Oncomine Dx Target Test

P160045 · Life Technologies Corporation · PQP · Jun 22, 2017 · Pathology

Device Facts

• Record ID: P160045
• Device Name: Oncomine Dx Target Test
• Applicant: Life Technologies Corporation
• Product Code: PQP · Pathology
• Decision Date: Jun 22, 2017
• Decision: APPR
• Device Class: Class 3

Intended Use

The Oncomine™ Dx Target Test is a qualitative in vitro diagnostic test that uses targeted high throughput, parallel-sequencing technology to detect single nucleotide variants (SNVs) and deletions in 23 genes from DNA and fusions in ROS1 from RNA isolated from formalin-fixed, paraffin-embedded (FFPE) tumor tissue samples from patients with non-small cell lung cancer (NSCLC) using the Ion PGM™ Dx System. The test is indicated to aid in selecting NSCLC patients for treatment with the targeted therapies listed in Table 1 in accordance with the approved therapeutic product labeling. Safe and effective use has not been established for selecting therapies using this device for the variants in Table 1 in tissue types other than NSCLC. Results other than those listed in Table 1 are indicated for use only in patients who have already been considered for all appropriate therapies (including those listed in Table 1). Analytical performance using NSCLC specimens has been established for the variants listed in Table 2. The test is not indicated to be used for standalone diagnostic purposes, screening, monitoring, risk assessment, or prognosis.

Device Story

The Oncomine™ Dx Target Test is an NGS-based IVD for detecting somatic variants in FFPE NSCLC tumor tissue. Input: DNA and RNA isolated from FFPE samples using the Ion Torrent™ Dx FFPE Sample Preparation Kit. Process: Nucleic acids are quantified, converted to cDNA (RNA only), and amplified using the Oncomine™ Dx Target Test DNA and RNA Panel. Libraries are prepared with specific barcodes, pooled, and sequenced on the Ion PGM™ Dx System. Data analysis is performed by Torrent Suite™ Dx software on the Ion Torrent™ Server, which maps reads to the hg19 reference genome and candidate fusion sequences to identify SNVs, deletions, and ROS1 fusions. Output: Variant calls used by clinicians to select targeted therapies (dabrafenib/trametinib, crizotinib, or gefitinib). Used in clinical laboratories by trained personnel. Results inform therapeutic selection for NSCLC patients, potentially improving clinical outcomes by matching patients to appropriate targeted therapies.

Clinical Evidence

Evidence based on retrospective analysis of clinical trial cohorts (BRF113928 for BRAF, Study 1001 for ROS1) and a concordance study for EGFR. BRAF V600E cohort (n=87) showed ORR of 72.7% (Cohort B) and 60.9% (Cohort C) in CDx(+) patients. ROS1 cohort (n=47) showed 83.3% ORR in CDx(+) patients. EGFR concordance study (n=225) demonstrated high agreement with QIAGEN therascreen® EGFR RGQ PCR Kit (PPA 98.6%, NPA 99.2% excluding no calls). Analytical performance established via LoD, precision, and interference studies.

Technological Characteristics

Targeted high-throughput parallel sequencing (NGS). Components: DNA/RNA panels, controls, FFPE extraction kit, Ion PGM™ Dx System (sequencer, server, software). Materials: Silica-based filter membranes for extraction. Connectivity: Networked server-based analysis. Software: Torrent Suite™ Dx (ADF-based). Sterilization: Not applicable (reagents).

Indications for Use

Indicated for NSCLC patients to aid in selecting treatment with dabrafenib/trametinib (BRAF V600E), crizotinib (ROS1 fusions), or gefitinib (EGFR L858R/Exon 19 deletions). Not for standalone diagnosis, screening, monitoring, risk assessment, or prognosis.

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

• QIAGEN therascreen® RGQ PCR Test
• BRAF V600 PCR Mutation Test
• ROS1 FISH test

Submission Summary (Full Text)

{0}

PMA P160045: FDA Summary of Safety and Effectiveness Data
Page 1

# 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 Target Test

Device Procode: PQP

Applicant's Name and Address: Life Technologies Corporation
5781 Van Allen Way
Carlsbad, CA 92008

Date(s) of Panel Recommendation: None

Premarket Approval Application (PMA) Number: P160045

Date of FDA Notice of Approval: June 22, 2017

Priority Review: Granted priority review status on November 8, 2016, because the device addresses an unmet medical need, as the test is based on breakthrough technology and is in the best interest of patients.

## II. INDICATIONS FOR USE

The Oncomine™ Dx Target Test is a qualitative *in vitro* diagnostic test that uses targeted high throughput, parallel-sequencing technology to detect single nucleotide variants (SNVs) and deletions in 23 genes from DNA and fusions in ROS1 from RNA isolated from formalin-fixed, paraffin-embedded (FFPE) tumor tissue samples from patients with non-small cell lung cancer (NSCLC) using the Ion PGM™ Dx System.

The test is indicated to aid in selecting NSCLC patients for treatment with the targeted therapies listed in Table 1 in accordance with the approved therapeutic product labeling.

Table 1. List of variants for therapeutic use

| Gene | Variant | Targeted therapy |
| --- | --- | --- |
| BRAF | BRAF V600E | TAFINLAR® (dabrafenib) in combination with MEKINIST® (trametinib) |
| ROS1 | ROS1 fusions | XALKORI® (crizotinib) |
| EGFR | L858R, Exon 19 deletions | IRESSA® (gefitinib) |

Safe and effective use has not been established for selecting therapies using this device for the variants in Table 1 in tissue types other than NSCLC.

{1}

Results other than those listed in Table 1 are indicated for use only in patients who have already been considered for all appropriate therapies (including those listed in Table 1).

Analytical performance using NSCLC specimens has been established for the variants listed in Table 2.

Table 2. List of variants with established analytical performance only
| Gene | Variant ID | Nucleotide change |
| --- | --- | --- |
| KRAS | COSM512 | c.34_35delGGinsTT |
| KRAS | COSM516 | c.34G>T |
| MET | COSM707 | c.3029C>T |
| PIK3CA | COSM754 | c.1035T>A |

The test is not indicated to be used for standalone diagnostic purposes, screening, monitoring, risk assessment, or prognosis.

## III. CONTRAINDICATIONS

There are no known contraindications.

## IV. WARNINGS AND PRECAUTIONS

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

## V. DEVICE DESCRIPTION

The Oncomine™ Dx Target Test is an 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 PGM Dx System and the Ion Torrent PGM Dx Reagents for detection of alterations in DNA and RNA isolated from non-small cell lung cancer (NSCLC) formalin-fixed, paraffin-embedded (FFPE) tumor specimens.

The Oncomine™ Dx Target Test consists of the following:

Oncomine™ Dx Target Test and Controls Kit (Combo Kit):
- Oncomine™ Dx Target Test DNA and RNA Panel
- Oncomine™ Dx Target DNA Control Kit
- Oncomine™ Dx Target RNA Control Kit
- Ion Torrent™ Dx No Template Control Kit

Ion Torrent™ Dx FFPE Sample Preparation Kit:
- Ion Torrent™ Dx Total Nucleic Acid Isolation Kit
- Ion Torrent™ Dx cDNA Synthesis Kit
- Ion Torrent™ Dx DNA Quantification Kit
- Ion Torrent™ Dx RNA Quantification Kit
- Ion Torrent™ Dx Dilution Buffer Kit

PMA P160045: FDA Summary of Safety and Effectiveness Data

{2}

Ion Torrent™ PGM™ Dx Reagents / Chips:
- Ion PGM™ Dx Library Kit
- Ion OneTouch™ Dx Template Kit
- Ion PGM™ Dx Sequencing Kit
- Ion 318™ Dx Chip Kit

Instrumentation and Software:
- The assay is run on the Ion Torrent™ PGM™ Dx System:
- Ion OneTouch™ Dx System:
- Ion OneTouch™ Dx Instrument
- Ion OneTouch™ ES Dx Instrument
- Ion PGM™ Dx Sequencer
- Ion PGM™ Dx Chip Minifuge
- Ion Torrent™ Server
- Torrent Suite™ Dx Software
- Other accessories:
- Ion PGM™ Wireless Scanner
- DynaMag™ 16 2mL Dx Magnet
- DynaMag™ 96 Side Dx Magnet

The system also utilizes specified accessories. The assay’s definition files are provided on a USB memory device along with the Oncomine™ Dx Target Test User Guides and ADF on a USB Memory Device:
- Oncomine™ Dx Target Assay Definition File (includes interpretive software)
- Oncomine™ Dx Target Test User Guide
- Veriti™ Dx Thermal Cycler Settings
- Electronic Document Instructions (provided to users both as a paper copy and a PDF document on the USB drive)

Nucleic Acid Extraction:
DNA and RNA extraction is performed using the proprietary Ion Torrent™ Dx FFPE Sample Preparation Kit. The deparaffinized sample is first subjected to protein digestion with Proteinase K at an elevated temperature in a guanidinium thiocyanate solution to facilitate release and protection of RNA and DNA by inhibiting nuclease activity. After a heating step to inactive the Proteinase K enzyme, the digested sample is transferred into a spin column containing a silica-based filter membrane.

The RNA is selectively eluted and separated from DNA which is retained on the filter. The eluted RNA is mixed with ethanol and captured onto a second spin column containing a silica-based membrane filter. The RNA is retained and cellular impurities are removed by a series of washes. The bound RNA is treated with DNase to reduce contaminating DNA. Following a series of washes to remove residual DNase and DNA degradation products, the purified RNA is eluted from the filter.

PMA P160045: FDA Summary of Safety and Effectiveness Data
Page 3

{3}

The DNA retained on the first filter is similarly subjected to a series of washes to remove cellular impurities and then purified DNA is eluted from the filter. The Elution Solution provided with the kit is a low ionic strength Tris-buffered solution containing EDTA that facilitates elution of nucleic acids from the silica filter. The solution provides appropriate pH for stability of RNA and DNA and inhibits nucleases by binding metal cofactors.

## Quantification:

RNA and DNA quantification is performed using a fluorescence dye-binding assay and a qualified fluorometer/fluorescence reader capable of operating at the specific excitation and emission wavelengths. First, working solutions consisting of buffer and proprietary fluorophores are prepared for both DNA and RNA samples, as well as the DNA and RNA standards supplied at different concentrations in the kit (0 ng/μL to 10 ng/μL). Second, the DNA and RNA samples are incubated with their respective solutions at room temperature where the fluorophores bind to the target DNA and RNA molecules. When bound to the DNA and RNA, the fluorophores exhibit fluorescence enhancement at a specific excitation wavelength. The emitted fluorescent signals are captured and converted into signal fluorescence units. Third, the concentration (in ng/μL) of the DNA and RNA samples are determined by performing a linear regression with the values obtained from the DNA and RNA standards.

Sample Dilution Buffer is provided in the kit to dilute the DNA and RNA samples to a specific concentration required for cDNA synthesis and library preparation.

## RT Step (RNA only):

RNA is enzymatically converted to cDNA using the Ion Torrent™ Dx cDNA Synthesis Kit. Ten nanograms (ng) of RNA is enzymatically converted to cDNA using an enzyme mix containing a proprietary engineered version of M-MLV reverse transcriptase (Superscript III RT), an RNase inhibitor, a proprietary helper protein, and a buffer containing random primers, dNTPs, and MgCl₂.

## Library Preparation workflow:

The process begins with polymerase chain reaction (PCR) and uses the Oncomine™ Dx Target Test DNA and RNA Panel and the Ion PGM™ Dx Library Kit to specifically amplify target regions of interest from cDNA (including cDNA from the RNA control) and DNA (including the DNA Control and No Template Control).

Two different libraries are generated and pooled for each sample; one for DNA targets and one for RNA targets. During library preparation for each sample, one of the 16 oligonucleotide barcodes in the Library Kit is used for the DNA-derived library and another oligonucleotide barcode is used for the RNA-derived library. This ensures the correct identification of each respective portion of the assay (DNA and RNA) from each patient sample. After library preparation, the DNA and RNA libraries for all samples and controls may be blended for the templating reaction.

## Data Analysis:

PMA P160045: FDA Summary of Safety and Effectiveness Data
Page 4

{4}

This process is executed by the Torrent Suite™ Dx software, which runs on the Ion Torrent™ Server. Together, these manage the complete end-to-end workflow from sample to variant call. The DNA reads are 'mapped' to the reference human genome (hg19) followed by detection of single nucleotide variants (SNV) and deletions (del) using a reference hotspot file. The RNA reads are 'mapped' to a reference containing control sequences and candidate gene fusion sequences. Gene fusions are detected as present if they map to these reference sequences and pass certain filtering criteria provided by the Oncomine™ Dx Target ADF.

## VI. ALTERNATIVE PRACTICES AND PROCEDURES

There are no FDA-cleared or -approved alternatives for BRAF mutation testing of formalin-fixed, paraffin-embedded (FFPE) NSCLC tissue for the selection of patients who are eligible for treatment with TAFINLAR® (dabrafenib) in combination with MEKINIST® (trametinib). There are no FDA-cleared or -approved alternatives for ROS1 translocation testing of FFPE NSCLC tissue for the selection of patients who are eligible for treatment with XALKORI® (crizotinib). There is one FDA-approved alternative, the QIAGEN therascreen® RGQ PCR Test, for EGFR (Exon 19 deletions and L858R mutation) testing of FFPE NSCLC tissue for the selection of patients who are eligible for treatment with IRESSA® (gefitinib). Each alternative has its own advantages and disadvantages. A patient should fully discuss any alternative with his/her physician to select the most appropriate method.

## VII. MARKETING HISTORY

The Oncomine™ Dx Target Test has not been marketed in the United States or any foreign country.

## 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 Oncomine™ Dx Target Test results and subsequently improper patient management decisions in NSCLC treatment. There is also a risk of delayed results, which may lead to delay of treatment with the appropriate targeted therapy. No adverse events were reported in connection with the clinical studies used to support this PMA as the studies were performed retrospectively using banked samples.

## IX. SUMMARY OF NONCLINICAL STUDIES

### A. Laboratory Studies

1. Analytical Accuracy

Concordance of the Oncomine™ Dx Target Test was assessed for the ability to detect SNVs, deletions, and fusions with two externally validated comparator methods. Four hundred twenty-two (422) of the 436 SNVs and deletion hotspot variants that could be detected by the Oncomine™ Dx Target Test were included

PMA P160045: FDA Summary of Safety and Effectiveness Data
Page 5

{5}

in an independently developed and validated NGS assay for SNVs and deletions. In addition, a validated ROS1 FISH test, consistent with the test used in the A8081001 clinical ROS1 study, was used as the comparator method to detect ROS1 fusions. Two additional comparator methods were included in the clinical concordance studies for BRAF V600E and EGFR for Exon 19 deletions (Ex. 19del) and L858R variants.

A bank of 566 uncharacterized FFPE tumor samples were procured and sent to the reference testing labs to be screened and enrolled into the study on a rolling basis until a minimum of 275 samples were identified per the study protocol.

A total of 461 samples were tested by the validated NGS comparator method with 291 samples (63.1%) yielding valid results. Eight (8) of the 291 positive samples were invalid for the ROS1 FISH test but were included in the analysis for the other valid variants and 1 of the 291 sample was not tested by the ROS1 FISH test. Among the 290 samples, 178 (61.4%) were identified as being positive for at least one alteration. The samples included the following variant types: simple SNVs, complex SNVs [SNVs in di-/tri-nucleotide repeat regions and multinucleotide variant (MNV)], and deletions that are targeted by the Oncomine™ Dx Target Test.

For ROS1, a total of 469 samples were tested with the validated ROS1 FISH comparator method and 392/469 (83.6%) samples generated valid results, while 77/469 (16.4%) samples failed the comparator test due to poor hybridization of the probes.

All 290 samples were then tested with the Oncomine™ Dx Target Test. One of the 290 samples (positive by comparator method) failed both DNA and RNA sample QC metrics. In addition, 22 of the 290 RNA samples failed to meet the RNA sample QC metrics and were considered invalid for fusion variant analysis.

The results at the variant, bin and sample levels are shown in the tables below. The tables and calculations below do not include the samples from the primary clinical studies (see individual clinical studies under section "X. Summary of Primary Clinical Studies", below); however the concurrence data from the clinical studies is also considered as a demonstration of analytical accuracy.

Table 3. Overall agreement results at bin level.

| Level | Agreement | No Calls Included | | | No Calls Excluded | | |
| --- | --- | --- | --- | --- | --- | --- | --- |
| | | # Comp1 | # ODx2 | % (95% CI) | # Comp | # ODx | % (95% CI) |
| Variant | PPA3 | 198 | 195 | 98.50% (95.6%, 99.7%) | 198 | 195 | 98.50% (95.6%, 99.7%) |
| | NPA4 | 122,012 | 118,155 | 96.80% (96.7%, 96.9%) | 118,159 | 118,155 | 99.997% (99.991%, 99.999%) |
| Bin | PPA | 181 | 176 | 97.20% | 181 | 176 | 97.20% |

PMA P160045: FDA Summary of Safety and Effectiveness Data

{6}

Table 4. Agreement results by bin type.

| Bin | Agreement | No Calls Included | | | No Calls Excluded | | |
| --- | --- | --- | --- | --- | --- | --- | --- |
| | | # Comp | # ODx | % (95% CI) | # Comp | # ODx | % (95% CI) |
| Simple SNV | PPA | 83 | 82 | 98.80% (93.47%, 99.97%) | 83 | 82 | 98.80% (93.47%, 99.97%) |
| | NPA | 200 | 65 | 32.50% (26.06%, 39.47%) | 206 | 206 | 100.00% (98.23%, 100.00%) |
| Complex SNV | PPA | 85 | 83 | 97.65% (91.76%, 99.71%) | 85 | 83 | 97.65% (91.76%, 99.71%) |
| | NPA | 203 | 82 | 40.39% (33.58%, 47.49%) | 204 | 202 | 99.02% (96.50%, 99.88%) |
| Deletion | PPA | 11 | 11 | 100.00% (71.51%, 100.0%) | 11 | 11 | 100.00% (71.51%, 100.0%) |
| | NPA | 278 | 252 | 90.65% (86.60%, 93.80%) | 276 | 276 | 100.00% (98.67%, 100.0%) |
| Fusion | PPA | 2 | 0 | 0.00% (0%, 84.2%) | 2 | 0 | 0.00% (0%, 84.2%) |
| | NPA | 258 | 258 | 100.00% (98.6%, 100.0%) | 258 | 258 | 100.00% (98.6%, 100.0%) |

Table 5. Overall PPA results, at variant level, by gene.

| Gene | No Calls Included | | | No Calls Excluded | | |
| --- | --- | --- | --- | --- | --- | --- |
| | # Comp (+) | # ODx (+) | PPA (95% CI) | # Comp (+) | # ODx (+) | PPA (95% CI) |
| AKT1 | 2 | 2 | 100.00% (15.8%, 100%) | 2 | 2 | 100.00% (15.8%, 100%) |
| ALK | 0 | 0 | NA | 0 | 0 | NA |
| BRAF | 26 | 25 | 96.20% (80.4%, 99.9%) | 26 | 25 | 96.20% (80.4%, 99.9%) |

PMA P160045: FDA Summary of Safety and Effectiveness Data

{7}

Table 6. Overall NPA results, at variant level, by gene.

| Gene | No Calls Included | | | No Calls Excluded | | |
| --- | --- | --- | --- | --- | --- | --- |
| | # Comp (+) | # ODx (+) | PPA (95% CI) | # Comp (+) | # ODx (+) | PPA (95% CI) |
| AKT1 | 287 | 238 | 82.90% (78.1%, 87.1%) | 238 | 238 | 100.00% (98.46%, 100%) |
| AKT2 | 287 | 238 | 82.90% (78.1%, 87.1%) | 238 | 238 | 100.00% (98.46%, 100%) |
| AKT3 | 287 | 238 | 82.90% (78.1%, 87.1%) | 238 | 238 | 100.00% (98.46%, 100%) |

PMA P160045: FDA Summary of Safety and Effectiveness Data

{8}

| Gene | No Calls Included | | | No Calls Excluded | | |
| --- | --- | --- | --- | --- | --- | --- |
| | # Comp (-) | # ODx (-) | NPA (95% CI) | # Comp (-) | # ODx (-) | NPA (95% CI) |
| ALK | 7225 | 7005 | 97.00% (96.5%, 97.3%) | 7005 | 7005 | 100.00% (99.9%, 100%) |
| BRAF | 3730 | 3621 | 97.10% (96.5%, 97.6%) | 3621 | 3621 | 100.00% (99.9%, 100%) |
| CDK4 | 2023 | 1950 | 96.40% (95.5%, 97.2%) | 1950 | 1950 | 100.00% (99.81%, 100%) |
| DDR2 | 578 | 539 | 93.30% (90.9%, 95.2%) | 539 | 539 | 100.00% (99.32%, 100%) |
| EGFR | 12700 | 12497 | 98.40% (98.2%, 98.6%) | 12498 | 12497 | 99.99% (99.96%, 100%) |
| ERBB2 | 3753 | 3685 | 98.20% (97.7%, 98.6%) | 3685 | 3685 | 100.00% (99.9%, 100%) |
| ERBB3 | 3756 | 3731 | 99.30% (99%, 99.6%) | 3731 | 3731 | 100.00% (99.9%, 100%) |
| FGFR2 | 4624 | 4482 | 96.90% (96.4%, 97.4%) | 4482 | 4482 | 100.00% (99.92%, 100%) |
| FGFR3 | 1732 | 1663 | 96.00% (95%, 96.9%) | 1663 | 1663 | 100.00% (99.78%, 100%) |
| HRAS | 4909 | 4687 | 95.50% (94.9%, 96%) | 4688 | 4687 | 99.98% (99.88%, 100%) |
| KIT | 6647 | 6504 | 97.80% (97.5%, 98.2%) | 6504 | 6504 | 100.00% (99.94%, 100%) |
| KRAS | 8590 | 8318 | 96.80% (96.4%, 97.2%) | 8319 | 8318 | 99.99% (99.93%, 100%) |
| MAP2K1 | 4045 | 3938 | 97.40% (96.8%, 97.8%) | 3938 | 3938 | 100.00% (99.91%, 100%) |
| MAP2K2 | 1445 | 1408 | 97.40% (96.5%, 98.2%) | 1408 | 1408 | 100.00% (99.74%, 100%) |
| MET | 4039 | 3952 | 97.80% (97.3%, 98.3%) | 3952 | 3952 | 100.00% (99.91%, 100%) |
| MTOR | 6067 | 5773 | 95.20% (94.6%, 95.7%) | 5773 | 5773 | 100.00% (99.94%, 100%) |
| NRAS | 6637 | 6528 | 98.40% (98%, 98.6%) | 6528 | 6528 | 100.00% (99.94%, 100%) |
| PDGFRA | 2890 | 2860 | 99.00% (98.5%, 99.3%) | 2860 | 2860 | 100.00% (99.87%, 100%) |
| PIK3CA | 15857 | 15183 | 95.70% (95.4%, 96.1%) | 15184 | 15183 | 99.99% (99.96%, 100%) |

PMA P160045: FDA Summary of Safety and Effectiveness Data

{9}

| Gene | No Calls Included | | | No Calls Excluded | | |
| --- | --- | --- | --- | --- | --- | --- |
| | # Comp (-) | # ODx (-) | NPA (95% CI) | # Comp (-) | # ODx (-) | NPA (95% CI) |
| RET | 2890 | 2767 | 95.70% (94.9%, 96.5%) | 2767 | 2767 | 100.00% (99.9%, 100%) |
| ROS1 | 547 | 544 | 99.50% (98.4%, 99.9%) | 544 | 544 | 100.00% (99.32%, 100%) |
| SMO | 578 | 570 | 98.60% (97.3%, 99.4%) | 570 | 570 | 100.00% (99.35%, 100%) |

2. Analytical Sensitivity:

a. Limit of Blank:

To assess the performance of the Oncomine™ Dx Target Test in the absence of template and to ensure that a variant-free ("blank") sample does not generate an analytical signal that might be classified as a mutation, wild-type (WT) samples were evaluated. Three (3) FFPE clinical samples that were WT for the targeted DNA variants and RNA fusion locations and five FFPE cell line samples that were WT at all locations, were included in this study. The samples were tested using two different lots of the Oncomine™ Dx Target Test. Each sample was extracted once. The clinical samples were then prepared into twelve independent DNA and RNA libraries; and six independent DNA and RNA libraries for the cell line samples. Sequencing was carried out in duplicate for each of the libraries. For all 8 samples, there were no positive calls at any of the variant locations analyzed by the test. The false positive rate was therefore 0%, and the limit of blank (LoB) of the test is zero.

b. Limit of Detection:

The limit of detection (LoD) based on positive calls for the Oncomine™ Dx Target Test was estimated to determine the lowest allele frequency (AF) of SNV (simple and complex), MNV or deletion variants, and the lowest number of RNA fusion reads, at which 95% of the test replicates produced correct calls. A total of 26 specimens (19 FFPE tissue samples and 7 plasmid constructs) were evaluated and included samples with deletions ranging up to 18 nucleotides, MNVs [two base pair (bp) substitution) and SNVs, including variants in di- or trinucleotide repeat regions and in regions with 38% to 73% GC content. Samples with gene fusions in ROS1 were also included in this study. At least 6 titration levels were tested and each level was tested with 10 replicates per sample for each of the two reagent lots.

The LoD is based on the highest %AF (DNA-based variants) or number of reads (RNA-based variants) with 95% correct calls observed for each representative variant. The claimed LoD for the variants for which clinical claims will be made are noted in Table 7 below, which were confirmed in the

PMA P160045: FDA Summary of Safety and Effectiveness Data

{10}

external panel reproducibility study. The LoDs for DNA variants were determined to have AFs of $6 - 13\%$ , and the LoD for the ROS1 RNA fusion was measured at 732 fusion reads. Data to confirm the estimated LoDs for both DNA and RNA variants will be provided in an additional reproducibility study (see IX.A.5.a, below).

Table 7. Estimated LoD for clinically relevant variants.

| Gene | Variant | Variant Category | LoD1 (%AF or # Reads) |
| --- | --- | --- | --- |
| BRAF | V600E | SNV | 12% AF |
| EGFR | L858R | SNV | 8% AF |
| EGFR | Ex. 19del | Deletion | 6% AF |
| ROS1 | ROS1 Fusion | RNA Fusion | 732 Reads |

Clinical specimens were tested for all variants for which clinical claims are being sought.

# c. DNA/RNA Input:

This study was conducted in two parts using samples which represented the SNV (simple and complex), deletions, and fusion variant types targeted by the Oncomine™ Dx Target Test. The initial study utilized 7 variant positive and WT cell lines containing the BRAF V600E, EGFR L858R, and EGFR Ex. 19del DNA variants and a ROS1 RNA fusion. DNA and RNA isolated from multiple $10\mu \mathrm{m}$ FFPE cell line specimens. The DNA and RNA for the individual variants were pooled, quantified, and blended with the WT samples to the target AF (15%) or fusion target reads (300-600 fusion reads). Fifteen (15) input level combinations were tested for each sample consisting of 5 DNA input levels and 5 RNA input levels ranging from three levels below the standard of $10\mathrm{ng}$ (5 ng, 6.5 ng, and 8.5 ng), the standard input (10 ng) and one level (15 ng) above the standard input were tested in combination with 3 levels (5 ng, 10 ng, and 15 ng) of RNA or DNA respectively. The variants present in all FFPE DNA and RNA sample blends tested were called correctly $100\%$ of the time across all DNA/RNA test combinations, including the $10\mathrm{ng}$ required input

Based on the results from the initial study, a second study using FFPE clinical specimens was conducted to confirm that the Oncomine™ Dx Target Test was able to detect variants accurately at the DNA and RNA input amount of 10 ng as specified in the Oncomine™ Dx Target Test User Guide. Four (4) pre-characterized FFPE clinical samples containing the EGFR L858R variant, the NRAS Q61L variant, and two with a ROS1 fusion variant were tested in this study. Five (5) DNA input levels and 5 RNA input levels ranging from three levels below the standard of 10 ng (5 ng, 6.5 ng and 8.5 ng), the standard input and one level (15 ng) above the standard input were tested. For each DNA and RNA test condition, 6 replicates were tested per sample. For each sample, results from the expected variant location were examined and the data generated from the study were analyzed using an analysis of variance

PMA P160045: FDA Summary of Safety and Effectiveness Data

{11}

(ANOVA) to determine the relationship of input level to allele frequency (for DNA variants) or the number of fusion reads (for RNA variants).

For samples with DNA variants, all tested DNA/RNA input combinations yielded similar allele frequencies, including input combination of 10 ng for both DNA and RNA. Additionally, the positive and negative call rates were greater than 95% for all DNA/RNA input combinations under investigation.

For one of the ROS1 RNA fusion samples, there was no significant differences in log-fusion reads across all RNA and DNA input combinations under investigation, and the RNA fusions were called correctly 100% of the time across all DNA/RNA input combinations. However, for the second ROS1 RNA fusion sample, the input combination of 15 ng RNA/15 ng DNA yielded ROS1 fusion reads which were significantly lower than the reads yielded by the input combination of 10 ng for both RNA and DNA. In addition, the positive call rates were less than 95% at the input combination of 15 ng RNA/15 ng DNA, and at the input combination of 8.5 ng RNA/15 ng DNA.

The study confirmed the input of 10 ng for both DNA and RNA as specified in the Oncomine™ Dx Target Test User Guide.

d. Tissue Input:

Sixty (60) slide-mounted FFPE samples were analyzed to determine if samples extracted using the Ion Torrent™ Dx Total Nucleic Acid Isolation Kit yield DNA and RNA at the concentrations required by the Oncomine™ Dx Target Test when tissue input requirements are met. The test requires DNA at a concentration ≥ 0.83 ng/μL and RNA at a concentration ≥ 1.43 ng/μL. Thirty (30) resection samples with ≥ 20% tumor content were prepared without macrodissection, 15 resection samples with < 20% to ≥ 10% tumor cell content were macrodissected, and 15 samples were collected by core needle biopsy (CNB). For the resection samples, 2 × 5 μm sections were used per extraction. For CNBs, 9 × 5 μm sections were used per extraction. DNA and RNA concentrations were determined using the Ion Torrent™ Dx DNA and RNA Quantification Kits, respectively. No sequencing was performed on the extracted samples. Of the 60 samples tested, 98.3% (59/60) had a DNA concentration ≥ 0.83 ng/μL and an RNA concentration ≥ 1.43 ng/μL. One CNB sample failed the minimum DNA and RNA concentration specifications, with values of 0.52 ng/μL and 1.23 ng/μL, respectively. The low concentrations were likely caused by the small tissue size. The results demonstrated that sufficient DNA and RNA could be isolated from tissue sections which meet the specified tumor cell content requirement of ≥ 20%.

e. Tumor Content:

A range of tumor contents was tested in this study to determine the impact of varied tumor content on the Oncomine™ Dx Target Test performance. Fifty-

PMA P160045: FDA Summary of Safety and Effectiveness Data
Page 12

{12}

five (55) pre-characterized clinical samples with different percentages of tumor cell content (3 to 58%) were analyzed using the Oncomine™ Dx Target Test. The samples included the clinically indicated variants BRAF V600E, EGFR (Ex. 19del and L858R), and ROS1 fusions, as well as multiple other variants to represent the different variant categories (SNVs, MNVs, and deletions) were tested with and without macrodissection. The tumor cell contents of each specimen and region of interest (ROI) were estimated before the study by an external pathology lab. For samples analyzed without macrodissection, all variants were called positive by the Oncomine™ Dx Target Test, including all the DNA and RNA variants. For those samples that were analyzed with macrodissection, all DNA and RNA variants, except 1 DNA variant were called positive by the Oncomine™ Dx Target Test. The missed call was a no call in a BRAF V600E variant sample with a tumor content of 16%. The data supports Oncomine™ Dx Target Test’s requirement of 20% tumor content, and the claim that if the tumor content in the region of interest is ≥ 10%, the tissue samples can be macrodissected and enriched for tumor content.

3. Analytical Specificity:

a. Inclusivity/Cross-Reactivity: An *in silico* cross-reactivity analysis was performed to evaluate the specificity of the primers in the Oncomine™ Dx Target Test Kit DNA and RNA panels. The primers were checked for specificity to the human genome, human transcriptome, and representative bacterial, fungal, and viral genomes. Any unintended amplification products were required to have a minimum of 2 bp mismatches to the intended amplification product sequences generated by the panels, because mismatches of ≥ 2 bp prevent mapping to the same location on the genome due to a low mapping score. For the DNA panel primers, *in silico* analysis predicted 20 unintended potential amplicon generating primer pairings against the human genome. Of these, 19 unintended amplification products had mismatches of 2-122 bp to intended products, and therefore would not cause false results. The remaining unintended primer pairing would only detect the existing WT or variant location and would not cause false results. For the RNA panel primers, unintended primer pairings against the human genome and the human transcriptome were predicted; however, due to the level of mismatches none were predicted to cause false results. The results demonstrated that the primers are specific for the intended targeted sequences.

4. Interference: To evaluate the potential impact of endogenous (necrotic tissue and hemoglobin) and exogenous interferents (paraffin, xylene, ethanol, proteinase K, and wash buffer), this study evaluated 8 clinical FFPE samples in 6 replicates for both DNA and RNA for every combination of sample and condition taken through the entire test workflow. The samples included all clinically indicated variants and represented SNV (simple and complex), deletions, and fusion variant types,

PMA P160045: FDA Summary of Safety and Effectiveness Data
Page 13

{13}

targeted by the Oncomine™ Dx Target Test. This included the sample preparation and processing steps (extraction, DNA and RNA quantification), steps prior to taking the purified nucleic acid through to library construction, templating, PGM sequencing and variant calling. The samples included WT and those with representative SNV, deletion, and fusion variants. The impact of potentially interfering substances on assay performance was evaluated, and the results were compared to the control (no interferents) condition.

a. Endogenous Interference:

A review of the clinical study data indicated that the presence of necrotic tissue up to approximately 20% did not appear to interfere with the assay. Results from an interference study using variant positive clinical samples will be provided in order to confirm that the presence of necrotic tissue does not interfere with the performance of the Oncomine™ Dx Target Test.

Hemoglobin was evaluated at 4 mg/mL. The allelic frequencies for SNV and deletion variants were ~10-20% and the fusion reads for the RNA variants were ~500-1000 reads. The positive concordance with the control condition (with no calls being excluded) across all samples, and the overall concordance with the control condition (no calls excluded) across all samples were calculated. One false positive result and one false negative result were observed across all study samples and all Oncomine™ Dx Target Test variant locations. There were 15 invalid results for two samples both for DNA (N = 1) and RNA (N = 14) samples across both control and test conditions. The overall concordance (no calls excluded) was 99.99%, and the overall data demonstrate that hemoglobin does not adversely impact the performance of the assay.

b. Exogenous Interference:

For the study with exogenous interferents, both the positive concordance (no calls excluded) and the overall concordance (no calls excluded) for all samples exceeded the study acceptance criteria of > 95%, with respect to each interferent under investigation. The data support that these interfering substances can be tolerated by the assay at the levels tested.

c. Anti-microbial Testing

Testing was performed to determine the effectiveness of the antimicrobial preservatives in 23 reagents used in the Oncomine™ Dx Target Test with five organisms (i.e., S. aureus, P. aeruginosa, C. albicans, E. coli, and A. brasiliensis). Two additional reagents (Templating (TMPL) Water and TMPL Tween Solution) which do not include antimicrobial preservatives were also included in the study. These reagents - which are included as part of the Ion Torrent Dx FFPE Sample Preparation Kit, the Ion PGM Dx Library Kit, Ion PGM Dx Templating Kit and Ion PGM Dx Sequencing Kit - are designed for storage at refrigerated or room temperature (15-30°C) conditions. Testing was performed at 20-25°C and tested at 5 time points post-inoculation (T₀ days, T₁

PMA P160045: FDA Summary of Safety and Effectiveness Data
Page 14

{14}

day, T₇ days, T₁₄ days and T₂₈ days). In addition to the initial inoculation and testing, aliquots of un-inoculated products were stored for future inoculation and testing again at the same 5 time points after 12 and 24 months storage at the recommended storage conditions. Twenty-three (23) of the 25 reagents (92%) met acceptance criteria for bacteria of ≥ 2 log reduction from T₀ to T₁₄ days; no increase from T₁₄ to T₂₈ days; and for yeast and mold no increase from initial (T₀) to T₁₄ and T₂₈ days. The 2 reagents which do not include antimicrobial preservatives did not demonstrate a reduction in colony counts for E. coli and P. aeruginosa. The product instructions for use includes a warning for users to visually inspect the TMPL Water and TMPL Tween Solution for potential bacterial or fungal contamination prior to use, and if evident, the user should discard and use new vials.

5. Precision and Reproducibility:

Two reproducibility studies, a multi-site assay reproducibility study and a separate sample processing reproducibility study, were performed to assess precision (within-run) and reproducibility (across different sample preparation reagent lots, sites, operators, and instrument systems).

a. External Panel Reproducibility Study (Assay Reproducibility):

The external reproducibility study was conducted across 4 US sites for the purposes of demonstrating within-run precision performance (repeatability) and variability across sites, operators and instrument platforms (reproducibility). At each site, the 4 operators were grouped into 2 pairs with each pair assigned to 2 instrument systems and were responsible for testing each of the pre-extracted DNA and RNA specimens representing all of the clinically indicated variants and each of the variant types (simple and complex SNVs, MNVs, deletions, and fusions) targeted by the Oncomine Dx Target Test. The 18 DNA samples consisting of 16 DNA samples with two or more DNA variants (blends of clinical samples and plasmids including BRAF V600E, EGFR L858R, and EGFR Ex.19del) and 2 were WT samples, and the 9 RNA FFPE clinical samples included 3 ROS1 fusions. Each specimen in the DNA and RNA panels were run in duplicate for 5 non-consecutive days using 2 different reagent lots at 3 of the study sites and all three reagent lots at the 4th study site for a total of 72 test determinations per DNA specimen and 144 test determinations per RNA specimen.

The positive call rates, with two-sided 95% confidence intervals using exact method, with and without no calls were calculated. When including no calls, the positive call rate for DNA variants (at the variant location) ranged from 87.1% (77%, 93.9%) to 100% (95%, 100%) and when no calls were excluded the positive call rate ranged from 97.1% (89.9%, 99.6%) to 100% (95%, 100%). Within-run precision ranged from 85.3% (68.9%, 95%) to 100% (90.3%, 100%) with no calls included and 93.9% (79.8%, 99.3%) to 100% (90.3%, 100%) when no calls are excluded. When including or excluding no calls, the positive call rate for the ROS1 fusion-containing samples (at the

PMA P160045: FDA Summary of Safety and Effectiveness Data
Page 15

{15}

variant location) ranged from 86.1% (79.4%, 91.3%) to 100% (97.5%, 100%). Within-run precision ranged from 75% (63.4%, 84.5%) to 100% (95.0%, 100%) with no calls included or excluded.

The average positive agreement (APA) and average negative agreement (ANA) at the variant-level were defined for a given pair of operators, calculated by comparing each possible set of sequencing results from the first operator with each possible set of sequencing results from the second operator at a given site. The two-sided 95% confidence intervals for APA and ANA were calculated using bootstrap method when APA or ANA was not 100%, or exact method when APA or ANA was 100%. For each DNA sample that carries at least one variant (including no calls), the APAs ranged from 78.1% (56.3%, 100%) to 100% (97.2%, 100%) and the ANAs ranged from 86.2% (73.7%, 94.7%) to 99.8% (99.6%, 99.8%). For the 2 WT DNA samples, the APAs (including no calls) were either not applicable or 8.3% (0%, 20.0%) due to false positive calls and ANAs ranged from 90.5% (80.4%, 99.0%) to 99.8% (97.4%, 99.8%). Excluding no calls, the APAs ranged from 78.1% (56.3%, 100%) to 100% (97.2%, 100%) and ANAs ranged from 99.8% (99.7%, 100%) to 100% (100%, 100%) comparing two operators.

For each RNA sample that carried at least one variant, the APAs for each specimen on the variant level from 75.0% (52.5%, 91.4%) to 100% (97.2%, 100%) and ANAs ranged from 99.7% (99.6%, 99.9%) to 100% (100%, 100%) with no call included or excluded.

In this study, the AFs ranged from 0.9x - 2.2x and 1.5x - 3.2x of the determined LoD values, which for the most part exceeded the LoD established in the LoD study. As a result, the reproducibility of the test near 1x LoD could not be determined, and an additional reproducibility study is to confirm the established LoD.

b. Precision:

Precision was estimated with respect to positive variant locations for within-run, between-system, between-operator, between-site, between-lot and total variability. When excluding No Calls from the assay reproducibility study data, the estimate of repeatability was at least 99.3%, with a lower limit of the corresponding 95% CI of ≥ 98.3%, at each of the clinical DNA variant locations. When No Calls were included, the estimate of repeatability was at least 95.2%, with a lower limit of the corresponding 95% CI of ≥ 93.2%, at each of the clinical DNA variant locations assessed. The estimate of repeatability for ROS1 was 94.4% with a lower limit of the corresponding 95% CI of 92.3%.

When the variance components were assessed, the between-run component had the greatest contribution to the total variability (up to 25.3% CV) for DNA samples.

PMA P160045: FDA Summary of Safety and Effectiveness Data
Page 16

{16}

c. External Sample Processing Reproducibility Study:

Two (2) WT samples and 10 variant-positive samples, which included simple and complex SNVs, MNVs, deletions, and fusions represented by BRAF V600E, EGFR L858R, 2 common EGFR Ex. 19del, ROS1 fusion transcripts, and other representative variants, were evaluated at 4 US testing sites, each with 4 Ion PGM™ Dx Systems and 4 operators, to determine the reproducibility and repeatability of sample processing. Each sample was tested 8 times at each site, for a total of 32 replicates per sample, or 768 sample sequencing results [12 samples × 32 replicates × 2 library types (RNA and DNA)]. The call rate, no call rate, positive call rate, negative call rate, and within-run repeatability were computed at each variant location of interest. Three DNA samples generated at least one no call, and the ROS1 RNA fusion sample generated 2 negative results out of the 32 replicates. Including or excluding no calls, all positive call rates from positive variant locations were >93%, with a lower bound of 79.2%.

6. Tissue Heterogeneity:

Tissue heterogeneity was assessed as part of the sample processing reproducibility study. Results from a comparison of unstained sections from across multiple sections of the same tissue sample demonstrated that the Oncomine™ Dx Target Test was able to produce repeatable and reproducible results across different measured tumor content levels and AF% or log (fusion reads).

7. Extraction Method Equivalency Studies:

Data were provided to demonstrate the equivalency of data generated from previously extracted DNA and RNA from clinical samples included in two of the clinical studies intended to support this PMA.

a. DNA:

This study assessed whether the Oncomine™ Dx Target Test produces comparable DNA variant from FFPE samples prepared with the Ion Torrent™ Dx FFPE Sample Preparation Kit and the QIAamp® DSP DNA FFPE Tissue Kit™ (IVD). Twenty (20) NSCLC FFPE tissue blocks with a mix of SNVs and deletions (including 10 samples with EGFR mutations) were evaluated. Among the 20 samples, 19 met the minimum DNA concentration requirement and produced equivalent variant calling using the two different sample preparation kits. One (1) sample was discordant due to significantly lower concentration obtained using the QIAamp protocol (QIAamp protocol requires elution in 120 μL, while the Ion Torrent Dx extraction elutes in 30 μL). The valid pairs of sample results demonstrated that all concordance statistics at the variant and sample levels were 100% when excluding no calls. The results from this study support that the Oncomine™ Dx Target Test system calls at DNA variant locations are equivalent when using either of the extraction methods.

PMA P160045: FDA Summary of Safety and Effectiveness Data
Page 17

{17}

b. RNA:

This study was performed to determine the comparability of the Ion Torrent Dx FFPE Sample Preparation Kit and a different commercially available RNA/DNA extraction system to prepare RNA to be used with the Oncomine™ Dx Target Test. Twenty (20) clinical FFPE specimens and FFPE cell lines, representing WT and RNA fusion positive specimens including ROS1, were evaluated. Positive concordance statistics at the variant level exceeded 85%, and negative concordance and overall concordance statistics at the variant level exceeded 95%. At the sample level, the concordance for the fusion-positive samples was greater than 85%. The negative and overall concordance were both 90% at the sample level. The relatively reduced negative concordance was due to a sample which had a low level of fusion expression, irrespective of the extraction method; and likely contamination from the assay control in another sample. Overall, the results support that results from the Oncomine™ Dx Target Test can be considered comparable when using either extraction system.

8. Contrived Sample Functional Characterization Study:

To support the use of plasmids and cell lines as an alternative for clinical samples for certain variants, which are difficult to acquire, a study was conducted to evaluate the functional behavior of the contrived samples to that of clinical specimens using the Oncomine™ Dx Target Test. Five (5) representative variants (for SNVs and deletions), which included all DNA clinically indicated variants, were tested at 6 AF levels (2%, 4%, 5%, 6%, 8%, and 12%) for each of the 3 sample types (FFPE clinical sample, FFPE cell line, and plasmid). AFs were verified by an analytically validated comparator method. Eight operators and multiple reagent lots were used to test 20 replicates for each of the 6 AF levels.

For each variant, the average AF associated with each sample level was computed, and the corresponding proportion of correct calls (hit rate) was determined and converted to probits (i.e., units from the cumulative standard normal distribution function). A generalized linear model was fitted, with the probit modeled as a function of AF level, utilizing the Firth penalized likelihood method. The goodness of fit of each resulting probit model was evaluated using Pearson's Chi-square test, and the fit of each model was found to be acceptable. Estimates of C25, C50, C75, and C95 (corresponding to hit rates of 25%, 50%, 75% and 95%, respectively) were computed for each sample type, along with the corresponding 95% confidence intervals. In addition, the differences of the predicted estimates of C25, C50, C75 or C95 - as well as the corresponding two-sided 95% confidence intervals using bootstrap method - between the plasmid blends or cell lines and the clinical samples were also calculated. The results show that differences in the predicted estimates of C25, C50, C75 or C95 between the plasmid blends or cell lines and the clinical samples were considered small and acceptable for all tested variants, demonstrating equivalent performance between sample types.

PMA P160045: FDA Summary of Safety and Effectiveness Data
Page 18

{18}

9. Guard Band Studies:

a. Workflow Tolerances:

The tolerances encompassing the workflow steps, library preparation, template preparation and sequencing were assessed across 20 separate studies corresponding to the test's most critical workflow steps which could lead to assay failure. Each study included 3 test points, which included testing in low condition, nominal condition as defined by the instructions for use, and high condition. The guard banding range for each experiment was designed such that the maximum and minimum test points challenged the system, while still being within operational error range. The study was conducted across multiple runs, utilizing multiple operators and instrument systems. For each study, 6 independent libraries were prepared per condition and pooled into three sets of pools with one pool tested at the low condition, one at the high condition and the remaining samples tested at the nominal condition. Each pool was tested in a single system run, resulting in a total of 3 pools tested in 3 runs, with 3 or 6 replicates per condition.

Of the 20 studies, one study, Thermal Cycling Temperature Offset, demonstrated a significant difference in performance when the temperature deviation in the PCR thermocycler during all thermocycling steps were increased by either +0.8°C to +1°C, which resulted in the samples failing the test QC parameters. The acceptable tolerance was therefore defined as -1°C to +0.5°C of the specified temperature. Of the remaining studies, 8 demonstrated no significant difference in results while 11 did show a statistically significant difference; however, it is recommended that each step in assay preparation and sequencing be followed according to the instructions for use. An additional study using fusion-positive clinical samples is ongoing to verify the tolerance ranges of the critical elements of the RNA workflow for the Oncomine™ Dx Target Test.

b. Tissue Fixation Study:

The impact of tissue fixation time on the performance of the Oncomine™ Dx Target Test was assessed by fixing WT cell line pellets in 10% neutral buffered formalin (NBF) for 12, 24, 48, 72, and 84 hours at room temperature. The results from the fixed specimens were compared to those from the same cell line that had not undergone any fixation with 10% NBF. One (1) replicate of each fixation time condition and the non-FFPE condition were tested on three different chips, for a total of three replicates per condition. The average AF observed at each of the 103 cytosine deamination-susceptible hotspots was determined for each condition tested. The results showed two cytosine deamination events as a result of the fixation process; however, neither event resulted in a false-positive call. DNA and RNA sequencing quality was evaluated by measuring percent reads, no call rate, and total mappable reads (TMR) for each condition tested. All DNA and RNA sequencing results met the DNA and RNA sample QC metrics. The results of this study support the

PMA P160045: FDA Summary of Safety and Effectiveness Data
Page 19

{19}

claim that specimens may be fixed for up to 84 hours in 10% NBF prior to testing with the Oncomine™ Dx Target Test.

c. Contamination Study:
Contamination events throughout the workflow from extraction to sequencing were assessed. Eight FFPE cell line samples, representing simple and complex SNVs, deletions, and RNA fusion variant categories, were used to ensure adequate sample quality and to allow testing the more challenging condition of high AFs (> 50%). Each FFPE cell line carried a unique variant, but was WT for all other variants evaluated in this study and WT RNA from two of the samples was also tested. The AFs ranged from 67% - 100% for the DNA variants (including EGFR Ex. 19del, KRAS G12D, NRAS, EGFR L858R), a WT sample, and for the RNA variants 14.7K fusion reads, which is much higher than the 40 read cutoff. Alternating a variant sample run and a run with samples that are WT at these variant locations replicated a situation where cross contamination could be observed. To evaluate cross contamination, each of the four operators tested all cell line samples side-by-side throughout the entire workflow. To evaluate carryover contamination, on each of the 4 systems, 5 rounds of the entire test protocol (extraction to variant calling) with the same set of samples were conducted consecutively. Samples were assigned alternating barcodes in consecutive runs conducted on the same instrument, per User Guide instruction.

The false positive (FP) rate at each of the DNA variant location examined in this study was 0% (0/100) and the FP rate for each of the RNA variants was no more than 1.25% (1/80). One FP result just above the fusion detection threshold of > 40 fusion target reads was observed in the study. The contamination event was likely caused by cross-contamination from a sample that was located in an adjacent well. The observed FP rate met the Oncomine™ Dx Target Test requirement of less than or equal to 3%.

10. Stability Studies:

a. Shelf-Life Stability:
Three (3) separately manufactured kit lots including all components of the Oncomine™ Dx Target Test and Controls Kit were stored according to the storage conditions specified in the product labeling. Stability of the reagents is being evaluated by testing 5 FFPE samples at specified time points from baseline, 2-4, and 7 months. For each sample each of the assay QC metrics was evaluated in addition to the final calls. The data demonstrate that the kit components of the Oncomine™ Dx Target Test and Controls Kit are stable for at least 6 months.

b. In-Use Stability:
The in-use stability study incorporated both open vial stability and freeze/thaw (FT) stability. Five (5) separately manufactured kit lots were tested at 6

PMA P160045: FDA Summary of Safety and Effectiveness Data
Page 20

{20}

functional testing time points and 2 mock workflow time points to cover 8 weeks for a total of 8 time points. A total of 72 samples were processed and DNA and RNA were tested across a total of 12 system runs. All runs and all samples except for one RNA sample passed the specifications and the results were used for analysis. In place of the failed RNA sample, the RNA sample using the same barcode from the back up run (Run 2) was used. The data supported in-use stability for 7 weeks and a maximum of 8 freeze/thaws.

c. Designated Hold Times:

The workflow for the Oncomine™ Dx Target Test incorporates several optional stopping points to hold assay intermediates. The stability of the intermediate products was evaluated by incorporating all of the 13 optional extended hold times specified in the User Guide. A total of 3 samples (2 FFPE clinical samples and 1 FFPE cell line sample) were included in this study. The SNV, deletions, and fusion variant types were represented by samples which contained EGFR L858R, EGFR Ex. 19del, BRAF V600E, a ROS1 fusion, and other representative variants. Each sample was tested under 3 different test conditions (Table 8 below).

Table 8. Designated hold time test conditions.

| Condition | Eluted Library Hold Time |
| --- | --- |
| A) Nominal Hold^{1} | No hold |
| B) Library Hold | 30 day hold
3 month hold |
| C) Combo Hold^{2} | No hold |
| 1Libraries prepared for condition A were used to test condition B
2Includes 12 of the 13 optional hold steps, excluding the eluted library hold tested in condition B. Each step was held at the maximum recommended time to approximate a “worst-case scenario”. | |

For DNA, AF and the log-transformed median absolute pairwise difference [log(MAPD)] were used as metrics to evaluate stability. For RNA, the log-transformed fusion reads and the log-transformed normalized read ratio [i.e., log(fusion reads/total mapped reads)] were used as metrics to evaluate stability. In all of the evaluations, the results of the test conditions with the incorporated hold times were compared to the samples tested without the hold times. The study results support the conclusion that the 30-day library hold and combo hold conditions did not result in a decrease in Oncomine™ Dx Target Test performance relative to the nominal test condition.

d. Kit Lot Interchangeability:

An initial study were provided using control samples to demonstrate that different lots of the Oncomine™ Dx Target Test Kit components (Oncomine™ Dx Target Test Control Kit DNA and Oncomine™ Dx Target Test Control Kit RNA, and the Oncomine™ Dx Target Test RNA and DNA

PMA P160045: FDA Summary of Safety and Effectiveness Data
Page 21

{21}

panels) are interchangeable. A study using clinical specimens to confirm that different lots of each of the components of the Oncomine™ Dx Target Test and Controls Kit may be used interchangeably will be conducted.

e. Sample Stability - Extracted RNA and DNA:

Storage and freeze-thaw (FT) stability of DNA and RNA extracted from FFPE clinical samples and FFPE cell line samples were assessed at baseline, 7 weeks, 3 months + 1 week, 6 months + 1 week, and 9 months + 1 week to support the stability of the extracted DNA and RNA at 9 months using the Ion Torrent™ Dx FFPE Sample Preparation Kit. DNA and RNA extracted from FFPE cell lines and clinical FFPE samples, representing SNVs, deletions, and RNA fusions by the clinically indicated variants, were aliquoted and stored at -30°C to -10°C and -90°C to -60°C, respectively. The DNA and RNA samples contained a SNV and a deletion variant and RNA fusions, respectively in both the respective clinical and cell line FFPE specimens to support stability across the variant categories and represented 3 of the clinically indicated variants.

For baseline, the samples were sequenced within one week of aliquoting. For the purpose of this study, the date of library preparation target amplification is considered to be start of the time point. At each time point, DNA and RNA aliquots were frozen and thawed once or 3 times and then sequenced using the Oncomine™ Dx Target Test. The data supported stability of extracted DNA stored at -30°C to -10°C and RNA stored at -90°C to -60°C for 9 months after 3 FT cycles.

f. Stored Slide Stability:

Eight (8) clinical sample FFPE blocks and 1 cell line FFPE block representing each of the variant categories (SNVs, deletions, and fusions) with each containing a known variant, including each of the clinically indicated variants plus one additional variant detected by the test, were sectioned to provide at least 30 sections per block and every other slide was dipped in paraffin. All slides were stored at 15°C to 30°C. The slides were tested at baseline plus three additional time points [7 week (49 days), 4 months+1 week (129 days), and 6 months+1 week (190 days)] to initially demonstrate stability at 8 months. At each time point, DNA and RNA were extracted from 2 paraffin dipped and 2 un-dipped slides per sample and then sequenced according to the approved stability protocol. Based on the data provided, both paraffin dipped and undipped slides are stable for at least 8 months when stored at 15°C to 30°C. Slide stability testing is intended to continue to demonstrate stability to at least 12 months when stored at 15°C to 30°C.

g. Stored Block Stability:

Seven (7) clinical FFPE tissue blocks and one FFPE cell line block comprising 6 DNA variants and 1 ROS1 fusion, representing each of the clinically indicated variants and variant categories, were tested in quadruplicate in 6 sequencing runs for each of five time points: baseline (0

PMA P160045: FDA Summary of Safety and Effectiveness Data
Page 22

{22}

days), 8 week (56 days), 4 months+2 weeks (134 days), 6 months+2 weeks (194 days), and 8 months+2 weeks (254 days). In addition, 1 FFPE cell line block and a single clinical FFPE RNA fusion positive sample was tested at four time points [baseline (0 days), 8 week (56 days), 4 months+2 weeks (134 days), 6 months+2 weeks (194 days)]. The mean AFs were calculated at each time point for the DNA samples and the mean fusion reads and the mean log(10)-fusion reads were calculated for the RNA sample, and comparisons were made to $\mathrm{T_0}$ values. The AF at baseline for the DNA variants ranged from $26.7\%$ to $47.3\%$ and the mean fusion reads for the RNA variants was 9410 (3.97 mean log-fusion reads). Based on the data provided, DNA extracted from DNA variant containing clinical samples is stable for 8 months and DNA extracted from cell line FFPE blocks for 6 months when the sample block is stored at $15^{\circ}\mathrm{C}$ to $30^{\circ}\mathrm{C}$ . RNA extracted from the RNA fusion clinical sample block is stable for 6 months when stored at $15^{\circ}\mathrm{C}$ to $30^{\circ}\mathrm{C}$ .

h. Transport Stability:

The transport stability study was performed to demonstrate that the shipping configurations developed for all kit components required for the Oncomine™ Dx Target Test provide adequate thermal and physical protection as packages are transported from the manufacturing site to customers. Three (3) separately manufactured kits and component reagents lots were exposed to simulated transport challenges intended to simulate the longest estimated domestic and international shipping times of 48 and 72 hours, respectively. The simulated transport conditions included both physical and temperature challenges, which included 7 packaging configurations (domestic vs. international shipping) and 4 temperature profiles (72-hour summer, 72-hour winter, 144-hour summer and 144-hour winter), respectively. The 48-hour domestic time point was assessed using the first 48-hours of the 72-hour profile and the 144-hour profile consisted of two consecutive 72-hour time periods. Each of the kits and components were functionally tested using a combination of DNA and RNA controls, synthetic, and FFPE samples to ensure reagent functionality in order to establish transport stability. Based on the results of the functional testing for each of the conditions, demonstrating that the shipping configurations provide adequate thermal and physical protection for the simulated time points.

## B. Animal Studies

Not Applicable

## X. SUMMARY OF PRIMARY CLINICAL STUDIES

Thermo Fisher Scientific conducted 3 separate clinical bridging studies to establish reasonable assurance of safety and effectiveness of the Oncomine™ Dx Target Test for detection of BRAF V600E and EGFR Ex. 19del and L858R mutations and ROS1 fusions in FFPE tumor specimens to select NSCLC patients for treatment with TAFINLAR®

PMA P160045: FDA Summary of Safety and Effectiveness Data
Page 23

{23}

(dabrafenib) in combination with MEKINIST® (trametinib), or IRESSA® (gefitinib), or XALKORI® (crizotinib), respectively. Justifications for the acceptance of foreign data were provided. Data from these bridging studies were the basis for the PMA approval decision. For each of the clinical claims, drug efficacy was based on clinical studies that enrolled only biomarker positive patients; therefore, a separate cohort of NSCLC patient samples from a population similar to those enrolled into the bridging studies was obtained and screened to be negative for the targeted BRAF, EGFR, and ROS1 abnormalities [acquired PCR(-) or FISH(-) samples] using the biomarker relevant comparator method assays. A summary of the clinical studies are presented below.

## A. BRAF Study Design

The clinical benefit of the Oncomine™ Dx Target Test was demonstrated in the analysis of efficacy and safety data obtained from two of three cohorts from a Phase II multi-center, multi-cohort, non-randomized, open-label trial (Study BRF113928). BRAF V600E mutation NSCLC patients enrolled into the two cohorts (B and C) received the combination therapy (dabrafenib 150 mg BID and trametinib 2 mg once daily) until disease progression or unacceptable toxicity. In the BRF113928 study, prospective patients were screened for a BRAF V600E mutation in NSCLC specimens and enrolled based on results from local lab tests (LLTs) used at each of the enrollment sites. Central confirmation testing for the BRAF V600E mutation was performed with the Oncomine™ Dx Target Test after enrollment.

Study BRF113928 included the following three cohorts:

- Cohort A – To evaluate dabrafenib monotherapy in 84 subjects with BRAF V600E NSCLC, that had relapsed or progressed on at least one platinum based chemotherapy regimen prior to enrollment. (Subjects receiving and adequately tolerating dabrafenib as a single agent and who continue to meet the inclusion and exclusion criteria had the option to switch to dabrafenib and trametinib combination treatment.) Cohort A was not included in the clinical efficacy analysis portion of the clinical bridging study, but utilized to establish concordance with the Oncomine™ Dx Target Test only.
- Cohort B – To evaluate dabrafenib and trametinib combination therapy in 57 BRAF V600E positive subjects who had progressed on at least one prior treatment for advanced / metastatic disease.
- Cohort C – To evaluate dabrafenib and trametinib combination therapy in 36 BRAF V600E positive NSCLC subjects who had not received prior treatment for advanced/metastatic disease (i.e., first line subjects).

Samples from a subset of enrolled patients from the intended population were retrospectively evaluated using the Oncomine™ Dx Target Test in a clinical bridging study (Study CDRB436E2201). The two objectives for the bridging study include: 1) agreement between results from the Oncomine™ Dx Target Test and the LLTs for BRAF V600E detection, and 2) the clinical outcomes (i.e., ORR) for the NSCLC patients (Cohorts B and C) who were positive for BRAF V600E mutations, as identified by the Oncomine™ Dx Target Test. Because no FDA-approved test was

PMA P160045: FDA Summary of Safety and Effectiveness Data
Page 24

{24}

available for detection of BRAF V600E in FFPE NSCLC specimens, a validated PCR assay, BRAF V600 PCR Mutation Test, served as a surrogate assay for the LLTs in the bridging study. This assay was similar to the LLTs used at more than one enrollment testing site and the validated assay used to screen the acquired PCR(-) samples.

Subjects with locally determined BRAF V600E mutant NSCLC were screened for clinical eligibility criteria for enrollment into the bridging study along with ~130 acquired PCR(-) samples.

1. Clinical Inclusion and Exclusion Criteria:
Enrollment in the bridging study CDRB436E2201 was limited to patients who met the following inclusion criteria:
- Patient informed consent provided for use of clinical sample in eligibility screening for the BRF113928 clinical trial and/or commercial samples with appropriate use consistent with guidance for leftover human specimens;
- Availability of adequate sample to generate an Oncomine™ Dx Target Test result including the following DNA requirement:
- DNA isolation - Use 10 ng of DNA at a concentration of ≥ 0.83 ng/μL.

Patients were not permitted to enroll in the bridging study CDRB436E2201 if they met any of the following exclusion criteria:
- Lack of clear subject identification or label on banked patient sample;
- Obvious physical damage of banked patient sample which precludes sample processing;
- No tumor present per pathology review.

2. Follow-up Schedule:
As the bridging study was conducted retrospectively to establish safety and effectiveness for selecting patients using the Oncomine™ Dx Target Test, there was no follow-up conducted.

3. Clinical Endpoints:
The objective response rate (ORR) by investigator assessment (IA) observed in Cohorts B and C were submitted to support the approval of dabrafenib and trametinib combination therapy in patients with BRAF V600E mutation positive NSCLC under NDAs 202806/S006 and 204114/S005.

Effectiveness of the Oncomine™ Dx Target Test was determined by measuring the objective ORR by investigator assessment for patients with stage IV NSCLC who tested positive for the BRAF V600E mutation by both the LLTs and the Oncomine™ Dx Target Test. The ORR was calculated for patients in two cohorts (B and C) who were selected for treatment with dabrafenib administered in combination with trametinib.

B. Accountability of PMA Cohort

PMA P160045: FDA Summary of Safety and Effectiveness Data
Page 25

{25}

At the time of the analysis cut-off date, August 8, 2016, for Study BRF113928, 84 patients were enrolled in Cohort A, 59 patients were enrolled in Cohort B, and 34 patients were enrolled in Cohort C. For Cohort A, 6 patients were excluded because they did not meet the enrollment criteria, an additional 11 samples were excluded due lack of sample availability and 22 were not tested due to insufficient DNA, RNA, lack of tumor, etc.

For Cohort B, 2 patients received first line treatment via protocol deviation, and therefore were included as part of the intended use population in Cohort C; samples from 2 subjects were not sent for central confirmation testing by the Oncomine™ Dx Target Test; and an additional 17 subjects were excluded due to either insufficient tumor tissue to process (n = 11), or insufficient DNA yield to proceed with testing (n = 6). As a result, 38 samples were available for testing with the Oncomine™ Dx Target Test. Of those, 14 samples were excluded due to not meeting the sequencing quality control (QC) metrics (n = 9) or insufficient coverage to generate a valid result (n = 5). The final number of samples available with a valid Oncomine™ Dx Target Test result was 24 samples (24/57 = 42.11%).

For Cohort C, there were a total of 36 patients, including 34 patients enrolled, and the 2 first line patients initially enrolled into Cohort B, who actually received first line treatment. Three (3) samples were not shipped for central confirmatory testing in time for inclusion into the study, 5 samples were excluded due to insufficient tissue or insufficient tumor to process, and therefore, did not meet the minimum DNA input requirement. An additional 5 samples were excluded due to not meeting the sequencing QC metrics (n = 3) or insufficient coverage to generate an valid result (n = 2). The final number of samples available with a valid Oncomine™ Dx Target Test result was 23 samples (23/36 = 63.89%).

## C. Study Population Demographics and Baseline Parameters

Baseline characteristics were compared between those patients with and without evaluable Oncomine™ Dx Target Test results. In general, for both cohorts B and C, the demographics for patients with and without a Oncomine™ Dx Target Test result are similar, apart from the specimen type for which there was a higher proportion of patients with resection specimens. Comparisons between the clinically relevant patient demographics of the study populations are shown for cohorts B and C in the tables below.

Table 9. Baseline characteristics between the patients included and excluded from the bridging study for Cohort B.

| Baseline characteristics | Patients in bridging study N=38 | Patients not in bridging study N=19 | All N=57 |
| --- | --- | --- | --- |
| Age (Years) | | | |
| Mean | 64.7 | 65.7 | 65.1 |
| SD | 10.40 | 9.85 | 10.14 |

PMA P160045: FDA Summary of Safety and Effectiveness Data
Page 26

{26}

| Baseline characteristics | Patients in bridging study N=38 | Patients not in bridging study N=19 | All N=57 |
| --- | --- | --- | --- |
| Median | 63.5 | 65.0 | 64.0 |
| Min | 41 | 50 | 41 |
| Max | 85 | 88 | 88 |
| Gender - n (%) | | | |
| Female | 18 (47.4) | 10 (52.6) | 28 (49.1) |
| Male | 20 (52.6) | 9 (47.4) | 29 (50.9) |
| Race - n (%) | | | |
| American Indian Or Alaskan Native | 0 (0) | 0 (0) | 0 (0) |
| Asian | 3 (7.9) | 1 (5.3) | 4 (7.0) |
| Black or African American | 2 (5.3) | 0 (0) | 2 (3.5) |
| Native Hawaiian Or Other Pacific Islander | 0 (0) | 0 (0) | 0 (0) |
| White | 32 (84.2) | 17 (89.5) | 49 (86.0) |
| Multiple | 1 (2.6) | 0 (0) | 1 (1.8) |
| Missing | 0 (0) | 1 (5.3) | 1 (1.8) |
| ECOG baseline - n (%) | | | |
| 0 | 13 (34.2) | 4 (21.1) | 17 (29.8) |
| 1 | 21 (55.3) | 14 (73.7) | 35 (61.4) |
| 2 | 4 (10.5) | 1 (5.3) | 5 (8.8) |
| Histology - n (%) | | | |
| Adenocarcinoma | 36 (94.7) | 17 (89.5) | 53 (93.0) |
| Adenosquamous Carcinoma - Predominantly Adenocarcinoma | 1 (2.6) | 0 (0) | 1 (1.8) |
| Bronchioloalveolar | 1 (2.6) | 1 (5.3) | 2 (3.5) |
| Larges Cells Lung Cancer | 0 (0) | 1 (5.3) | 1 (1.8) |
| AJCC tumor stage - n (%) | | | |
| IIIA | 0 (0) | 0 (0) | 0 (0) |
| IV | 38 (100) | 19 (100) | 57 (100) |
| Smoking history - n (%) | | | |
| Never Smoked | 11 (28.9) | 5 (26.3) | 16 (28.1) |
| Current Smoker | 4 (10.5) | 2 (10.5) | 6 (10.5) |
| Former Smoker | 23 (60.5) | 12 (63.2) | 35 (61.4) |
| Specimen type - n (%) | | | |
| Resection | 15 (39.5) | 0 (0) | 15 (26.3) |
| Core needle biopsy | 21 (55.3) | 8 (42.1) | 29 (50.9) |
| Fine needle aspirate | 2 (5.3) | 0 (0) | 2 (3.5) |
| Not recorded | 0 (0) | 9 (47.4) | 9 (15.8) |
| Other | 0 (0) | 2 (10.5) | 2 (3.5) |

All percentages calculated using N as denominator

Table 10. Baseline characteristics between the patients included and excluded from the bridging study for Cohort C.

PMA P160045: FDA Summary of Safety and Effectiveness Data

{27}

| Baseline characteristics | Patients in bridging study N=28 | Patients not in bridging study N=8 | All N=36 |
| --- | --- | --- | --- |
| Age (Years) | | | |
| Mean | 67.0 | 70.6 | 67.8 |
| SD | 8.56 | 17.63 | 11.00 |
| Median | 66.5 | 71.0 | 67.0 |
| Min | 53 | 44 | 44 |
| Max | 86 | 91 | 91 |
| Gender - n (%) | | | |
| Female | 18 (64.3) | 4 (50.0) | 22 (61.1) |
| Male | 10 (35.7) | 4 (50.0) | 14 (38.9) |
| Race - n (%) | | | |
| American Indian Or Alaskan Native | 0 (0) | 0 (0) | 0 (0) |
| Asian | 2 (7.1) | 1 (12.5) | 3 (8.3) |
| Black or African American | 1 (3.6) | 0 (0) | 1 (2.8) |
| Native Hawaiian Or Other Pacific Islander | 1 (3.6) | 0 (0) | 1 (2.8) |
| White | 23 (82.1) | 7 (87.5) | 30 (83.3) |
| Multiple | 0 (0) | 0 (0) | 0 (0) |
| Missing | 1 (3.6) | 0 (0) | 1 (2.8) |
| ECOG baseline - n (%) | | | |
| 0 | 12 (42.9) | 1 (12.5) | 13 (36.1) |
| 1 | 15 (53.6) | 7 (87.5) | 22 (61.1) |
| 2 | 1 (3.6) | 0 (0) | 1 (2.8) |
| Histology - n (%) | | | |
| Adenocarcinoma | 25 (89.3) | 7 (87.5) | 32 (88.9) |
| Adenosquamous Carcinoma - Predominantly adenocarcinoma | 1 (3.6) | 0 (0) | 1 (2.8) |
| Adenosquamous Carcinoma - Predominantly Squamous Cell Carcinoma | 1 (3.6) | 0 (0) | 1 (2.8) |
| Large Cell Carcinoma | 0 (0) | 1 (12.5) | 1 (2.8) |
| Non-Small Cell Carcinoma Without Other | 1 (3.6) | 0 (0) | 1 (2.8) |
| AJCC tumor stage - n (%) | | | |
| IIIA | 1 (3.6) | 0 (0) | 1 (2.8) |
| IV | 27 (96.4) | 8 (100) | 35 (97.2) |
| Smoking history - n (%) | | | |
| Never Smoked | 9 (32.1) | 1 (12.5) | 10 (27.8) |
| Current Smoker | 3 (10.7) | 2 (25.0) | 5 (13.9) |
| Former Smoker | 16 (57.1) | 5 (62.5) | 21 (58.3) |
| Specimen type - n (%) | | | |
| Resection | 10 (35.7) | 0 (0) | 10 (27.8) |
| Core needle biopsy | 17 (60.7) | 1 (12.5) | 18 (50.0) |
| Fine needle aspirate | 1 (3.6) | 2 (25.0) | 3 (8.3) |
| Not recorded | 0 (0) | 2 (25.0) | 2 (5.6) |
| Other | 0 (0) | 3 (37.5) | 3 (8.3) |

All percentages calculated using N as denominator

PMA P160045: FDA Summary of Safety and Effectiveness Data

{28}

Clinically important demographic and prognostic features including ECOG performance status and tumor stage were well balanced between the Oncomine™ Dx Target Test-evaluable and the Oncomine™ Dx Target Test-unevaluable populations.

## D. Safety and Effectiveness Results

### 1. Safety Results

The safety with respect to treatment with TAFINLAR® (dabrafenib) in combination with MEKINIST® (trametinib) will not be addressed in detail in the SSED for the Oncomine™ Dx Target Test. Briefly, in the drug study, the most commonly occurring adverse reactions (≥ 20%) in these patients were: pyrexia, fatigue, nausea, vomiting, diarrhea, dry skin, decreased appetite, edema, rash, chills, hemorrhage, cough, and dyspnea. Adverse reactions resulting in discontinuation of TAFINLAR occurred in 18% of patients; the most common were pyrexia (2.2%), ejection fraction decreased (2.2%), and respiratory distress (2.2%). Adverse reactions leading to dose reductions of TAFINLAR occurred in 35% of patients; the most common were pyrexia (10%), diarrhea (4.3%), nausea (4.3%), vomiting (4.3%), and neutropenia (3.2%). Adverse reactions leading to dose interruptions of TAFINLAR occurred in 62% of patients; the most common were pyrexia (27%), vomiting (11%), neutropenia (8%), and chills (6%). Adverse reactions resulting in discontinuation of MEKINIST occurred in 19% of patients; the most common were pyrexia (2.2%), ejection fraction decreased (2.2%), and respiratory distress (2.2%). Adverse reactions leading to dose reductions of MEKINIST occurred in 30% of patients receiving MEKINIST plus dabrafenib; the most common were pyrexia (5%), nausea (4.3%), vomiting (4.3%), diarrhea (3.2%), and neutropenia (3.2%). Adverse reactions leading to dose interruptions of MEKINIST occurred in 57% of patients receiving MEKINIST plus dabrafenib; the most common were pyrexia (16%), vomiting (10%), neutropenia (8%), nausea (5%), and ejection fraction decreased (5%). Please refer to the drug labels for more information.

In the PMA clinical study, no adverse events were reported in connection with the studies used to support this PMA as the studies were performed retrospectively using banked samples.

### 2. Effectiveness Results

#### a. Concordance Study:

The concordance between the Oncomine™ Dx Target Test and BRAF V600 PCR Mutation Test (surrogate assay) were assessed in the combined Cohorts A, B and C as well as in Cohort B and C separately.

The agreements between the Oncomine™ Dx Target Test and the BRAF V600 surrogate assay are shown in the table below for the combined cohorts.

PMA P160045: FDA Summary of Safety and Effectiveness Data
Page 29

{29}

Table 11. Agreements between Oncomine™ Dx Target Test and PCR comparator method results in combined cohorts A, B and C.

| Measure of agreement | Including No Calls | | Excluding No Calls | |
| --- | --- | --- | --- | --- |
| | % agreement (N) | 95% CI¹ | % agreement (N) | 95% CI¹ |
| PPA | 91.8% (67/73) | (83.0%, 96.9%) | 100.0% (67/67) | (94.6%, 100.0%) |
| NPA | 97.4% (114/117) | (92.7%, 99.5%) | 100.0% (114/114) | (96.7%, 100.0%) |
| OPA² | 95.3% (181/190) | (91.2%, 97.8%) | 100.0% (181/181) | (97.9%, 100.0%) |

¹ 95%CI calculated using Pearson-Clopper Exact method
² Overall Percent Agreement

A subset of LLT positive [LLT(+) ] specimens were retrospectively tested using the Oncomine™ Dx Target Test (CDx). Since LLT-negative [LLT(-)] specimens were not available for retesting, the acquired PCR(-) cohort samples were used to derive the NPA. The acquired PCR(-) cohort samples were randomly assigned to each study cohort in proportion to the size of the enrolled population of each cohort. The numbers of Acquired PCR(-) samples assigned to each cohort were as follows: 59 for Cohort A, 43 for Cohort B, and 27 for Cohort C.

In the combined cohorts, the agreements between the Oncomine™ Dx Target Test with LLT are shown in the tables below.

Table 12. Agreement between the Oncomine™ Dx Target Test and LLT/Acquired PCR(-) samples tested in combined Cohorts A, B and C.

| Oncomine™ Dx Target Test | Available enrolled LLT(+) samples | Acquired PCR(-) samples | Total |
| --- | --- | --- | --- |
| Positive | 72 | 1 | 73 |
| Negative | 8 | 114 | 122 |
| No call | 8 | 2 | 10 |
| Invalid | 23 | 7 | 30 |
| Total | 111 | 124 | 235 |

Table 13. Agreement between Oncomine™ Dx Test and LLT/Acquired PCR(-) sample results in combined Cohorts A, B and C.

| Agreement Measure | Including No Calls | | Excluding No Calls | |
| --- | --- | --- | --- | --- |
| | % agreement (N) | 95% CI¹ | % agreement (N) | 95% CI¹ |
| PPA | 64.9% (72/111) | (55.2%, 73.7%) | 90.0% (72/80) | (81.2%, 95.6%) |
| NPA | 91.9% (114/124) | (85.7%, 96.1%) | 99.1% (114/115) | (95.3%, 100.0%) |
| OPA | 79.1% (186/235) | (73.4%, 84.2%) | 95.4% (186/195) | (91.4%, 97.9%) |

¹ The 95%CI calculated using Pearson-Clopper Exact method

In Cohort B, the agreements between the Oncomine™ Dx Target Test with LLT are shown in the tables below.

PMA P160045: FDA Summary of Safety and Effectiveness Data

{30}

Table 14. Agreement between the Oncomine™ Dx Target Test and LLT/Acquired PCR(-) samples from Cohort B.

| Oncomine™ Dx Target Test | Available enrolled LLT(+) samples | Acquired PCR(-) samples | Total |
| --- | --- | --- | --- |
| Positive | 22 | 1 | 23 |
| Negative | 2 | 36 | 38 |
| No Call | 5 | 1 | 6 |
| Invalid | 9 | 3 | 12 |
| Total | 38 | 41 | 79 |

Table 15. Agreement between Oncomine™ Dx Test and LLT/Acquired PCR(-) samples based on LLT/Acquired PCR(-) samples results in Cohort B.

| Agreement Measure | Including No Calls | | Excluding No Calls | |
| --- | --- | --- | --- | --- |
| | % agreement (N) | 95% CI^{1} | % agreement (N) | 95% CI^{1} |
| PPA | 75.9% (22/29) | (56.5%, 89.7%) | 91.7% (22/24) | (73.0%, 99.0%) |
| NPA | 94.7% (36/38) | (82.3%, 99.4%) | 97.3% (36/37) | (85.8%, 99.9%) |
| OPA | 86.6% (58/67) | (76.0%, 93.7%) | 95.1% (58/61) | (86.3%, 99.0%) |

1 95% CI calculated using Pearson-Clopper Exact method

In Cohort C, the agreements between the Oncomine™ Dx Target Test with LLT are shown in the tables below.

Table 16. Agreement between the Oncomine™ Dx Target Test and LLT/Acquired PCR(-) samples from Cohort C.

| Oncomine™ Dx Target Test | Available enrolled LLT(+) samples | Acquired PCR(-) Samples | Total |
| --- | --- | --- | --- |
| Positive | 23 | 0 | 23 |
| Negative | 0 | 22 | 22 |
| No call | 2 | 1 | 3 |
| Invalid | 3 | 2 | 5 |
| Total | 28 | 25 | 53 |

Table 17. Agreements between Oncomine™ Dx Test and LLT/Acquired PCR(-) samples in Cohort C.

| Agreement Measure | Including No Calls | | Excluding No Calls | |
| --- | --- | --- | --- | --- |
| | % agreement (N) | 95% CI^{1} | % agreement (N) | 95% CI^{1} |
| PPA | 82.1% (23/28) | (63.1%, 93.9%) | 100.0% (23/23) | (85.2%, 100.0%) |
| NPA | 88.0% (22/25) | (68.8%, 97.5%) | 100.0% (22/22) | (84.6%, 100.0%) |
| OPA | 84.9% (45/53) | (72.4%, 93.3%) | 100.0% (45/45) | (92.1%, 100.0%) |

1 95% CI calculated using Pearson-Clopper Exact method

b. Bridging Study:
The analysis of effectiveness was based on a bridging study conducted using Cohorts B and C of the clinical study which included 87 evaluable patients at

PMA P160045: FDA Summary of Safety and Effectiveness Data

{31}

the time of database lock and samples from the acquired PCR(-) cohort which were randomized as part of the retrospective bridging study.

The analyses of primary efficacies were based on ORR by investigator assessment for Cohorts B and C separately. The efficacies were calculated in the $\mathrm{LLT}(+)$ and the Oncomine™ Dx Target Test positive [CDx(+)] population [denoted as “ $\mathrm{LLT}(+)$ , CDx(+)]” and the $\mathrm{LLT}(+)$ population using Cohorts B and C at the analysis cut-off date of August 8, 2016. Key effectiveness outcomes are presented in the tables…