1 | 0 | 9292 | 100.0% (14/14) [78.5%, 100.0%] | 100.0% (9292/9293) [99.9%, 100.0%] |
| Level 3 | 52/13 | 9 | 1 | 0 | 69343 | 100.0% (9/9) [70.1%, 100.0%] | 100.0% (69343/69344) [100.0%, 100.0%] |
| CNVs | All | 10/10 | 114 | 32 | 7 | 4294 | 94.2% (114/121) [88.5%, 97.2%] | 99.3% (4294/4326) [99.0%, 99.5%] |
| Level 2 | 2/2 | 14 | 3 | 0 | 238 | 100.0% (14/14) [78.5%, 100.0%] | 98.8% (238/241) [96.4%, 99.6%] |
| Level 3 | 8/8 | 100 | 29 | 7 | 4056 | 93.5% (100/107) [87.1%, 96.8%] | 99.3% (4056/4085) [99.0%, 99.5%] |
| RNA Variants | All | 990/18 | 47 | 30 | 2 | 261123 | 95.9% (47/49) [86.3%, 98.9%] | 100.0% (261123/261153) [100.0%, 100.0%] |
| Level 2 | 809/10 | 25 | 11 | 0 | 93561 | 100.0% (25/25) [86.7%, 100.0%] | 100.0% (93561/93572) [100.0%, 100.0%] |

PMA P240040: FDA Summary of Safety and Effectiveness Data

{13}

| Variant Type | FDA Biomarker Category1 | Total Unique Variants/Genes | ODxET (+) | Comparator (+) | ODxET (-) | Comparator (-) | ODxET (-) | Comparator (+) | ODxET (-) | Comparator (-) | PPA (n/N) [95% CIs]2 | NPA (n/N) [95% CIs]2 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| | Level 3 | 181/16 | 22 | 19 | 2 | 167562 | 91.7% (22/24) [74.2%, 97.7%] | 100.0% (167562/167581) [100.0%, 100.0%] |
| 1Level 2 = Cancer Mutations with Evidence of Clinical Significance, Level 3 = Cancer Mutations with Evidence of Potential Clinical Significance295% 2-sided confidence interval calculated via the Wilson score method.* N/E: Not evaluable. The statistic cannot be calculated as there were no sample variant results in this category.‘Total Unique Variants/Genes’ refers to the total number of unique variants detected in the study and the number of unique genes in which those variants are located. |

A summary of accuracy results for SNVs in predefined challenging genomic regions (e.g., simple sequence repeats [SSRs], homopolymers, GC content), and for insertions and deletions stratified by predefined size bins (e.g., 1-5 bp, 6-10 bp, 11-15 bp, $>15$ bp) and by challenging genomic regions, excluding unknown results, is presented in Table 9.

Table 9. Accuracy Results for SNVs, Insertions, and Deletions in Clinically and Analytically Meaningful Bins – Accuracy Study I

| Variant Type | Bin | Total Unique Variants | ODxET (+) | Comparator (+) | ODxET (-) | Comparator (-) | ODxET (-) | Comparator (+) | ODxET (-) | Comparator (-) | PPA (n/N) [95% CIs]1 | NPA (n/N) [95% CIs]1 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| SNVs | All | 1324 | 397 | 47 | 2 | 387643 | 99.5% (397/399) [98.2%, 99.9%] | 100.0% (387643/387690) [100.0%, 100.0%] |
| Non Homopolymers | 344 | 90 | 10 | 0 | 106693 | 100.0% (90/90) [95.9%, 100.0%] | 100.0% (106693/106703) [100.0%, 100.0%] |
| Homopolymers | 980 | 307 | 37 | 2 | 280950 | 99.4% (307/309) [97.7%, 99.7%] | 100.0% (280950/280987) [100.0%, 100.0%] |
| Non SSRs | 813 | 230 | 27 | 1 | 241707 | 99.6% (230/231) [97.6%, 99.9%] | 100.0% (241707/241734) [100.0%, 100.0%] |
| SSRs | 511 | 167 | 20 | 1 | 145936 | 99.4% (167/168) [96.7%, 99.9%] | 100.0% (145936/145956) [100.0%, 100.0%] |
| GC [30%-40%] | 282 | 111 | 3 | 0 | 87285 | 100.0% (111/111) [96.7%, 100.0%] | 100.0% (87285/87288) [100.0%, 100.0%] |
| GC (40%-50%] | 237 | 98 | 5 | 0 | 69985 | 100.0% (98/98) [96.2%, 100.0%] | 100.0% (69985/69990) [100.0%, 100.0%] |

PMA P240040: FDA Summary of Safety and Effectiveness Data

{14}

| Variant Type | Bin | Total Unique Variants | ODxET (+) | Comparator (+) | ODxET (-) | Comparator (-) | ODxET (-) | Comparator (+) | PDA (n/N) [95% CIs]1 | NPA (n/N) [95% CIs]1 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| | GC (50%-60%) | 552 | 143 | 29 | 2 | 163944 | 98.6% (143/145) [95.1%, 99.6%] |
| GC (60%-75%) | 253 | 45 | 10 | 0 | 66429 | 100.0% (45/45) [92.1%, 100.0%] |
| MNVs + Complex | All | 287 | 0 | 0 | 0 | 6466 | N/E |
| Non Homopolymers | 40 | 0 | 0 | 0 | 0 | N/E |
| Homopolymers | 247 | 0 | 0 | 0 | 6466 | N/E |
| Non SSRs | 223 | 0 | 0 | 0 | 1484 | N/E |
| SSRs | 64 | 0 | 0 | 0 | 4982 | N/E |
| GC [30%-40%] | 63 | 0 | 0 | 0 | 110 | N/E |
| GC (40%-50%) | 157 | 0 | 0 | 0 | 8 | N/E |
| GC (50%-60%) | 49 | 0 | 0 | 0 | 1366 | N/E |
| GC (60%-75%) | 18 | 0 | 0 | 0 | 4982 | N/E |
| Insertions | All | 75 | 5 | 1 | 0 | 61874 | 100.0% (5/5) [56.6%, 100.0%] |
| 1-5 bp | 22 | 1 | 1 | 0 | 15958 | 100.0% (1/1) [20.7%, 100.0%] |
| 6-10 bp | 27 | 4 | 0 | 0 | 30858 | 100.0% (4/4) [51.0%, 100.0%] |
| 11-15 bp | 12 | 0 | 0 | 0 | 9839 | N/E |
| >15bp | 14 | 0 | 0 | 0 | 5219 | N/E |
| Non Homopolymers | 8 | 0 | 0 | 0 | 2169 | N/E |
| Homopolymers | 67 | 5 | 1 | 0 | 59705 | 100.0% (5/5) [56.6%, 100.0%] |

PMA P240040: FDA Summary of Safety and Effectiveness Data

{15}

| Variant Type | Bin | Total Unique Variants | ODxET (+) | Comparator (+) | ODxET (-) | Comparator (-) | ODxET (-) | Comparator (-) | PPA (n/N) [95% CIs]1 | NPA (n/N) [95% CIs]1 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| | Non SSRs | 65 | 0 | 1 | 0 | 19201 | N/E |
| SSRs | 10 | 5 | 1 | 0 | 42673 | 100.0% (5/5) [56.6%, 100.0%] |
| GC [30%-40%] | 22 | 0 | 0 | 0 | 7387 | N/E |
| GC (40%-50%] | 7 | 0 | 0 | 0 | 2320 | N/E |
| GC (50%-60%] | 41 | 0 | 0 | 0 | 11643 | N/E |
| GC (60%-75%] | 5 | 5 | 1 | 0 | 40524 | 100.0% (5/5) [56.6%, 100.0%] |
| Deletions | All | 258 | 23 | 2 | 0 | 78635 | 100.0% (23/23) [85.7%, 100.0%] |
| 1-5 bp | 52 | 8 | 2 | 0 | 15160 | 100.0% (8/8) [67.6%, 100.0%] |
| 6-10 bp | 49 | 0 | 0 | 0 | 14697 | N/E |
| 11-15 bp | 63 | 14 | 0 | 0 | 19988 | 100.0% (14/14) [78.5%, 100.0%] |
| >15bp | 94 | 1 | 0 | 0 | 28790 | 100.0% (1/1) [20.7%, 100.0%] |
| Non Homopolymers | 25 | 0 | 1 | 0 | 7944 | N/E |
| Homopolymers | 233 | 23 | 1 | 0 | 70691 | 100.0% (23/23) [85.7%, 100.0%] |
| Non SSRs | 210 | 20 | 2 | 0 | 64874 | 100.0% (20/20) [83.9%, 100.0%] |
| SSRs | 48 | 3 | 0 | 0 | 13761 | 100.0% (3/3) [43.9%, 100.0%] |
| GC [30%-40%] | 142 | 7 | 1 | 0 | 43018 | 100.0% (7/7) [64.6%, 100.0%] |

PMA P240040: FDA Summary of Safety and Effectiveness Data

{16}

| Variant Type | Bin | Total Unique Variants | ODxET (+) | Comparator (+) | ODxET (-) | Comparator (-) | ODxET (-) | Comparator (+) | ODxET (-) | Comparator (-) | PPA (n/N) [95% CIs]1 | NPA (n/N) [95% CIs]1 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| | GC (40%-50%) | 75 | 14 | 0 | 0 | 24834 | 100.0% (14/14) [78.5%, 100.0%] | 100.0% (24834/24834) [100.0%, 100.0%] |
| GC (50%-60%) | 15 | 1 | 1 | 0 | 5081 | 100.0% (1/1) [20.7%, 100.0%] | 100.0% (5081/5082) [99.9%, 100.0%] |
| GC (60%-75%) | 26 | 1 | 0 | 0 | 5702 | 100.0% (1/1) [20.7%, 100.0%] | 100.0% (5702/5702) [99.9%, 100.0%] |
| 195% 2-sided confidence interval calculated via the Wilson score method.*N/E: Not evaluable. The statistic cannot be calculated as there were no sample variant results in this category. |

A summary of accuracy results for CNVs and RNA variants (including fusions and splice variants), excluding unknown results, is presented at the gene level in Table 10.

Table 10. Accuracy Results for CNV and RNA Variants at the Gene Level - Accuracy Study I

| Variant Type | Gene Name | ODxET (+) | Comparator (+) | ODxET (-) | Comparator (-) | ODxET (-) | Comparator (+) | ODxET (-) | Comparator (-) | PPA (n/N) [95% CIs]1 | NPA (n/N) [95% CIs]1 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| CNV | All | 114 | 32 | 7 | 4294 | 94.1% (112/119) [88.4%, 97.1%] | 99.2% (3861/3893) [98.8%, 99.4%] |
| AR | 3 | 0 | 1 | 441 | 75.0% (3/4) [30.1%, 95.4%] | 100.0% (441/441) [99.1%, 100.0%] |
| EGFR | 25 | 8 | 0 | 409 | 100.0% (25/25) [86.7%, 100.0%] | 98.1% (409/417) [96.3%, 99.0%] |
| ERBB2 | 22 | 4 | 0 | 419 | 100.0% (22/22) [85.1%, 100.0%] | 99.1% (419/423) [97.6%, 99.6%] |
| ERBB3 | 2 | 1 | 0 | 442 | 100.0% (2/2) [34.2%, 100.0%] | 99.8% (442/443) [98.7%, 100.0%] |
| FGFR1 | 19 | 1 | 1 | 424 | 95.0% (19/20) [76.4%, 99.1%] | 99.8% (424/425) [98.7%, 100.0%] |
| FGFR2 | 9 | 0 | 0 | 436 | 100.0% (9/9) [70.1%, 100.0%] | 100.0% (436/436) [99.1%, 100.0%] |
| FGFR3 | 9 | 14 | 1 | 421 | 90.0% (9/10) [59.6%, 98.2%] | 96.8% (421/435) [94.7%, 98.1%] |
| KRAS | 10 | 0 | 2 | 433 | 83.3% (10/12) [55.2%, 95.3%] | 100.0% (433/433) [99.1%, 100.0%] |
| MET | 9 | 3 | 2 | 431 | 81.8% (9/11) [52.3%, 94.9%] | 99.3% (431/434) [98.0%, 99.8%] |
| PIK3CA | 6 | 1 | 0 | 438 | 100.0% (6/6) | 99.8% (438/439) |

PMA P240040: FDA Summary of Safety and Effectiveness Data

{17}

| Variant Type | Gene Name | ODxET (+) | Comparator (+) | ODxET (+) | Comparator (-) | ODxET (-) | Comparator (+) | ODxET (-) | Comparator (-) | PPA (n/N) [95% CIs]1 | NPA (n/N) [95% CIs]1 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| | | | | | | [61.0%, 100.0%] | [98.7%, 100.0%] |
| Fusions | All | 32 | 20 | 1 | 260085 | 97.0% (32/33) [84.7%, 99.5%] | 100.0% (260085/260105) [100.0%, 100.0%] |
| ALK | 2 | 2 | 0 | 56654 | 100.0% (2/2) [34.2%, 100.0%] | 100.0% (56654/56656) [100.0%, 100.0%] |
| BRAF | 2 | 0 | 0 | 49739 | 100.0% (2/2) [34.2%, 100.0%] | 100.0% (49739/49739) [100.0%, 100.0%] |
| ESR1 | 3 | 0 | 0 | 8507 | 100.0% (3/3) [43.9%, 100.0%] | 100.0% (8507/8507) [100.0%, 100.0%] |
| FGFR1 | 0 | 0 | 0 | 9310 | N/E | 100.0% (9310/9310) [100.0%, 100.0%] |
| FGFR2 | 1 | 1 | 0 | 21012 | 100.0% (1/1) [20.7%, 100.0%] | 100.0% (21012/21013) [100.0%, 100.0%] |
| FGFR3 | 8 | 5 | 1 | 11158 | 88.9% (8/9) [56.5%, 98.0%] | 100.0% (11158/11163) [99.9%, 100.0%] |
| MET | 1 | 0 | 0 | 8245 | 100.0% (1/1) [20.7%, 100.0%] | 100.0% (8245/8245) [100.0%, 100.0%] |
| NRG1 | 1 | 1 | 0 | 9306 | 100.0% (1/1) [20.7%, 100.0%] | 100.0% (9306/9307) [99.9%, 100.0%] |
| NTRK1 | 1 | 0 | 0 | 23673 | 100.0% (1/1) [20.7%, 100.0%] | 100.0% (23673/23673) [100.0%, 100.0%] |
| NTRK2 | 0 | 1 | 0 | 8245 | N/E | 100.0% (8245/8246) [99.9%, 100.0%] |
| NTRK3 | 4 | 0 | 0 | 8772 | 100.0% (4/4) [51.0%, 100.0%] | 100.0% (8772/8772) [100.0%, 100.0%] |
| NUTM1 | 0 | 1 | 0 | 4255 | N/E | 100.0% (4255/4256) [99.9%, 100.0%] |
| RET | 0 | 0 | 0 | 23674 | N/E | 100.0% (23674/23674) [100.0%, 100.0%] |
| ROS1 | 9 | 4 | 0 | 15146 | 100.0% (9/9) [70.1%, 100.0%] | 100.0% (15146/15150) [99.9%, 100.0%] |
| RSPO2 | 0 | 0 | 0 | 1064 | N/E | 100.0% (1064/1064) [99.6%, 100.0%] |
| RSPO3 | 0 | 5 | 0 | 1325 | N/E | 99.6% (1325/1330) [99.1%, 99.8%] |
| Splice Variants /RNA Exon Variants | All | 15 | 10 | 1 | 1038 | 93.8% (15/16) [71.7%, 98.9%] | 99.0% (1038/1048) [98.3%, 99.5%] |
| AR | 6 | 9 | 0 | 251 | 100.0% (6/6) [61.0%, 100.0%] | 96.5% (251/260) [93.6%, 98.2%] |
| EGFR | 6 | 0 | 0 | 260 | 100.0% (6/6) [61.0%, 100.0%] | 100.0% (260/260) [98.5%, 100.0%] |
| MET | 3 | 1 | 1 | 527 | 75.0% (3/4) [30.1%, 95.4%] | 99.8% (527/528) [98.9%, 100.0%] |
| 195% 2-sided confidence interval calculated via the Wilson score method.*N/E: Not evaluable. The statistic cannot be calculated as there were no sample variant results in this category. |

PMA P240040: FDA Summary of Safety and Effectiveness Data

{18}

CNVs in the three genes $AR$ , $KRAS$ , and $MET$ showed relatively low PPA: $75\%$ (3/4) for $AR$ CNV, $83.3\%$ (10/12) for $KRAS$ CNV, and $81.8\%$ (9/11) for $MET$ CNV. All discordant calls were attributed to the presence of target variants in these samples below the limit of detection (LoD) established by ODxET.

A summary of accuracy results (PPA and NPA) for all reported variant types, stratified by the LoD level established by ODxET (i.e., at or above $1 \times$ LoD and below LoD) and excluding unknown results, is presented in Table 11. For analysis conditional on the ODxET LoD levels, PPV was calculated for all reported variant types using the same stratification and exclusion criteria. For RNA variants, 'fusion read' refers to a sequencing read that spans a gene fusion breakpoint, i.e., a read that provides evidence of a gene fusion event by mapping to two different genomic regions, and 'molecular count' refers to the number of unique fusion molecules present in a sample, as identified by molecular tags.

Table 11. Accuracy Results for All Variant Types Stratified by LoD Level – Accuracy Study I

| Variant Type | LoD Level | ODxET (+) | Comparator (+) | ODxET (-) | Comparator (-) | ODxET (-) | Comparator (-) | PPA (n/N) [95% CI]1 | NPA (n/N) [95% CIs]1 | PPV (n/N) [95% CIs]1 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| SNVs | All | 397 | 47 | 2 | 387643 | 99.5% (397/399) [98.2%, 99.9%] | 100.0% (387643/387690) [100.0%, 100.0%] |
| ≥ 1x LoD | 391 | 12 | 0 | - | 100.0% (391/391) [99.0%, 100.0%] | N/E |
| < 1x LoD | 6 | 35 | 2 | - | 75.0% (6/8) [40.9%, 92.9%] | N/E |
| Insertions | All | 5 | 1 | 0 | 61874 | 100.0% (5/5) [56.6%, 100.0%] | 100.0% (61874/61875) [100.0%, 100.0%] |
| ≥ 1x LoD | 5 | 1 | 0 | - | 100.0% (5/5) [56.6%, 100.0%] | N/E |
| < 1x LoD | 0 | 0 | 0 | - | N/E | N/E |
| Deletions | All | 23 | 2 | 0 | 78635 | 100.0% (23/23) [85.7%, 100.0%] | 100.0% (78635/78637) [100.0%, 100.0%] |
| ≥ 1x LoD | 23 | 2 | 0 | - | 100.0% (23/23) [85.7%, 100.0%] | N/E |
| < 1x LoD | 0 | 0 | 0 | - | N/E | N/E |
| CNVs | All | 114 | 32 | 7 | 4294 | 94.2% (114/121) [88.5%, 97.2%] | 99.3% (4294/4326) [99.0%, 99.5%] |
| ≥ 1x LoD | 113 | 29 | 0 | - | 100.0% (113/113) [96.7%, 100.0%] | N/E |

PMA P240040: FDA Summary of Safety and Effectiveness Data

{19}

| Variant Type | LoD Level | ODxET (+) | Comparator (+) | ODxET (+) | Comparator (-) | ODxET (-) | Comparator (+) | ODxET (-) | Comparator (-) | PPA (n/N) [95% CI]1 | NPA (n/N) [95% CIs]1 | PPV (n/N) [95% CIs]1 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| | < 1x LoD | 1 | 3 | 7 | - | 12.5% (1/8) [2.2%, 47.1%] | N/E | 25% (1/4) [4.56%, 69.94%] |
| Fusions | All | 32 | 20 | 1 | 260085 | 97.0% (32/33) [84.7%, 99.5%] | 100.0% (260085/260105) [100.0%, 100.0%] | 61.5% (32/52) [48.0%, 73.5%] |
| ≥ 1x LoD (Molecular Counts) | 32 | 13 | 0 | - | 100.0% (32/32) [89.3%, 100.0%] | N/E | 71.1% (32/45) [56.6%, 82.3%] |
| < 1x LoD (Molecular Counts) | 0 | 7 | 1 | - | 0.0% (0/1) [0.0%, 79.3%] | N/E | 0.0% (0/7) [0.0%, 35.4%] |
| ≥ 1x LoD (Fusion Reads) | 30 | 15 | 0 | - | 100.0% (30/30) [88.6%, 100.0%] | N/E | 66.7% (30/45) [52.1%, 78.6%] |
| < 1x LoD (Fusion Reads) | 2 | 5 | 1 | - | 66.7% (2/3) [20.8%, 93.9%] | N/E | 28.6% (2/7) [8.2%, 64.1%] |
| RNA Splice Variants | All | 15 | 10 | 1 | 1038 | 93.8% (15/16) [71.7%, 98.9%] | 100.0% (1038/1048) [100.0%, 100.0%] | 60.0% (15/25) [40.7%, 76.6%] |
| ≥ 1x LoD (Molecular Counts) | 14 | 3 | 0 | - | 100.0% (14/14) [78.5%, 100.0%] | N/E | 82.4% (14/17) [59.0%, 93.8%] |
| < 1x LoD (Molecular Counts) | 1 | 7 | 1 | - | 50.0% (1/2) [9.5%, 90.5%] | N/E | 12.5% (1/8) [2.2%, 47.1%] |
| ≥ 1x LoD (Fusion Reads) | 14 | 4 | 0 | - | 100.0% (14/14) [78.5%, 100.0%] | N/E | 77.8% (14/18) [54.8%, 91.0%] |
| < 1x LoD (Fusion Reads) | 1 | 6 | 1 | - | 50.0% (1/2) [9.5%, 90.5%] | N/E | 14.3% (1/7) [2.6%, 51.3%] |
| 195% 2-sided confidence interval calculated via the Wilson score method.*N/E: Not evaluable. The statistic cannot be calculated as there were no sample variant results in this category. |

SNVs (75.0%), CNVs (12.5%), and RNA variants (33.3%) showed low PPA when the target variants in these samples were present below the LoD established by ODxET.

# Accuracy Study II - Tumor Profiling Variants

The accuracy study II was conducted to evaluate the accuracy of ODxET for detecting ERBB2 CNVs in breast cancer, ALK fusions in NSCLC, and RET fusions in NSCLC and thyroid cancer by retrospectively testing FFPE clinical samples. The samples were sourced from qualified commercial vendors and were pre-screened for target variants using an orthogonal method similar to

PMA P240040: FDA Summary of Safety and Effectiveness Data

{20}

ODxET. All variants evaluated in this study were Level 2 variants, classified according to FDA's three-tiered approach for reporting biomarkers in tumor profiling NGS tests.

For ERBB2 CNVs, 74 breast cancer samples were tested using both the ODxET and the fluorescence in situ hybridization (FISH)-based orthogonal method. Of these, 66 samples (29 positive and 37 negative) had concordant results. Four samples were discordant, showing positive results with the ODxET but negative with the orthogonal method, and four samples were negative with the ODxET but had unknown results with the orthogonal method. Table 12 shows the $3 \times 3$ contingency table accounting for all 74 breast ERBB2 CNV samples tested by both the ODxET and the orthogonal method, along with the PPA and NPA estimates.

Table 12. Concordance Between ODxET and Orthogonal Method for Detection of Level 2 ERBB2 CNVs in Breast Cancer – Accuracy Study II

| | Orthogonal Method | | | | |
| --- | --- | --- | --- | --- | --- |
| | | Positive | Negative | Unknown1 | Total |
| ODxET | Positive | 29 | 4 | 0 | 33 |
| | Negative | 0 | 37 | 4 | 41 |
| | Unknown1 | 0 | 0 | 0 | 0 |
| | Total | 29 | 41 | 4 | 74 |
| Excluding Unknown | PPA% (n/N) [95% CIs]2 | 100.0% (29/29) [88.3%, 100.0%] | | | |
| | NPA% (n/N) [95% CIs]2 | 90.2% (37/41) [77.5%, 96.1%] | | | |
| Including Unknown | PPA% (n/N) [95% CIs]2 | 100.0% (29/29) [88.3%, 100.0%] | | | |
| | NPA% (n/N) [95% CIs]2 | 90.2% (37/41) [77.5%, 96.1%] | | | |
| 1Unknowns are defined as invalid due to insufficient sample, or sample QC sequencing failure resulting in an invalid result or No Call for the variant. 295% 2-sided confidence interval calculated via the Wilson score method. | | | | | |

For $ALK$ fusions, 93 NSCLC samples were tested using the FISH-based orthogonal method. Of these, 67 samples (26 positive and 41 negative) had concordant results. One sample was discordant, showing a positive result with the ODxET but negative with the orthogonal method. Thirteen samples were positive with the ODxET but produced unknown results with the orthogonal method. Ten samples were negative with the ODxET but produced unknown results with the orthogonal method. One sample was positive with the orthogonal method but produced an unknown result with the ODxET, and one sample produced unknown results with both methods. Table 13 shows the $3 \times 3$ contingency table accounting for all 93 NSCLC $ALK$ fusion samples tested by both the ODxET and the orthogonal method, as well as the concordance analysis.

PMA P240040: FDA Summary of Safety and Effectiveness Data

{21}

Table 13. Concordance Between ODxET and Orthogonal Method for Detection of Level 2 ALK Fusions in NSCLC – Accuracy Study II

| | Orthogonal Method | | | | |
| --- | --- | --- | --- | --- | --- |
| | | Positive | Negative | Unknown1 | Total |
| ODxET | Positive | 26 | 1 | 13 | 40 |
| | Negative | 0 | 41 | 10 | 51 |
| | Unknown1 | 1 | 0 | 1 | 2 |
| | Total | 27 | 42 | 24 | 93 |
| Excluding Unknown | PPA% (n/N) [95% CIs]2 | 100.0% (26/26) [87.1%, 100.0%] | | | |
| | NPA% (n/N) [95% CIs]2 | 97.6% (41/42) [87.7%, 99.6%] | | | |
| Including Unknown | PPA% (n/N) [95% CIs]2 | 96.3% (26/27) [81.7%, 99.4%] | | | |
| | NPA% (n/N) [95% CIs]2 | 97.6% (41/42) [87.7%, 99.6%] | | | |
| 1 Unknowns are defined as invalid due to insufficient sample, or sample QC sequencing failure resulting in an invalid result or No Call for the variant. 2 95% 2-sided confidence interval calculated via the Wilson score method. | | | | | |

For RET fusions, 54 NSCLC and thyroid samples were tested using the NGS orthogonal method. Of these, 51 samples (17 positive and 34 negative) had concordant results. One sample was discordant, showing a positive result with the ODxET but negative with the orthogonal method. Two samples were negative with the ODxET but produced unknown results with the orthogonal method. Table 14 shows the $3 \times 3$ contingency table accounting for all 54 NSCLC and thyroid RET fusion samples tested by both the ODxET and the orthogonal method, along with the concordance analysis.

Table 14. Concordance Between ODxET and Orthogonal Method for Detection of Level 2 RET Fusions in NSCLC and Thyroid Cancer – Accuracy Study II

| | Orthogonal Method | | | | |
| --- | --- | --- | --- | --- | --- |
| | | Positive | Negative | Unknown1 | Total |
| ODxET | Positive | 17 | 1 | 0 | 18 |
| | Negative | 0 | 34 | 2 | 36 |
| | Unknown1 | 0 | 0 | 0 | 0 |
| | Total | 17 | 35 | 2 | 54 |
| Excluding Unknown | PPA% (n/N) [95% CIs]2 | 100.0% (17/17) [81.6%, 100.0%] | | | |
| | NPA% (n/N) [95% CIs]2 | 97.1% (34/35) [85.5%, 99.5%] | | | |
| Including Unknown | PPA% (n/N) [95% CIs]2 | 100.0% (17/17) [81.6%, 100.0%] | | | |
| | NPA% (n/N) [95% CIs]2 | 97.1% (34/35) [85.5%, 99.5%] | | | |
| 1 Unknowns are defined as invalid due to insufficient sample, or sample QC sequencing failure resulting in an invalid result or No Call for the variant. 2 95% 2-sided confidence interval calculated via the Wilson score method. | | | | | |

Concordance analysis results presented in Tables 12-14 demonstrate robust detection performance of ODxET for ERBB2 CNVs in breast cancer, ALK fusions in NSCLC, and RET fusions in NSCLC and thyroid cancer, all of which are classified as Level 2 according to the FDA's Biomarker Class Level for tumor profiling NGS tests.

PMA P240040: FDA Summary of Safety and Effectiveness Data

{22}

A summary of accuracy results (PPA and NPA) for ERBB2 CNV, ALK fusion, and RET fusion variants, stratified by the LoD level established by ODxET (i.e., at or above $1 \times$ LoD and below LoD), excluding unknown results, is presented in Table 15. For analysis conditional on the ODxET LoD levels, PPV was calculated for all reported variant types using the same stratification and exclusion criteria.

Table 15. Accuracy Results for Level 2 ERBB2 CNVs, ALK fusions, and RET Fusions Tumor Profiling Variants Stratified by LoD Level – Accuracy Study II

| Gene Alterations | LoD Level | ODxET (+) | Comparator (+) | ODxET (-) | Comparator (-) | ODxET (-) | Comparator (-) | OPDxET (-) | Comparator (-) | PPA (n/N) [95% CIs]1 | NPA (n/N) [95% CIs]1 | PPV (n/N) [95% CIs]1 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| ERBB2 CNV | All | 29 | 4 | 0 | 37 | 100.0% (29/29) [88.3%, 100.0%] | 90.2% (37/41) [77.5%, 96.1%] | 87.9% (29/33) [72.7%, 95.2%] |
| ≥ 1x LoD | 29 | 4 | 0 | 37 | 100.0% (29/29) [88.3%, 100.0%] | 90.2% (37/41) [77.5%, 96.1%] | 87.9% (29/33) [72.7%, 95.2%] |
| < 1x LoD | 0 | 0 | 0 | 0 | N/E | N/E | N/E |
| ALK Fusion | All | 26 | 1 | 0 | 41 | 100.0% (26/26) [87.1%, 100.0%] | 97.6% (41/42) [87.7%, 99.6%] | 96.3% (26/27) [81.7%, 99.3%] |
| ≥ 1x LoD (Molecular Counts) | 19 | 1 | 0 | 0 | 100.0% (19/19) [83.2%, 100.0%] | N/E | 95.0% (19/20) [76.4%, 99.1%] |
| < 1x LoD (Molecular Counts) | 7 | 0 | 0 | 41 | 100.0% (7/7) [64.6%, 100.0%] | N/E | 100.0% (7/7) [64.6%, 100.0%] |
| ≥ 1x LoD (Fusion Counts) | 22 | 1 | 0 | 0 | 100.0% (22/22) [85.1%, 100.0%] | N/E | 95.7% (22/23) [79.0%,99.2%] |
| < 1x LoD (Fusion Counts) | 4 | 0 | 0 | 41 | 100.0% (4/4) [51.0%, 100.0%] | N/E | 100.0% (4/4) [51.0%, 100.0%] |
| RET Fusion | All | 17 | 1 | 0 | 34 | 100.0% (17/17) [81.6%, 100.0%] | 97.1% (34/35) [85.5%, 99.5%] | 94.4% (17/18) [74.2%, 99.0%] |
| ≥ 1x LoD (Molecular Counts) | 17 | 1 | 0 | 34 | 100.0% (17/17) [81.6%, 100.0%] | 97.1% (34/35) [85.5%, 99.5%] | 94.4% (17/18) [74.2%, 99.0%] |
| < 1x LoD (Molecular Counts) | 0 | 0 | 0 | 0 | N/E | N/E | N/E |
| ≥ 1x LoD (Fusion Counts) | 17 | 1 | 0 | 34 | 100.0% (17/17) [81.6%, 100.0%] | 97.1% (34/35) [85.5%, 99.5%] | 94.4% (17/18) [74.2%, 99.0%] |
| < 1x LoD (Fusion Counts) | 0 | 0 | 0 | 0 | N/E | N/E | N/E |
| 195% 2-sided confidence interval calculated via the Wilson score method.*N/E: Not evaluable. The statistic cannot be calculated as there were no sample variant results in this category. |

PMA P240040: FDA Summary of Safety and Effectiveness Data

{23}

All four discordant ERBB2 CNV samples had tumor content levels between 20% and 39%. The CNV algorithm used by ODxET requires a minimum tumor content threshold of 40%. If a value below 40% is entered, the algorithm defaults to using 40% as the input. Entering a tumor content value higher than the actual level may deflate the copy number estimate, potentially leading to false negatives. Conversely, entering a lower value than the actual tumor content may inflate the estimate, potentially resulting in false positives. Given this threshold, the detection of ERBB2 CNVs in samples with 20%–39% tumor content may be unreliable, with an increased risk of inaccurate results, including false negatives. According to the ODxET User Guide, when tumor content is below 20% and tumor content within the region of interest (ROI) is at least 10% in resection samples, macrodissection is performed to enrich tumor content. However, this process does not mitigate the risk of inaccurate CNV results within the 20%–39% tumor content range. To address this issue, limitation language has been included in the device labeling to state that ERBB2 CNV calls may not be accurate when tumor content is between 20% and 39%.

## Accuracy Study III – Tumor Profiling Variants

The accuracy study III was conducted to evaluate the accuracy of ODxET for detecting EGFR exon 19 deletions, L858R, and T790M variants in NSCLC FFPE tissue samples by comparing results between ODxET and an orthogonal PCR method. All EGFR variants evaluated in this study were Level 2 variants, classified according to FDA’s three-tiered approach for reporting biomarkers in tumor profiling NGS tests. Samples were sourced from qualified commercial vendors and were pre-screened for target variants using an orthogonal method similar to ODxET.

Of the 70 samples tested, 68 showed concordance between ODxET and the orthogonal method. Among these, 23 were positive for the variants of interest (11 for EGFR L858R and T790M, and 12 for EGFR exon 19 deletions), and 45 were negative for EGFR variants. One sample was discordant, showing an EGFR-positive result with the ODxET but negative with the orthogonal method. One sample yielded an unknown result with the orthogonal PCR method.

Concordance analysis was performed at the sample level, and PPA and NPA estimates, along with their respective two-sided 95% CIs, were calculated using the orthogonal test results as the reference, excluding unknown results. The accuracy results (PPA and NPA), summarized by gene and further stratified by the FDA’s Biomarker Class Level for tumor profiling NGS tests and by predefined challenging genomic regions (e.g., homopolymers, SSRs, GC content), are presented in Table 16. All EGFR variants were tested at levels above 1× LoD.

PMA P240040: FDA Summary of Safety and Effectiveness Data
24 of 85

{24}

Table 16. Accuracy Results for EGFR exon 19 deletions, L858R, and T790M Tumor Profiling Variants – Accuracy Study III

| Variant Type | ODxET (+) | Comparator (+) | ODxET (+) | Comparator (-) | ODxET (-) | Comparator (+) | ODxET (-) | Comparator (-) | PPA (n/N) [95% CIs]¹ | NPA (n/N) [95% CIs]¹ |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| All EGFR Variants | 23 | 1 | 0 | 45 | 100.0% (23/23)
[85.7%, 100.0%] | 97.8% (45/46)
[88.7%, 99.6%] |
| Level 2 | 23 | 1 | 0 | 45 | 100.0% (23/23)
[85.7%, 100.0%] | 97.8% (45/46)
[88.7%, 99.6%] |
| Level 3 | 0 | 0 | 0 | 0 | N/E | N/E |
| EGFR SNVs | 11 | 1 | 0 | 45 | 100.0% (11/11)
[74.1%, 100.0%] | 97.8% (45/46)
[88.7%, 99.6%] |
| EGFR Deletions | 12 | 0 | 0 | 45 | 100.0% (12/12)
[75.8%, 100.0%] | 100.0% (45/45)
[92.1%, 100.0%] |
| EGFRs_non Homopolymers | 4 | 0 | – | – | 100.0% (4/4)
[51.0%, 100.0%] | N/E |
| EGFRs_Homopolymers | 19 | 1 | – | – | 100.0% (19/19)
[83.2%, 100.0%] | N/E |
| EGFRs_non SSRs | 19 | 1 | – | – | 100.0% (19/19)
[83.2%, 100.0%] | N/E |
| EGFRs_SSRs | 4 | 0 | – | – | 100.0% (4/4)
[51.0%, 100.0%] | N/E |
| EGFRs_GC [30%-40%] | 0 | 0 | – | – | N/E | N/E |
| EGFRs_GC [40%-50%] | 11 | 0 | – | – | 100.0% (11/11)
[74.1%, 100.0%] | N/E |
| EGFRs_GC [50%-60%] | 12 | 1 | – | – | 100.0% (12/12)
[75.8%, 100.0%] | N/E |
| EGFRs_GC [60%-75%] | 0 | 0 | – | – | N/E | N/E |
| ¹95% 2-sided confidence interval calculated via the Wilson score method.
*N/E: Not evaluable. The statistic cannot be calculated as there were no sample variant results in this category. |

The discordance observed between the ODxET and the orthogonal assays can be attributed to inherent design differences, including variant calling thresholds and rules between the two assays.

## 2. Analytical Sensitivity

### a. Limit of Blank

The limit of blank (LoB) study was conducted to estimate the false positive rate (FPR) for each variant class based on replicate testing of FFPE tumor samples known to be WT for all DNA and RNA variants detectable by ODxET. A total of 24 FFPE clinical samples from 10 different tissue types were used: bile duct (2), bladder (2), brain (4), breast (2), colon (2), endometrium (2), lung (4), pancreas (2), prostate (2), and skin (2). These samples were prescreened using an orthogonal method and served as blank

PMA P240040: FDA Summary of Safety and Effectiveness Data

{25}

controls for all reportable variants across all genes in the ODxET panel. The study was conducted using two reagent lots. For each tissue sample, four replicates were tested per reagent lot, resulting in a total of 192 measurements per DNA and RNA sample (4 replicates $\times$ 2 lots $\times$ 24 samples).

A total of 19 false positive calls were observed in the SNV variant type across 24 blank samples, resulting in an FPR of $0.006\%$ (19/315,072) for overall SNVs when calculated using the total number of SNV positions sequenced as the denominator. The per-position and per-sample FPRs for Level 3 SNVs across 24 blank samples representing all 10 tissue types were $0.0007\%$ (19/2,750,976) and $9.9\%$ (19/192), respectively (Table 17). The per-position FPR was calculated for each variant type based on the total number of sequenced SNV positions associated with clinically relevant variants across all tested sample replicates, while the per-sample FPR was calculated for each variant type using the total number of tested sample replicates as the denominator. All observed false positives were classified as Level 3 SNVs, with allele frequencies ranging from approximately $2.7\%$ to $5.2\%$ (mean: $3.5\%$ ). These values are below both the predicted LoD of the orthogonal assay used for prescreening, and the majority fall at or near the lower end of the established LoD of ODxET for SNVs ( $3.24\% - 7.34\%$ ), corresponding to approximately $0.56\times$ to $1.06\times$ relative to the ODxET LoD. These findings suggest that the observed false positives occurred below or near the assay's LoD for SNVs. Additionally, detection of these Level 3 SNVs exhibited tissue-specific patterns. The tissue-specific per-position FPR ranged from $0\%$ to $0.06\%$ , detailed as follows: $0\%$ for colon, endometrium, lung, and skin; $0.005\%$ for breast; $0.006\%$ for brain glioma; $0.009\%$ for bile duct, bladder, and pancreas; $0.02\%$ for pancreas; and $0.06\%$ for brain glioblastoma. The tissue-specific per-sample FPR ranged from $0\%$ to $75\%$ , with $0\%$ observed in colon, endometrium, lung, and skin; $6.25\%$ (1/16) in breast; $8.33\%$ (2/24) in brain glioma; $12.5\%$ (2/16) in bile duct, bladder, and pancreas; $25\%$ (4/16) in prostate; and $75\%$ (6/8) in brain glioblastoma. To mitigate the risk of reporting false positives for Level 3 SNVs, Level 3 SNVs with VAFs below $5\%$ were masked from reporting. After masking these 15 Level 3 SNVs from reporting, the overall FPR for SNVs was $0.0003\%$ (1/315,072) when calculated using the total number of SNV positions sequenced as the denominator. Additionally, after masking these 15 Level 3 SNVs from reporting, the per-position and per-sample FPRs for Level 3 SNVs across 24 blank samples representing all 10 tissue types remained $0.00004\%$ (1/2,750,976) and $0.52\%$ (1/192), respectively (Table 17).

Table 17. False Positive Rates Per Position and Per Sample Before and After Masking of Level 3 SNVs with VAFs Below $5\%$

| Category1 | Before Masking Certain Level 3 SNVs with VAFs Below 5% | | After Masking Certain Level 3 SNVs with VAFs Below 5% | |
| --- | --- | --- | --- | --- |
| | Per Position False Positive Rate2 | Per Sample False Positive Rate2 | Per Position False Positive Rate2 | Per Sample False Positive Rate2 |
| Level 1: EGFR exon 20 insertion (230) | 0% (0/(230×192)) | 0% (0/192) | 0% (0/(230×192)) | 0% (0/192) |

PMA P240040: FDA Summary of Safety and Effectiveness Data

{26}

| Category1 | Before Masking Certain Level 3 SNVs with VAFs Below 5% | | After Masking Certain Level 3 SNVs with VAFs Below 5% | |
| --- | --- | --- | --- | --- |
| | Per Position False Positive Rate2 | Per Sample False Positive Rate2 | Per Position False Positive Rate2 | Per Sample False Positive Rate2 |
| Level 2: Panel-wide SNVs (236) | 0% (0/(236×192)) | 0% (0/192) | 0% (0/(236×192)) | 0% (0/192) |
| Level 2: Panel-wide Indels (454) | 0% (0/(454×192)) | 0% (0/192) | 0% (0/(454×192)) | 0% (0/192) |
| Level 2: Panel-wide fusions and splice variants (3286) | 0% (0/(3286×192)) | 0% (0/192) | 0% (0/(3286×192)) | 0% (0/192) |
| Level 2: Panel-wide CNVs (4) | 0% (0/(4×192)) | 0% (0/192) | 0% (0/(4×192)) | 0% (0/192) |
| Level 3: Panel-wide SNVs (14328) | 0.0007% (19/(14328×192)) | 9.9% (19/192) | 0.00004% (1/(14328×192)) | 0.52% (1/192) |
| Level 3: Panel-wide Indels (8666) | 0% (0/(8666×192)) | 0% (0/192) | 0% (0/(8666×192)) | 0% (0/192) |
| Level 3: Panel-wide fusions and splice variants (7604) | 0% (0/(7604×192)) | 0% (0/192) | 0% (0/(7604×192)) | 0% (0/192) |
| Level 3: Panel-wide CNVs (105) | 0% (0/(105×192)) | 0% (0/192) | 0% (0/(105×192)) | 0% (0/192) |
| 1 The number in parentheses in the ‘Category’ column represents the sum of all variants across all tested cancer type included in the study for each variant category and reporting level.2 The FPR was calculated from 2 reagent lots (n=192 data points). | | | | |

A supplemental LoB study was conducted using twelve leftover tissue samples representing 10 different tissue types selected from the original LoB and tested using a different orthogonal assay. The orthogonal assay detected 44 DNA variants confirmed to be true positives, all present at or above the assay's LoD. None of these variants are targeted by ODxET, as they are not included in the ODxET panel. Among the 12 samples used in the supplemental study, ODxET did not detect any false positive reportable variants.

The results and software validation documentation from regression testing on the configuration with restricted reporting of certain Level 3 tumor profiling SNVs with variant allele frequencies below 5%, intended to prevent reporting of false positives, have not yet been provided. These results and documentation will be submitted in a post-market setting.

## b. Limit of Detection

### i. Limit of Detection for CDx Variants

The LoD of the ODxET for EGFR exon 20 insertion variants was determined using four patient-derived samples containing insertions of 3, 6, 9, and 12 bp. DNA from these samples was blended with DNA extracted from a WT NSCLC sample to create five or six variant allele frequency (VAF) levels per insertion type. Each blend level was tested using two reagent lots, with 10 replicates per lot, resulting in a total of 20 replicates per level. Detection rates across the blend levels were analyzed using probit regression to estimate the LoD, defined as the lowest VAF predicted to achieve a detection rate of at least 95%. Table 18 summarizes the LoD results for EGFR exon 20 insertions. The estimated LoD for EGFR exon 20 insertion variants ranged from 3.24% to 4.38% VAF (mean = 3.78%).

PMA P240040: FDA Summary of Safety and Effectiveness Data

{27}

Table 18. Summary of LoD Estimates for EGFR Exon 20 Insertion Variants

| Gene | Exon 20 Insertion Size | Tumor Type | COSMIC ID | Insertion Variant | LoD (% VAF) |
| --- | --- | --- | --- | --- | --- |
| EGFR | 3 bp | NSCLC | COSM12377 | c.2316_2317insGTT | 4.38 |
| EGFR | 6 bp | NSCLC | COSM1238028 | c.2320_2321insCCCACG | 4.16 |
| EGFR | 9 bp | NSCLC | COSM12376 | c.2308_2309insCCAGCGTGG | 3.24 |
| EGFR | 12 bp | NSCLC | COSM26720 | c.2290_2291insTCCAGGAAGCCT | 3.32 |

# ii. Limit of Detection for Tumor Profiling Variants

The LoD of the ODxET was determined for SNVs, insertions, deletions, CNVs, fusions, and splice variants. The LoD for SNVs, insertions, and deletions was established using a representative variant approach, while the LoD for CNVs, fusions, and splice variants was assessed at the gene level. The variant set evaluated comprised 19 SNVs, 14 insertion and deletion variants, 15 CNVs, 26 fusions and exon-skipping variants, and 4 RNA imbalance variants. DNA and RNA blends were prepared using clinical specimens representing ten cancer types: bladder, breast, cholangiocarcinoma, colorectal, endometrial, glioma, non-small cell lung, melanoma, pancreatic, and thyroid cancers. For each variant type, at least 120 data points were generated by testing six blend levels using two reagent lots, with 10 replicates per level per lot (10 replicates per reagent lot $\times$ 2 lots $\times$ 6 blend levels). LoD values were calculated using probit regression based on the detection rates across blend levels. The LoD was defined as the lowest VAF or copy number predicted to yield a $\geq 95\%$ detection rate. If the Probit model did not provide a reasonable fit to the data, the empirical hit rate method was used instead, in which the LoD was defined as the lowest level with an observed detection rate of at least $95\%$ .

The estimated LoDs for SNVs ranged from $3.24\%$ to $7.34\%$ VAF, with a mean of $5.29\%$ . For insertions, LoDs ranged from $2.98\%$ to $5.24\%$ VAF (mean $= 4.11\%$ ), and for deletions, from $3.08\%$ to $4.67\%$ VAF (mean $= 3.88\%$ ). CNVs had LoDs ranging from 4.56 to 5.78 copies, with a mean of 4.99 copies. A summary of the LoD results for tumor profiling DNA variants is presented in Table 19.

Table 19. LoD Summary for Tumor Profiling DNA Variants

| Variant Type | Gene | Variant ID | COSMIC ID | Cancer Type | LoD (%VAF or CN) |
| --- | --- | --- | --- | --- | --- |
| SNV | ERBB2 | ERBB2 G776V | COSM18609 | NSCLC | 3.24 |
| SNV | IDH1 | IDH1 R132C | COSM28747 | CRC | 4.94 |
| SNV | BRAF | BRAF V600E | COSM476 | NSCLC | 4.37 |
| SNV | BRAF | BRAF V600E | COSM476 | Thyroid | 3.68 |
| SNV | ERBB2 | ERBB2 SNV | COSM48358 | Bladder | 3.56 |
| SNV | KRAS | KRAS G12C | COSM516 | CRC | 4.43 |

PMA P240040: FDA Summary of Safety and Effectiveness Data

{28}

| Variant Type | Gene | Variant ID | COSMIC ID | Cancer Type | LoD (%VAF or CN) |
| --- | --- | --- | --- | --- | --- |
| SNV | KRAS | KRAS G12C | COSM516 | Pancreatic | 4.6 |
| SNV | KRAS | KRAS G12C | COSM516 | NSCLC | 3.81 |
| SNV | MET | MET | COSM6108462 | NSCLC | 4.45 |
| SNV | EGFR | EGFR L858R | COSM6224 | NSCLC | 5.44 |
| SNV | EGFR | EGFR T790M | COSM6240 | NSCLC | 4.75 |
| SNV | FGFR3 | FGFR3 R248C | COSM714 | Bladder | 7.03 |
| SNV | FGFR3 | FGFR3 S249C | COSM715 | Bladder | 4.4 |
| SNV | FGFR3 | FGFR3 G370C | COSM716 | Bladder | 5.21 |
| SNV | PIK3CA | PIK3CA C420R | COSM757 | Breast | 7.34 |
| SNV | PIK3CA | PIK3CA E545K | COSM763 | Breast | 5.69 |
| SNV | PIK3CA | PIK3CA H1047R | COSM775 | Breast | 4 |
| SNV | RET | RET C618Y | COSM980 | Thyroid | 5.91 |
| MNV | BRAF | BRAF V600K | COSM473 | Melanoma | 5.37 |
| Insertion | ERBB2 | ERBB2 exon 20 ins (3bp) | COSM12553 | NSCLC | 2.98 |
| Insertion | ERBB2 | ERBB2 Exon 20 Ins (9bp) | COSM12556 | NSCLC | 4.07 |
| Insertion | ERBB2 | ERBB2 Exon 20 ins (12bp) | COSM12558 | NSCLC | 3.87 |
| Insertion | EGFR | EGFR Exon 20 Ins (9bp) | COSM13428 | Endometrial | 4.27 |
| Insertion | ERBB2 | ERBB2 Exon 20 Ins (12 bp) | COSM20959 | Breast | 5.24 |
| Insertion | ERBB2 | ERBB2 Exon 20 Ins (6bp) | COSM303939 | NSCLC | 3.91 |
| Deletion | EGFR | EGFR Exon 19 del (18bp) | COSM12370 | NSCLC | 4.15 |
| Deletion | EGFR | EGFR Exon 19 Del (9bp) | COSM12382 | NSCLC | 4.19 |
| Deletion | EGFR | EGFR Exon 19 del (3bp) | COSM18420 | NSCLC | 3.5 |
| Deletion | PIK3CA | PIK3CA E110del | COSM24710 | CRC | 4.03 |
| Deletion | PIK3CA | PIK3CA E110del | COSM24710 | Bladder | 3.46 |
| Deletion | TP53 | TP53 Del 18 bp | COSM43570 | Breast | 3.14 |
| Deletion | EGFR | EGFR Exon 19 Del (15bp) | COSM6223 | NSCLC | 3.08 |
| Deletion | RET | RET D898 E901del | COSM962 | Thyroid | 4.67 |
| CNV | ERBB2 | ERBB2 CNV* | N/A | Bladder | 4.84 |
| CNV | ERBB2 | ERBB2 CNV* | N/A | CRC | 4.69 |
| CNV | ERBB2 | ERBB2 CNV* | N/A | Breast | 5.78 |
| CNV | ERBB3 | ERBB3 CNV* | N/A | Bladder | 4.91 |
| CNV | AR | AR CNV* | N/A | Breast | 5.78 |
| CNV | EGFR | EGFR CNV* | N/A | Glioma | 4.9 |
| CNV | FGFR1 | FGFR1 CNV* | N/A | Breast | 5.36 |
| CNV | FGFR2 | FGFR2 CNV* | N/A | Breast | 4.56 |
| CNV | FGFR3 | FGFR3 CNV* | N/A | Endometrial | 4.63 |
| CNV | KRAS | KRAS CNV* | N/A | NSCLC | 5.49 |
| CNV | MET | MET CNV* | N/A | NSCLC | 5.23 |
| CNV | MET | MET CNV* | N/A | Glioma | 5.09 |
| CNV | MET | MET CNV* | N/A | Melanoma | 4.76 |
| CNV | PIK3CA | PIK3CA CNV* | N/A | Bladder | 5.09 |
| CNV | PIK3CA | PIK3CA CNV* | N/A | NSCLC | 4.75 |
| *These CNV variants failed the probit regression fit, hence the LoD was estimated using the empirical hit-rate method. | | | | | |

PMA P240040: FDA Summary of Safety and Effectiveness Data

{29}

The LoDs for the evaluated RNA fusions and splice variants ranged from 5.0 to 15.5 molecular counts (mean = 8.68), as shown in Table 20. Four fusions assessed using imbalance scores based on unequal expression of the two alleles of gene demonstrated LoDs ranging from 1.5 to 4.93 (mean = 2.75), as presented in Table 21. The LoD estimates summarized in Table 20 and Table 21 were determined using either the probit regression method or the empirical hit-rate method, depending on the quality of model fit.

Table 20. LoD Summary for Tumor Profiling RNA Variants

| Variant Type | Gene | Variant ID | Cancer Type | LoD (Molecular Count) | LoD (Read Count) |
| --- | --- | --- | --- | --- | --- |
| Fusion | NTRK2 | BCR-NTRK2.B1N17 | Glioma | 5.93 | 46.67 |
| Fusion | RET | CCDC6-RET.C1R12.COSF1271.1 | NSCLC | 10.14 | 122.16 |
| Fusion | RET | CCDC6-RET.C1R12.COSF1271.1 | Thyroid | 8.31 | 112.8 |
| Fusion | NRG1 | CD74-NRG1.C6N6.1 | NSCLC | 5.11 | 101.67 |
| Fusion | RSPO2 | EIF3E-RSPO2.E1R2.COSF1307.1 | CRC | 5.52 | 126.17 |
| Fusion | ALK | EML4-ALK.E13A20.COSF408.2 | NSCLC | 7.78 | 69.55 |
| Fusion | ALK | EML4-ALK.E20A20.COSF409.2 | NSCLC | 5 | 55.95 |
| Fusion | ESR1 | ESR1-CCDC170.E2C7.1 | Breast | 15.5 | 251.6 |
| Fusion | ROS1 | EZR-ROS1.E10R34.COSF1267 | NSCLC | 8.92 | 73.29 |
| Fusion | FGFR1 | FGFR1-TACC1.F17T7.COSF1362.1 | NSCLC | 6.82 | 61.19 |
| Fusion | FGFR2 | FGFR2-CREB5.F17C8.1 | Cholangiocarcinoma | 7.83 | 99.29 |
| Fusion | FGFR2 | FGFR2-TACC2.F17T11 | NSCLC | 5.75 | 108.84 |
| Fusion | FGFR3 | FGFR3-TACC3.F17T11.COSF1348.1 | Bladder | 12.65 | 249.33 |
| Fusion | BRAF | KIAA1549-BRAF.K15B9.COSF481.2 | Glioma | 7.11 | 66.74 |
| Fusion | RET | KIF5B-RET.K15R12.COSF1232.1 | NSCLC | 8.59 | 108.9 |
| Fusion | RSPO3 | PTPRK-RSPO3.P1R2.COSF1311.1 | CRC | 8.93 | 171.77 |
| Fusion | MET | PTPRZ1-MET.P1M2.1 | Glioma | 6.57 | 56.61 |
| Fusion | ROS1 | SDC4-ROS1.S2R32.COSF1265 | NSCLC | 7.76 | 87.72 |
| Fusion | NTRK3 | SQSTM1-NTRK3.S5N14 | Thyroid | 7.61 | 118.93 |
| Fusion | NTRK1 | TPM3-NTRK1.T7N10.COSF1329 | CRC | 7.76 | 97.89 |
| Fusion | NTRK1 | TPM3-NTRK1.T7N10.COSF1329 | Thyroid | 8.13 | 115.04 |
| Fusion | NUTM1 | WHSC1L1-NUTM1.W7N2.1 | Thyroid | 8.93 | 164.87 |
| Alt Splice Variant | AR | AR-AR.A3A4.V7 | Prostate | 11.28 | 159.26 |
| Alt Splice Variant | EGFR | EGFRvIII.E1E8.Del1 | Glioma | 11 | 121.67 |
| Alt Splice Variant | MET | MET-MET.M13M15.1 | NSCLC | 13.63 | 114.7 |

PMA P240040: FDA Summary of Safety and Effectiveness Data

{30}

PMA P240040: FDA Summary of Safety and Effectiveness Data

| Variant Type | Gene | Variant ID | Cancer Type | LoD (Molecular Count) | LoD (Read Count) |
| --- | --- | --- | --- | --- | --- |
| Alt Splice Variant | MET | MET-MET.M13M15.1 | NSCLC | 10.78 | 54.78 |

Table 21. LoD Summary for Tumor Profiling Imbalance Variants

| Variant Type | Gene | Variant ID | Cancer Type | LoD (Imbalance Score) |
| --- | --- | --- | --- | --- |
| Imbalance | FGFR3 | FGFR3 | Bladder | 1.5 |
| Imbalance | RET | RET | Thyroid | 2.1 |
| Imbalance | NTRK1 | TFG-NTRK1.T6N10.1* | Thyroid | 2.45 |
| Imbalance | ALK | EML4-ALK.E13A20.COSF408.2* | NSCLC | 4.93 |
| *These Imbalance variants failed the probit regression fit, hence, the LoD was estimated using the empirical hit-rate method. | | | | |

## c. Nucleic Acid Input

### i. DNA and RNA Input for CDx Variants

To verify the performance of ODxET at the product requirement of 10 ng nucleic acid input and to establish the acceptable range of DNA input for accurate detection of EGFR exon 20 insertion variants, five DNA-RNA input combinations were evaluated (5 ng, 7.5 ng, 10 ng [standard input], 12.5 ng, and 15 ng). Each combination was tested as a single sample to assess the impact of varying RNA quantities relative to DNA input. The study used DNA extracted from four FFPE NSCLC patient samples containing EGFR exon 20 insertions of 3, 6, 9, and 12 bp, each blended with WT FFPE NSCLC DNA to achieve VAFs at 1–3× the LoD.

Analysis of variance (ANOVA) was performed to evaluate the relationship between input levels and VAF. No statistically significant differences in VAF were observed across the different input levels. The PPA for all DNA reference locations was 100% across all input levels. lowest observed NPA was 99.94% excluding no-calls, and 97.39% including no-calls, across all DNA input levels. These results support the performance of ODxET at the recommended input of 10 ng and confirm its robustness across a range of DNA and RNA input levels for the accurate detection of EGFR exon 20 insertion variants.

### ii. DNA and RNA Input for Tumor Profiling Variants

The recommended input level for FFPE DNA and RNA samples is 10 ng for detection of tumor profiling variants. To assess the impact of nucleic acid input amounts on ODxET performance, DNA and RNA extracted from clinical FFPE samples were tested at three input levels (5 ng, 10 ng [standard input], and 15 ng) for tumor profiling variants, and at five input levels (5 ng, 7.5 ng, 10 ng [standard input], 12.5 ng, and 15 ng) for

31 of 85

{31}

selected tumor profiling variants. These selected variants included NSCLC ERBB2 exon 20 insertion, NSCLC EGFR exon 19 deletion, EGFR L858R and T790M, BRAF V600E, KRAS G12C, glioma IDH1 and IDH2 SNVs, NSCLC ALK, RET, and ROS1 fusions, and NSCLC MET exon 14 skipping variants.

The FFPE specimens included variants across 25 genes and represented 11 cancer types, including bladder, breast, cholangiocarcinoma, colorectal, endometrial, glioma, melanoma, non-small cell lung, pancreatic, prostate, and thyroid cancers. The analysis included a range of DNA and RNA variant types, including SNVs, MNVs, deletions, insertions, CNVs, RNA fusions, and splice variants. A total of 37 DNA blends were prepared to achieve VAFs at 1--3× the LoD, and 34 RNA blends were prepared to achieve read counts at 1--3× the LoD for fusions and splice variants. Each blend was tested at each input level using either 6 or 12 replicates per variant to assess the effect of input amount on positive and negative call rates.

The results demonstrated that the PPA for SNVs, insertions, deletions, and CNVs was 100% across all DNA input levels. The NPA, including no calls at negative variant locations, was ≥99.64% across all DNA input levels, with the exception of one ERBB2 exon 20 insertion variant, which showed an NPA of 97.39% including no calls and 99.95% excluding no calls at the 7.5 ng input level. For all other DNA blends, the NPA excluding no calls was ≥99.84% at all input levels. For RNA fusions and splice variants evaluated, the PPA was 100% across all input levels. The NPA including no calls, was ≥99.26%, and the NPA excluding no calls was ≥99.90% across all input levels. These results confirm the performance of ODxET at the recommended 10 ng input level for accurate detection of tumor profiling variants.

The minimum nucleic acid concentration required for the ODxET is 0.5 ng/μL. To verify ODxET performance at the minimum concentration for detecting both CDx and tumor profiling variants, variant-positive DNA and RNA samples were tested in six replicates under three conditions: low concentration (Level 1, 0.25 ng/μL), standard concentration (Level 2, 0.50 ng/μL), and high concentration (Level 3, 0.75 ng/μL). Summary statistics were generated for each concentration level, and equivalence across levels was assessed using Welch's ANOVA test at a significance level of α = 0.05. Among 11 CNVs, FGFR1 CNV, and among 10 RNA fusion variants, six fusions showed no statistical equivalence in summary statistics across concentration levels. Variant call concordance data (PPA and NPA) were not assessed in this study.

{32}

d. Tissue Input

This study was performed to verify that FFPE sections cut from resections, core needle biopsies (CNBs), and cell blocks from fine needle aspirate (FNA) samples can yield the minimum nuclei…