BIOMERIEUX THXID BRAF ASSAY KIT

P120014 · bioMerieux, Inc. · OWD · May 29, 2013 · Pathology

Device Facts

• Record ID: P120014
• Device Name: BIOMERIEUX THXID BRAF ASSAY KIT
• Applicant: bioMerieux, Inc.
• Product Code: OWD · Pathology
• Decision Date: May 29, 2013
• Decision: APWD
• Device Class: Class 3
• Attributes: Expedited Review

Intended Use

The THxID™ BRAF kit is an In Vitro Diagnostic device intended for the qualitative detection of the BRAF V600E and V600K mutations in DNA samples extracted from formalin-fixed paraffin-embedded (FFPE) human melanoma tissue. The THxID™ BRAF kit is a real-time PCR test on the ABI 7500 Fast Dx system and is intended to be used as an aid in selecting melanoma patients whose tumors carry the BRAF V600E mutation for treatment with dabrafenib [Tafinlar®] and as an aid in selecting melanoma patients whose tumors carry the BRAF V600E or V600K mutation for treatment with trametinib [Mekinist™].

Device Story

Device detects BRAF V600E and V600K mutations in FFPE melanoma tissue; utilizes nucleic acid isolation, real-time PCR amplification, and automated detection. Input: DNA extracted from FFPE skin/lymph node tissue. Process: Amplification Refractory Mutation System (ARMS) technology; Taq DNA polymerase distinguishes mutant/wild-type alleles; fluorescent probes detect BRAF exon 15 mutations and exon 13 internal control. Output: Qualitative mutation status report generated by THxID™-BRAF software. Used in clinical laboratories; operated by trained personnel. Output informs physician selection of patients for dabrafenib or trametinib therapy. Benefits: Enables targeted therapy for patients with specific BRAF mutations, improving clinical outcomes in advanced/metastatic melanoma.

Clinical Evidence

PMA approval based on retrospective bridging studies using specimens from Phase 3 BREAK-3 (dabrafenib) and METRIC (trametinib) clinical trials. Analytical concordance between THxID™ BRAF and clinical trial assays (CTA) showed >95% overall agreement. Clinical concordance demonstrated that PFS hazard ratios for patients identified by THxID™ BRAF were consistent with those identified by the CTA in the randomized populations.

Technological Characteristics

Real-time PCR platform (ABI 7500 Fast Dx). Materials: FFPE tissue, reagents for nucleic acid isolation, primer pairs, oligonucleotide probes with fluorescent dyes. Connectivity: Standalone system with dedicated computer for THxID™-BRAF software. Software: Rule-based interpretation of Ct values and ΔCt calculations. Sterilization: Not applicable (reagents).

Indications for Use

Indicated for qualitative detection of BRAF V600E and V600K mutations in FFPE human melanoma tissue DNA. Used to select patients with BRAF V600E mutation for dabrafenib treatment and patients with BRAF V600E or V600K mutations for trametinib treatment.

Regulatory Classification

Identification

The somatic gene mutation detection system is an in vitro diagnostic nucleic acid amplification and real-time detection system for the identification of genetic mutations in DNA extracted from human tumor tissues.

Submission Summary (Full Text)

{0}

SUMMARY OF SAFETY AND EFFECTIVENESS DATA (SSED)

I. GENERAL INFORMATION

Device Generic Name: Real Time PCR

Device Trade Name: THxID™ BRAF Kit for use on the ABI 7500 Fast Dx Real-Time PCR Instrument

Device Procode: OWD

Applicant’s Name and Address: bioMérieux, Inc.
595 Anglum Rd.
Hazelwood, MO 63042

Date(s) of Panel Recommendation: None

Premarket Approval Application (PMA) Number: P120014

Date of FDA Notice of Approval: May 29, 2013

Expedited: Granted priority review status on August 31, 2012 because the availability of the device is in the best interest of the patients.

II. INDICATIONS FOR USE

The THxID™ BRAF kit is an In Vitro Diagnostic device intended for the qualitative detection of the BRAF V600E and V600K mutations in DNA samples extracted from formalin-fixed paraffin-embedded (FFPE) human melanoma tissue. The THxID™ BRAF kit is a real-time PCR test on the ABI 7500 Fast Dx system and is intended to be used as an aid in selecting melanoma patients whose tumors carry the BRAF V600E mutation for treatment with dabrafenib [Tafinlar®] and as an aid in selecting melanoma patients whose tumors carry the BRAF V600E or V600K mutation for treatment with trametinib [Mekinist™].

III. CONTRAINDICATIONS

None.

IV. WARNINGS AND PRECAUTIONS

The warnings and precautions can be found in the THxID™ BRAF Kit labeling.

V. DEVICE DESCRIPTION

The BRAF assay is intended for the detection of V600E (T1799A) and V600K (GT1798-

PMA P120014: FDA Summary of Safety and Effectiveness Data
Page 1

{1}

1799AA) mutations in exon 15 of the BRAF oncogene in DNA samples extracted from formalin-fixed, paraffin-embedded (FFPE) melanoma skin tissue and invaded lymph nodes. The following components comprise the overall device:

- THxID™ BRAF PUR Kit – provides reagents for nucleic acid isolation from FFPE tissue sections.
- THxID™ BRAF AMP Kit – provides reagents for automated real-time PCR amplification and detection of target DNA present in the extracted nucleic acids using primer pairs and oligonucleotide probes labeled with different fluorescent dyes.
- Applied BioSystems® 7500 Fast Dx Real-time instrument with the Sequence Detection System (SDS) Software version 1.4 with THxID™ BRAF templates.
- THxID™ -BRAF Software – BRAF assay specific software converts the SDS file into a B-RAF mutation test report on a separate computer for this purpose.

**Specimen Requirements and DNA extraction:** The claimed specimens for this assay are formalin-fixed paraffin embedded melanoma skin and lymph node specimens. At least 20 mm² of tissue when using a 10 μM section, or 40 mm² of tissue when using a 5 μM section, is required to perform the assay. FFPE tissue sections are macro-dissected with a scalpel according to the pathologists indication if less than 80% of the tumor cells are present in the section, or if the section contains necrotic tissue, fatty tissue, hemorrhagic tissue, or non-tumor, melanin rich area. The tissue is deparaffinized and lysed by incubation with proteinase K to digest tissue and enhance removal of formaldehyde. The solution is centrifuged through a column to remove contaminants from the sample. The DNA is washed and eluted following a short incubation with sodium azide solution. The DNA is eluted in a total volume of 55 μL to 60 μL.

**Amplification and Detection:** The THxID™ BRAF assay uses Amplification Refractory Mutation System (ARMS) technology which selectively amplifies the targeted mutant allele when the terminal 3′-end of the primer is hybridized to the target DNA sequence. The Taq DNA polymerase distinguishes between the match and mismatch at the 3′-end of each unique forward primer. In the presence of a wild-type sequence, the forward primers will not completely hybridize and the amplification will not occur. One primer pair is specific to the BRAF gene and allows the amplification of a non-polymorphic gene area of exon 13, which is used as an internal control (IC). Two primer pairs are specific to the mutations V600E and V600K respectively and allow the amplification of mutated fragments leading to the identification of these BRAF mutations in exon 15. The assay is not designed to detect the wild-type sequence for codon 600. Two different target probes, each labeled with a different dye, allow the simultaneous detection of the BRAF internal control and a BRAF mutation if present. The amount of DNA present in the sample is determined by the amount of fluorescence detected by the instrument during allele amplification and is expressed as a function of Ct values. Crossing threshold (Ct) values for BRAF mutant and internal control (IC) are calculated using the ABI Instrument automatic baseline and automatic threshold features of the SDS Software version 1.4 (auto Ct).

PMA P120014: FDA Summary of Safety and Effectiveness Data
Page 2

{2}

Instrument and Software: The $\mathrm{THxID}^{\mathrm{TM}}$ BRAF assay is run on the Applied Biosystems® 7500 Fast Dx Real-Time PCR Instrument with the Sequence Detection System (SDS) Software version 1.4. This system is a real-time nucleic acid amplification and detection platform that measures fluorescence signals and converts them into comparative quantitative readouts using fluorescent detection of dual-labeled hydrolysis probes. The $\mathrm{THxID}^{\mathrm{TM}}$ BRAF assay uses the automatic baseline and threshold features of the SDS Software version 1.4 [Auto Crossing threshold (Ct)] to analyze the data obtained after the amplification run. The use of these features is mandatory for the editing of the results report, i.e., any manual change of these settings by the user will be detected and lead to an error preventing the issuance of a diagnostic result.

Four $\mathrm{THxID}^{\mathrm{TM}}$ BRAF templates configure the instrument for a BRAF run, reducing the risk of misuse of the instrument and wrong positioning of samples. The proper template is chosen by the user according to the number of clinical samples to be tested in the run. The $\mathrm{THxID}^{\mathrm{TM}}$ BRAF templates file is loaded on the computer that runs the instrument and on which the SDS software is installed. The user creates an .sds run file for the instrument according to the instructions given in the Package Insert.

The $\mathrm{THxID}^{\mathrm{TM}}$ BRAF Software gives an interpretation of data obtained from the ABI 7500 Fast Dx Real-Time PCR instrument and SDS software version 1.4. The $\mathrm{THxID}^{\mathrm{TM}}$ BRAF Software imports raw data from the specified SDS file, checks the validity of the run and automatically interprets the results of the SDS file created by the SDS Software by indicating whether the mutation has been detected or not for each clinical sample. It uses an algorithm to interpret the patient samples results and generates a PDF formatted report with qualitative assessment of the BRAF mutation status for patient samples. The interpretation of the sds file requires 2 steps: (1) Transfer a copy of the sds result file to the $\mathrm{THxID}^{\mathrm{TM}}$ BRAF computer; and (2) Generate the $\mathrm{THxID}^{\mathrm{TM}}$ BRAF Mutation Test Report (PDF formatted report).

Test Controls: Controls are used throughout the entire procedure from sample preparation to PCR to ensure the performance of the $\mathrm{THxID}^{\mathrm{TM}}$ BRAF test.

Internal Control (IC): The internal control is a non-polymorphic region of exon 13 in the BRAF gene. This target should be detected (within a pre-established acceptance range) in every sample reaction. It controls the cell lysis efficiency, DNA extraction and purification efficiency, and absence of PCR inhibitors.

Positive Control: The positive control is composed of a mixture of V600E and V600K and WT plasmids. The positive control is carried through the entire PCR process in each run to detect $\mathrm{THxID}^{\mathrm{TM}}$ BRAF assay failures (reagents and instruments). In the event of the positive control failing, the whole run is deemed invalid (no result is provided for clinical samples) All samples and controls should be re-tested.

Negative Control: The negative control (i.e., a no template control) consists of an empty tube that follows the whole process beginning with the xylene treatment step. The negative control is performed in each run to control the absence of

PMA P120014: FDA Summary of Safety and Effectiveness Data
Page 3

{3}

contamination in the whole process. The expected result is a negative outcome. Failure of the negative control invalidates the whole run.

## Acceptable Ct Values for Positive and Negative Run Controls

| | Positive Control | Negative Control |
| --- | --- | --- |
| Internal Control (IC) Ct values | 26.5 < Ct value <31.5 | Ct value > 37 |
| | AND | AND |
| Mutant target Ct values (V600E or V600K) | 28.9 < Ct value < 35.7 | Ct value > 39 |
| | AND | AND |
| ΔCt values | 1.0 < Ct value < 5.6 | Not applicable |

**Interpretation of Results:** The $\mathrm{THxID^{TM}}$ BRAF software interprets the results automatically and highlights the presence of valid or invalid results in the generated report. The 2 possible outcomes for Positive and Negative Controls are "Valid" or "Invalid". The result validity of a clinical specimen is determined first by the Internal Control (IC) Ct value that should fall within pre-specified limits to guarantee that the samples contained adequate DNA. If a specific amplification is detected for the mutant target, a $\Delta \mathrm{Ct}$ value is calculated by subtracting the IC Ct from the Ct value of the mutant reaction (V600E or V600K).

## Ct Values Used to Interpret Results

| IC Ct values | | Mutant target Ct values (V600E or V600K) | | ΔCt values (Ct mutant - Ct IC) | Result for the considered mutation |
| --- | --- | --- | --- | --- | --- |
| Ct ≤ 20 | Values not considered when Ct ≤ 20 | | | | Invalid |
| 20 < Ct < 31.5 | AND | Ct < 38.5 | AND | ΔCt ≤ -10 | Invalid |
| | | | | -10 < ΔCt < 7 | Mutant detected |
| | | | | ΔCt ≥ 7 | Mutant not detected |
| | | Ct ≥ 38.5 or no Ct* | AND | ΔCt not considered | Mutant not detected |
| 31.5 ≤ Ct < 32.5 | AND | Ct < 38.5 | AND | ΔCt ≤ -10 | Invalid |
| | | | | -10 < ΔCt < 7 | Mutant detected |
| | | | | ΔCt ≥ 7 | Invalid |
| | | Ct ≥ 38.5 or no Ct* | AND | ΔCt not considered | Invalid |
| Ct ≥ 32.5 or no Ct* | Values not considered when Ct IC ≥ 32.5 | | | | Invalid |

The final result is based on the $\Delta \mathrm{Ct}$ value: $(\mathrm{Ct~mutant} - \mathrm{Ct~IC}) = \Delta \mathrm{Ct}$

- If the $\Delta \mathrm{Ct}$ value is below the threshold value then a V600E or V600K BRAF mutation is present,
- If the $\Delta \mathrm{Ct}$ value is above the threshold value then no V600E or V600K BRAF mutation is present or it is below the limit of detection.
- If no specific amplification is detected for the mutant targets (V600E and V600K not detected), the sample will be characterized as BRAF mutation-negative.

## Result Reports

PMA P120014: FDA Summary of Safety and Effectiveness Data
Page 4

{4}

| Reported Result | Description |
| --- | --- |
| BRAF mutation-negative | V600E and V600K mutations not detected |
| V600E mutation | V600E mutation detected |
| V600K mutation | V600K mutation detected |
| V600E and V600K mutations | V600E and V600K mutations detected |
| Invalid | Run was invalid |

VI. ALTERNATIVE PRACTICES AND PROCEDURES

There are no other FDA-cleared or approved alternatives for the testing of formalin-fixed, paraffin-embedded melanoma tissue for BRAF V600E and BRAF V600K mutation status in the selection of patients who are eligible for dabrafenib (Tafinlar®) and tramatenib (Mekinist™) treatment.

VII. MARKETING HISTORY

The THxID™ BRAF Kit 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 BRAF test results and subsequently improper patient management decisions on melanoma treatment. For the specific adverse events that occurred in the clinical studies, please see Section X below.

IX. SUMMARY OF PRECLINICAL STUDIES

Formalin-fixed, paraffin-embedded (FFPE) specimens used in the analytical performance studies were assessed for tumor content (% tumor cells), melanin content, and presence of necrotic tissue by a pathological review. The genetic status of the samples on the V600 locus was determined by bi-directional Sanger sequencing. All samples were macrodissected when the tumor content in the sample was less than 80%, or diluted consistent with instructions to users. Additionally, because melanin inhibition may lead to an invalid result, for samples with invalid results, a 1:4 dilution of the sample was made and retested, as indicated in the instructions for use.

A. Laboratory Studies

1. Correlation to Reference Method

The primary study objective was to evaluate the performance of the THxID™ BRAF test in the detection of BRAF V600E and V600K mutations in clinical samples in malignant melanoma specimens collected from clinical trials with dabrafenib and tramatenib by assessing the positive percent agreement (PPA), negative percent agreement (NPA) and overall percent agreement (OPA) with bidirectional sequencing Sanger as a reference method. Consecutive samples were

PMA P120014: FDA Summary of Safety and Effectiveness Data

{5}

collected and compared to bi-directional Sanger sequencing results. The number of samples was selected on the basis of supporting a pre-specified point estimates for PPA and NPA. Pre-specified acceptance criteria for the sequencing Phred score were $&gt;40$ . There were 898 samples available for testing. Excluding all invalids and QNS samples (total 43) there were 35 discordant cases $[35/(898-43) = 4.1\%]$ . Two samples determined to be V600D were detected by the $\mathrm{THxID}^{\mathrm{TM}}$ BRAF assay as V600E. The overall results are shown in the table below:

Agreement Between THxID - BRAF assay and Bi-directional Sequencing for All Samples

| | | Bi-directional Sequencing | | | | | | | |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| THxID-BRAF Result | BRAF V600 mutations | | | E and K mutations not detected | | | | | |
| | | V600E | V600K | V600D | V600R | WT | Invalid1 | QNS2 | Total |
| | V600E | 341 | 2 | 2 | | 21 | 7 | | 373 |
| | V600K | 1 | 57 | | | 2 | | | 60 |
| | V600E&K | | 2 | | | 2 | | | 4 |
| | E and K mutations negative | 6 | 2 | | 11 | 406 | 5 | | 430 |
| | Invalid1 | 6 | 1 | | | 20 | 2 | | 29 |
| | QNS2 | | | | | 2 | | 1 | 3 |
| | Total | 354 | 64 | 2 | 11 | 453 | 14 | | 899 |

${}^{1}$ No result was obtained; ${}^{2}\mathrm{{QNS}} =$ Quantity Not Sufficient for testing

For the purposes of analyzing agreement between the $\mathrm{THxID^{TM}}$ BRAF kit and Sanger sequencing, any specimen that was deemed E or K was considered mutation positive and any sequencing result not E or K was deemed E and K mutation negative. Analyses were conducted with and without the $\mathrm{THxID^{TM}}$ BRAF assay invalids and QNS samples. Agreement was not impacted by specimen type or melanin content (data not shown). The results met the prespecified criteria.

Agreement between the $\mathrm{THxID^{TM}}$ BRAF assay and Sanger sequencing for all samples

| | Including THxID™ BRAF invalids | Without THxID Invalids |
| --- | --- | --- |
| | No. of concordance / No of tests (%) [95% CI] | No. of concordance / No of tests (%) [95% CI] |
| Positive Percent Agreement (PPA) for V600E and V600K | 403/418 96.4% [94.2%; 97.8%] | 403/411 98.1% [96.2%; 99.0%] |
| Negative Percent Agreement (NPA) | 417/464 89.9% *[86.8%; 92.3%] | 417/444 93.9% [91.3%; 95.8%] |

PMA P120014: FDA Summary of Safety and Effectiveness Data

{6}

| | Including THxID™ BRAF invalids | Without THxID Invalids |
| --- | --- | --- |
| | No. of concordance / No of tests (%) [95% CI] | No. of concordance / No of tests (%) [95% CI] |
| Overall Agreement | 820/882 92.3% [91.1%; 94.5%] | 820/855 95.9% [94.4%; 97.0%] |

The accuracy of the V600E and V600K allele were individually assessed. $\mathrm{THxID^{TM}}$ BRAF invalids were included in this analysis (QNS and Sanger invalids excluded). 1Negative agreement for V600E was based on the total non-V600E alleles. Negative agreement for V600K was based on the total non-V600K alleles. Agreement was not impacted by specimen type (data not shown). Accuracy for the V600E allele was $96.3\%$ and accuracy for the V600K alleles was $92.2\%$ .

| | V600E –with invalids | V600K –with invalids |
| --- | --- | --- |
| | No. of concordance / No of tests (%) [95% CI] | No. of concordance / No of tests (%) [95% CI] |
| Positive Percent Agreement (PPA) | 341/354 96.3% [(93.8, 97.8] | 59/64* 92.2 (79.7; 94.7) |
| Negative Percent Agreement (NPA) | 503/528 99.2% (93.1, 96.8) | 813/817 99.5 (98.8 to 99.8) |

*Two samples with V600E&amp;K and Sanger shows V600K, and Sanger can only report one mutation

# 2. Analytical Sensitivity

# a) Limit of Blank (LoB) - No Template/Exclusivity

To assess performance of the $\mathrm{THxID^{TM}}$ BRAF Kit in the absence of the template and to ensure that sample with wild-type (WT) DNA or a sample with a V600E or V600K mutation does not generate an analytical signal that results in a false result, cell lines and clinical FFPE melanoma samples covering all testing conditions (i.e., WT, V600E and V600K for skin &amp; lymph node) were tested at high DNA concentrations to determine the level of background amplification in the absence of the target. A total of 3 cell lines and 6 clinical samples were tested at the highest DNA concentration per reaction $(350\mathrm{ng} / \mu \mathrm{L}$ ; total $750\mathrm{ng}$ ) claimed. A total of 60 replicates per sample were evaluated over 3 runs and one assay reagent lot. Skin samples were limited to $150\mathrm{ng} / \mu \mathrm{L}$ and 20 replicates due to lack of tissue. No background amplification of the unexpected target DNA occurred. Ct values are reported as undetermined for the appropriate reactions. Four results were reported as invalid and the remaining results reported as "mutant not detected" and no unexpected amplification or wrong results occurred

PMA P120014: FDA Summary of Safety and Effectiveness Data

{7}

b) Limit of Detection (LoD)

The claimed range for DNA input for the assay is 20-700 ng DNA per reaction. For the THxID™ BRAF Kit, the limit of detecting mutant DNA in a background of wild-type DNA is defined as the lowest mutation level in a specimen for which the assay yields a positive result in 95% of the test replicates for each mutation positive sample. Determination of the LoD was performed by testing (1) melanoma cell line FFPE blocks; and (2) FFPE melanoma skin and lymph node clinical specimens. Two different test lots were used for these evaluations.

Clinical samples: LoD values were calculated independently for both lymph and skin specimens. LoD values were determined using serial dilutions of DNA extracted from macrodissected mutant specimens blended with DNA extracted from wild-type specimens. Serial dilutions from three DNA concentrations spanning the claimed input range were evaluated (20ng, 350 ng, and 700ng) for V600E and two concentrations for V600K (because a high concentration specimen could not be obtained.) Mutant samples were presumed to be 100% mutant based on the ΔCt value. Mutant content was diluted from this value.

The LoD was estimated by calculating the 95% predicted hit rate using Probit analysis. Results for skin and lymph node were similar The LoD was determined as the highest value obtained. The data support a claimed LoD of 5% mutant DNA in a background of wild-type DNA for V600E and V600K positive FFPE skin and lymph node specimens across the DNA input range.

FFPE cell line DNA: Dilution series using DNA extracted from FFPE cell lines with either the V600E or V600K mutation were diluted to a final concentration of 5% mutant in a background of wild-type DNA extracted from FFPE cell line. A total of 24 replicates (12 replicates per lot) at very low DNA input concentration (30ng/reaction) was evaluated for each mutation on two reagent lots. The V600E FFPE cell line was homozygous and the V600K FFPE cell line contained 66.7% mutant DNA. LoD was confirmed to be &lt; 5%.

Confirmatory Testing: DNA was extracted from WT and mutant clinical FFPE specimens, or melanoma cell line FFPE blocks, and combined to create the required samples, each containing 5% mutation. The FFPE specimens were from skin or lymph node tissue. Samples represented a range of total DNA inputs within the claimed range (20-700 ng/reaction) and outside of this range (10 and 1000 ng/reaction). Two FFPE clinical specimens with high melanin contents were included. A total of 20 replicates each were evaluated. For each sample, the 20 replicates were tested in one PCR run. The study included one lot of the THxID™ BRAF assay, which was a different lot than the two used for LoD Determination (above).

For each sample in the study, including clinical FFPE specimens and cell line FFPE blocks, the observed hit rate (mutation positive rate) was 100% (20/20

PMA P120014: FDA Summary of Safety and Effectiveness Data
Page 8

{8}

replicates) [95% confidence interval 83.9% – 100%], meeting the acceptance criteria of ≥ 95% hit rate (≥19/20 replicates) for each sample. A summary of confirmation of LoD results for the clinical FFPE specimens and the melanoma cell line FFPE blocks is shown in the following table.

| Condition | Tissue Type | Mutant allele at 5% | Mutation positive / tested |
| --- | --- | --- | --- |
| 10 ng/reaction (below DNA range) | Lymph node | V600E | 20/20 |
| | Skin | V600K | 20/20 |
| 20 ng/reaction (lower limit of DNA range) | Lymph node | V600E | 20/20 |
| | Skin | V600E | 20/20 |
| | Lymph node | V600K | 20/20 |
| | Skin | V600K | 20/20 |
| High melanin content* | Skin | V600E | 20/20 |
| | Lymph node | V600K | 20/20** |
| 700 ng/reaction (high end of range) | Lymph node | V600E | 20/20 |
| | Skin | V600E | 20/20 |
| | Lymph node | V600K | 20/20 |
| | Skin | V600K | 20/20 |
| 1000 ng/reaction (above DNA range) | Lymph node | V600E | 20/20 |
| | Lymph node | V600K | 20/20 |
| Melanoma FFPE cell line blocks | Melanoma cell line | V600E | 20/20 |
| | Melanoma cell line | V600K | 20/20 |

* DNA input for high melanin samples were at 60 ng/reaction for the V600E sample and 526 ng/reaction for the V600K sample and melanin content were respectively of 80.7% and 73.6%.
** After 1:4 dilution of the eluate. Before dilution each of the 20 replicates was invalid. In accordance with the troubleshooting section, the eluate was diluted 1:4 in Buffer ATE and re-tested with results as described in the table.

c) Genomic DNA Input Range

The claimed genomic input range is 10ng/μL to 350ng/ μL, i.e., a total of 20ng of DNA to 700ng of DNA input when 2μL is used for each reaction. The THxID - BRAF Kit does not require the user to obtain a specific concentration of DNA as determined by spectrophotometry because an internal control Ct value is used to measure adequate DNA integrity. The claimed range was validated two ways: (1) Validation of the DNA input range was shown in the limit of detection, and

PMA P120014: FDA Summary of Safety and Effectiveness Data
Page 9

{9}

linearity studies demonstrated that replicates across the claimed DNA input range produce correct results; and (2) DNA from the 891 eluates from the patient specimens evaluated in the accuracy studies were measured by spectrophotometry prior to testing to ensure that users will typically obtain between 10ng/μL to 350ng/μL per extraction using the range of allowable tumor specimen. The results showed that 94.6% of DNA values were within 10-350ng/μL (total range was 2 to 1764ng/μL). Of these samples, 6.3% had DNA concentrations outside of the claimed range with 25 clinical specimens exceeding the claimed upper limit and 31 specimens had concentrations below the claimed lower limit. Of the 56 results, 8 were invalid. The package insert indicates that the optimal DNA concentration is less than 350ng/reaction.

## 3. Validation of Control Ct Values

The cut-off values for the positive control (PC), negative control (NC) and Internal Control (IC) using the BRAF interpretation algorithm for the THxID™ BRAF assay were validated in a study designed to ensure that errors due to failures occurring during set-up and amplification would be detected by the controls.

**Positive control:** The positive control is composed of a mixture of V600E, V600K and WT plasmids. The positive control is carried through the entire amplification process in each run to detect THxID™ BRAF assay failures (reagents and instruments). In the event of the positive control failure, the run is deemed invalid and no result is provided for clinical samples. Two independent studies were conducted to demonstrate the effectiveness of the positive control by simulating (1) 11 potential user-driven errors that can occur in assay set-up and amplification process using 3 lymph node specimens representing WT, V600E, and V600K, and belned to represent the claimed LoD of 5% mutant in WT ratio; and (2) 6 potential user driven errors using 6 melanoma FFPE specimens blends with approximate LoD equivalent to 15% mutant to WT ratio. The study included the following six melanoma FFPE samples: 2 WT samples (1 lymph node, 1 skin); 2 V600E samples (1 lymph node, 1 skin); 2 V600K samples (1 lymph node, 1 skin). Both positive control and clinical procured specimens were tested with and without the simulated errors. The potential errors that were simulated were as follows:

- Error 1: no vortexing during preparation of the PCR amplification to mimic a breakdown of the vortex-type mixer (steps 2, 6 and 9).
- Error 2: no centrifugation before starting the amplification on the instrument, only tested on 96-well plate (step 20).
- Error 3: Primers &amp; probes sphere re-suspended by master mix instead of dedicated diluent (step1).
- Error 4: substitution of master mix with primer diluent (no Master Mix in the reaction) (step 1).
- Error 5: Master mix frozen for at least 18 hours (recommended storage being +4°C) (step 5).
- Error 6: Use master mix re-suspension volume (110μl) instead of diluent volume (85μl) (step 1).

PMA P120014: FDA Summary of Safety and Effectiveness Data
Page 10

{10}

- Error 7: Primers &amp; probes sphere re-suspended in 60μl instead of 85μl to mimic confusion with the elution volume (step 1).
- Error 8: 5μl of eluate or control added in the reaction instead of 2μl (step 12).
- Error 9: Re-suspend 2 Primers &amp; probes spheres with 255μl to mimic the loss of one sphere in a mix (note in step 1).
- Error 10: Positive control sphere re-suspended in 85μl instead of 150μl to mimic confusion with the Primers &amp; probes sphere re-suspension volume (step 1).
- Error 11: Primers &amp; probes sphere re-suspended in 150μl instead of 85μl to mimic confusion with the control re-suspension volume (step 1).

The Positive Control was considered effective if it was reported as invalid in 4 / 4 replicates when tested under the same error condition that generated an invalid or incorrect result for an FFPE specimen. The results showed that for each of the PCR set-up errors evaluated in both studies, the Positive Control was invalid in 4 of 4 replicates whenever one or more replicates of the FFPE specimen tested under the same condition is incorrect or invalid. The results also showed that the user errors did not impact the results in many cases demonstrating the robustness of the assay overall.

**Negative Control:** The negative control follows the entire assay process, starting from the sample paraffin removal step. The negative control is performed in each run to control the absence of contamination in the whole process. In a valid negative control, no target amplification should be detected (Ct values are undetermined) or, if a late target specific amplification is detected, the corresponding Ct values should be greater than a value to be determined. Validation of the negative control was assessed by intentional contamination of key components with DNA extracted from mutant specimens: the lysis buffer (ATL), the elution buffer (ATE), and the primer diluent. The DNA samples used as contaminating materials were prepared from DNA extracted from procured clinical samples blended to reach 5% mutant to WT ratio and used at a very low concentration to represent a challenging sample. Each test with and without a contaminated component (lysis buffer, elution buffer or primer diluent) was carried out in quadruplicate. The results demonstrated that the negative control is more sensitive to contamination than a WT sample and that no false positive results occurred.

**Internal Control:** The internal control is a non-polymorphic region of exon 13 in the BRAF gene that should be detected (within the pre-established acceptance range) in every sample reaction. It controls the cell lysis efficiency, DNA extraction and purification efficiency, and absence of PCR inhibitors. Two independent studies were conducted using the 6 FFPE melanoma specimens 2 WT samples (1 lymph node, 1 skin); 2 V600E samples (1 lymph node, 1 skin); 2 V600K samples (1 lymph node, 1 skin) to demonstrate the effectiveness of the internal control by simulating 8 different error conditions as follows:

- Error 1: no deparaffinization with the xylene and ethanol before starting the lysis step - Start at the step 11.

PMA P120014: FDA Summary of Safety and Effectiveness Data
Page 11

{11}

- Error 2: no addition of the proteinase K -omit step 12.
- Error 3: no addition of 200μL of the Buffer AL to the sample and no addition of 200μL of ethanol -omit steps 17 to 19.
- Error 4: no addition of 500μL of buffer AW2 (wash buffer) -omit steps 30 to 34.
- Error 5: no centrifugation at full speed (approximately 20000 g) for 3 minutes at 18-25°C to dry the membrane completely and no incubation at 18-25°C for at least 1 minute -omit step 35 and step 40.
- Error 6: perform the extraction without adding ethanol in the buffer AW1 (25mL) and AW2 (30mL).
- Error 7: apply 120 μL instead of 60 μL of Buffer ATE to the center of the membrane -step 38 (it is tested if this step is done 2 times).
- Error 8: spiking of AL buffer (lysis buffer) in the eluate to mimic the presence of inhibitor- add 5μL of AL buffer in the eluate after step 41.

The results demonstrated the effectiveness of the IC control with the exception of error 8 (addition of ATL to the eluate) which failed in generating an invalid result and generated false negative results for the V600E PCR specimens and a V600K PCR specimen. This error condition did not systematically occur and does not reflect a realistic situation (5μL lysis buffer ending in the eluate). While it is very unlikely to occur (as confirmed by the performance studies shown here), a limitation in the package insert highlights the possible error.

## 4. Minimum Specimen Requirements

### a) Tissue Area

A total of 199 clinical specimens with tissue areas ranging from 8-300 mm² were tested with the THxID BRAF assay. The specimens were both skin and lymph node, exhibited melanin content ranging 0 to 100%, were V600E, V600K or V600 wild-type, had varying tumor proportion, and were processed as either 1 x 10μm or 2x 5μm sections. The Ct values for these specimens were compared to the assay cut-offs and used to confirm the minimum quantity of tissue that can be used to give a valid result when compared to bi-directional Sanger sequencing. The observed DNA concentrations of eluates for 199 procured clinical specimens ranged from 5ng/μL to 517ng/μL with an average of 122ng/μL. The claimed tumor area for the assay is 20mm² to 250mm². Smaller tissue (i.e., &lt; 20mm²) areas cannot ensure reliable results.

### b) Thickness

To demonstrate that both 5μm and 10μm sections can be used with the THxID™ BRAF assay, 30 FFPE melanoma specimens were each tested using one 10μm section, and 2 x 5μm sections and testing the eluates of each in parallel. There were 10 V600E (7 lymph and 3 skin), 5 V600K (3 lymph and 2 skin), and 15 wild-type (11 lymph and 4 skin) specimens. Melanin content ranged from 0 to 80%. Acceptance criteria were set to an absolute

PMA P120014: FDA Summary of Safety and Effectiveness Data
Page 12

{12}

mean difference of less than or equal to 1.0 Ct or 1.0 delta Ct between results obtained with the two conditions. Both sections and curls were included in the evaluation depending on tumor content and the requirement for macrodissection. Tumor content ranged from 40-95% prior to macrodissection. The results of the study confirm that the use of one 10 μm section or two 5 μm sections of the same FFPE melanoma sample will give comparable results with the THxID™ BRAF assay. The total surface of tissue should not exceed 250 mm² for 1 10 μm, or 500 mm² if 2 × 5 μm sections are prepared.

c) Tumor Content and Macrodissection

The instructions for use require the user to macrodissect samples with less than 80% tumor content. To determine whether the THxID BRAF assay specifically fails to detect the V600E and V600K mutant alleles when the tumor content is below 80% and the section has not undergone macrodissection, 40 mutation-positive melanoma FFPE specimens with tumor content ranging from 5% to 70% for each specimen type (skin and lymph node) were tested without macrodissection. Of the 40 samples, 33 were V600E (10 skin and 22 lymph) and 7 V600K (3 skin and 4 lymph). Melanin content ranged from 0 to 50% across the specimens. Results were compared to bi-directional Sanger sequencing. All results were correct.

d) Curl vs. Slide Equivalency

To demonstrate that results generated with the THxID™ BRAF assay starting from FFPE sections on slide versus from FFPE sections in tube (curls) are comparable for samples with ≥ 80% tumor content, a total of 30 FFPE melanoma clinical samples containing ≥ 80% tumor cell were processed in parallel as sections on slides and curls in a tube. The study included 14 V600E (10 lymph node and 4 skin), 2 V600K (both lymph node) and 14 wild-type (13 lymph node and 1 skin) samples. Melanin content ranged from 0- to 50%. Ct and ΔCt values obtained from slides and curls on the same sample were compared. The results demonstrated that 29 of the 30 specimens were concordant with less than a mean difference in ΔCt value of 1.0. One wild-type lymph node, curl was false positive for V600E. Further testing of 10 additional curls from the same specimen were all wild-type suggesting contamination in the original sample. The results met the acceptance criterion.

5. Linearity/Amplification Efficiency as a Function of %Mutant DNA

To demonstrate the linearity of the THxID™ BRAF assay across the range of mutant DNA content, a series of dilutions with varying concentrations of mutant DNA in a background of wild-type DNA at various 3 fixed DNA inputs (20 ng, 350 ng and 700 ng per reaction) from both skin and lymph node specimens for each mutation (V600E and V600K) was evaluated. Mutant to wild-type ratios tested were 100%, 50%, 25%, 10%, 5%, 2.5%, and 1.25%. Initial samples were presumed to have mutant content of 100% by using a macrodissected specimen with a ΔCt value

PMA P120014: FDA Summary of Safety and Effectiveness Data
Page 13

{13}

&lt;1 (i.e., mutant DNA content at the highest concentration possible). A pool of wild-type DNA was generated from 3 different samples and used to dilute the mutant DNA to create the ratios. The highest 3 concentrations were tested in 6 replicates while the remaining were tested in 7 replicates. Linear regression, second and third order polynomial regressions were performed on the obtained Ct or delta Ct values and the log10 of the mutant to WT ratios. The acceptance criteria for linearity above the claimed LoD of $5\%$ for skin and lymph node were met, however the lymph node high V600K samples was not linear at the high range. This did not impact the correct calls though. At three, fixed DNA input concentrations, amplification efficiency when the mutant DNA concentration ranges from $1.25\%$ to $100\%$ (presumed) is summarized below:

| | | Intercept (SE) | Intercept (95% CI) | Calculated Slope | Slope (SE) | Two-Sided 95% CI (slope) | Amplification Efficiency |
| --- | --- | --- | --- | --- | --- | --- | --- |
| Skin V600E | high | 31.12 (0.11) | [30.89, 31.34] | -3.13 | 0.08 | [-3.30, -2.96] | 109 % |
| | Med | 30.87 (0.08) | [30.72, 31.02] | -2.77 | 0.06 | [-2.90, -2.64] | 130 % |
| | Low | 34.64 (0.14) | [34.36, 34.92] | -3.01 | 0.11 | [-3.24, -2.78] | 115 % |
| Lymph V600E | high | 32.01 (0.08) | [31.86, 32.16] | -3.86 | 0.06 | [-3.98, -3.73] | 82 % |
| | Med | 29.88 (0.06) | [29.76, 29.99] | -3.06 | 0.05 | [-3.15, -2.96] | 112 % |
| | Low | 35.24 (0.16) | [34.92, 35.56] | -2.92 | 0.13 | [-3.19, -2.66] | 120 % |
| Skin V600K | high | 31.64 (0.19) | [31.26, 32.02] | -3.14 | 0.13 | [-3.41, -2.87] | 108 % |
| | Med | 31.00 (0.08) | [30.83, 31.17] | -3.47 | 0.07 | [-3.61, -3.33] | 94 % |
| | Low | 32.93 (0.09) | [32.75, 33.11] | -3.06 | 0.08 | [-3.22, -2.91] | 112 % |
| Lymph V600K | high | 32.53 (0.10) | [32.33, 32.72] | -4.59 | 0.08 | [-4.74, -4.44] | 65 % |
| | Med | 28.97 (0.09) | [28.80, 29.14] | -3.00 | 0.07 | [-3.15, -2.85] | 115 % |
| | Low | 35.65 (0.16) | [35.32, 35.98] | -3.07 | 0.13 | [-3.33, -2.81] | 112 % |

# 6. Linearity/Amplification Efficiency as a Function of DNA Input

The linearity and amplification efficiency of PCR were evaluated using a dilution series of DNA inputs extracted from FFPE cell line blocks and spanning the DNA range for two fixed BRAF mutations (5% to represent the LoD, 100% for V600E and 66% for the V600K). The DNA inputs tested were 700ng, 350 ng, 175ng, 87.5ng, 43.8ng, 21.9 ng, and 10.9 ng. For each condition, 6 replicates per DNA input were tested with one lot. The $\Delta$ Ct values were used to select the specimens. Linear regression, second and third order polynomial regressions were performed on

PMA P120014: FDA Summary of Safety and Effectiveness Data

{14}

the obtained Ct or delta Ct values and the log10 of the DNA. The acceptance criteria were met. The amplification efficiencies are shown below:

V600E FFPE cell lines:

| Mutation content | Variable | Intercept (SE) | Intercept 95% CI | Slope | Slope (SE) | Slope 95% CI | Efficiency |
| --- | --- | --- | --- | --- | --- | --- | --- |
| 5 % | Ct V600E | 35.43 (0.05) | [35.34, 35.53] | -3.55 | 0.02 | [-3.59, -3.50] | 91 % |
| 5 % | Ct IC V600E | 32.50 (0.09) | [32.32, 32.67] | -3.53 | 0.04 | [-3.62, -3.45] | 92 % |
| 100 % | Ct V600E | 31.78 (0.03) | [31.72, 31.85] | -3.49 | 0.01 | [-3.52, -3.46] | 93 % |
| 100 % | Ct IC V600E | 31.94 (0.06) | [31.82, 32.06] | -3.41 | 0.03 | [-3.47, -3.36] | 96 % |

V600K FFPE cell lines:

| Mutation content | Variable | Intercept (SE) | Intercept 95% CI | Slope | Slope (SE) | Slope 95% CI | Efficiency |
| --- | --- | --- | --- | --- | --- | --- | --- |
| 5 % | Ct V600K | 35.23 (0.04) | [35.15, 35.31] | -3.44 | 0.02 | [-3.48, -3.41] | 95 % |
| 5 % | Ct IC V600K | 32.18 (0.05) | [32.07, 32.29] | -3.50 | 0.03 | [-3.55, -3.45] | 93 % |
| 66.6 % | Ct V600K | 32.23 (0.04) | [32.14, 32.31] | -3.44 | 0.02 | [-3.48, -3.41] | 95 % |
| 66.6 % | Ct IC V600K | 31.51 (0.06) | [31.39, 31.62] | -3.50 | 0.03 | [-3.55, -3.45] | 93 % |

The PCR efficiency and linearity were evaluated for each mutation (V600E and V600K) on lymph nodes and skin as well. The DNA inputs tested were 700ng, 350 ng, 175ng, 87.5ng, 43.8ng, 21.9 ng, and 10.9 ng. For each condition, 6 replicates per DNA input were tested with one lot. Linearity was assessed on the range in which all DNA quantities gave consistent results. The results demonstrated for the clinical specimens evaluated. All calls were correct. The linear ranges observed for clinical samples at high (100%) and low (5%) mutant DNA concentrations are summarized below.

| | Variable | |
| --- | --- | --- |
| Specimen | CT V600mutant | CT ICV600mutant |
| Skin V600E Low | 10.9 - 350 | 10.9 - 700 |
| Skin V600 High | 10.9 - 350 | 10.9 - 700 |
| Skin V600K Low | 10.9 - 175 | 10.9 - 350 |
| Skin V600K High | 10.9 - 350 | 10.9 - 700 |
| Lymph node V600E Low | 10.9 - 700 | 10.9 - 700 |
| Lymph node V600E High | 10.9 - 700 | 10.9 - 700 |
| Lymph node V600K Low | 10.9 - 87.5 | 10.9 - 350 |
| Lymph node V600K High | 10.9 - 350 | 10.9 - 700 |

PMA P120014: FDA Summary of Safety and Effectiveness Data

{15}

PMA P120014: FDA Summary of Safety and Effectiveness Data
Page 16

# 7. Analytical Specificity

## a) Primer and Probe Specificity

An in silico study was performed to evaluate primer and probe specificity using publicly available sequence and mutation databases. Two different types of analysis were performed (1) the potential for cross-reactivity to primer pairs and probe sequences was investigated using public databases and specialized multiple sequence alignment software, and (2) amplification of all homologous sequences in Genbank were predicted in a simulation analysis using specialized bioinformatics software designed for this purpose. The results showed that amplification of non-BRAF genes (including human, fungal, bacterial and viral), BRAF pseudogenes, and ARAF/CRAF genes is not predicted, however, the assay is expected to cross-react to the V600D mutation in BRAF exon 15. The results also showed that other BRAF mutations reported in somatic mutation databases may impact detection, but none of these mutations were reported as occurring in melanoma. The results support the specificity of the assay primer and probes.

## b) Inclusivity

The THxID™ BRAF assay is designed to detect the V600E (T1799A) and V600K (GT1798/1799AA) mutations. In addition, the THxID™ BRAF assay was shown to detect a rare form of the V600E mutation (i.e., rare codon GAA) and the V600E/K601E mutation (also referred to as V600E2). using 2 FFPE lymph node specimens and plasmids. Two FFPE samples from lymph node tissues with the rare V600E allele as determined by bidirectional Sanger sequencing were tested with 3 replicates each. Melanin content was 5% and 50%. All results were correctly detected. The sample was then combined with WT DNA to obtain a 5% mutant DNA content in background of wild-type DNA. The sample was tested with 20 replicates. All results were called correctly. The expected results for V600E and V600K mutations are as follows:

| | V600E result | V600K result |
| --- | --- | --- |
| V600E (GAG) | Mutant detected | Mutant not detected |
| V600K | Mutant not detected | Mutant detected |
| V600E rare (GAA) | Mutant detected | Mutant not detected |
| V600E/K601E | V600E detected | Mutant not detected |

## c) Cross-Reactivity

i. Non V600E mutations: Cross-reactivity of the THxID –BRAF assay was assessed by testing other non-V600E and non-V600K BRAF mutations using plasmids and clinical samples. Plasmids containing the Internal Control PCR target (region of BRAF gene in exon 13) and the

{16}

mutant PCR target for the following variants were tested: V600, V600D, V600E/K601E (also referred to as V600E2), V600R, V600L, V600M, V600G, V600A and the BRAF pseudogene (a homologous gene present on the X chromosome). Plasmids were spiked into ATE buffer to mimic a gene copy number equivalent to approximately 700ng of DNA (the maximum claimed range). Two clinical samples (one skin and one lymph node) characterized as V600R by Sanger sequencing were also tested. Three replicates of each variant were tested. The results demonstrated that none of the variants with the exception of V600D and V600E/K601E, cross-reacted with the V600E or V600K PCR reactions. Cross-reactivity for V600D and V600E/K601E, persisted following dilutions or plasmids with wild-type plasmid at ratios of $50\%$ , $5\%$ , $2\%$ , and $1\%$ . PCR signals were observed for the V600G plasmid, however, the signal was sufficiently high that a false result did not occur.

| | V600E result | V600K result | Cross-reactivity detected |
| --- | --- | --- | --- |
| BRAF pseudogene | Mutant not detected | Mutant not detected | No |
| V600D plasmid | V600E detected | Mutant not detected | Yes |
| V600R plasmid | Mutant not detected | Mutant not detected | No |
| V600R skin sample | Mutant not detected | Mutant not detected | No |
| V600R lymph sample | Mutant not detected | Mutant not detected | No |
| V600L plasmid | Mutant not detected | Mutant not detected | No |
| V600M plasmid | Mutant not detected | Mutant not detected | No |
| V600G plasmid | Mutant not detected | Mutant not detected | No |
| V600A plasmid | Mutant not detected | Mutant not detected | No |

ii. Microorganisms: A study was conducted to assess whether there are risks of wrong results when a specimen is exposed to microorganism contamination. Five (5) common human microorganisms (E. coli, S. aureus, P. aeruginosa, A. niger/brasiliensis, and C. albicans) along with 5 of the most prevalent microorganisms (including 3 skin organisms) detected in the manufacturing facilities (M. luteus, R. picketti, K. vairans, P. fluorescens, and S. paucimobilis) were investigated. The primer and probes of the kit were spiked with a cocktail of the 10 microorganisms to a final concentration of 104 and $102\mathrm{CFU / mL}$ . No cross detection of THxID -BRAF false results due to contaminants was observed. An in silico analysis was also performed to check the risks of cross-detection, including 6 additional skin microorganisms (S.epidermis, S. aureus, C. xerosis, C. jeikeiu, C. minutissimum, and C. ulcerans). There was no cross-detection predicted.

# 8. Interference - Melanin

PMA P120014: FDA Summary of Safety and Effectiveness Data

{17}

The effect of melanin as an interfering substance on the invalid rate and accuracy of the $\mathrm{THxID^{TM}}$ BRAF assay was evaluated using 56 FFPE samples (30 skin and 26 lymph node) containing melanin levels ranging from $50 - 100\%$ (as determined by pathology review). All three genotypes were represented (i.e., samples were either wild-type (WT), V600E or V600K). Samples with invalid results were tested in accordance with the instructions for use which indicates that first a 1:4 dilution of the extracted sample be tested and if that fails a new section of the sample be reextracted using higher tissue amount. Specimens were macro-dissected according to protocol. Out of the 56 samples tested, a total of 16 samples had invalid results after the first test; 9 lymph node samples $(34.6\%)$ and 7 skin samples $(23.3\%)$ . Nine (9) of these 16 sample $(16.1\%)$ had invalid results not resolved with retesting (4 lymph node and 5 skin). Eight (8) of the 9 samples had melanin content over $80\%$ .

| Melanin Content (%) | Invalid Samples | Valid Samples* | Total No. samples |
| --- | --- | --- | --- |
| 50 | 1 | 11 | 12 |
| 60 | 0 | 5 | 5 |
| 70 | 0 | 8 | 8 |
| 80 | 2 | 12 | 14 |
| 90 | 5 | 10 | 15 |
| 100 | 1 | 1 | 2 |
| Total : 56 | | | |

*After troubleshooting per IFU.

For the 48 samples with Sanger results, there were no false negatives (i.e., all V600E and V600K samples correctly called). There were 5 samples deemed wild-type by Sanger that were called V600E positive by the $\mathrm{THxID^{TM}}$ BRAF assay which is attributed to the more sensitive $\mathrm{THxID^{TM}}$ BRAF assay. To confirm melanin does not generate false results, an additional 5 highly-pigmented, non-melanoma FFPE skin samples (i.e., BRAF V600 wild-type) from African Americans were evaluated. For four of five samples, the original triplicate results (from independent extractions of each specimen) agreed with the expected WT result determined by bi-directional Sanger sequencing. One specimen gave invalid results for two of three replicates. The instructions for use indicate that samples with high melanin may interfere with the assay.

# 9. Interference - Effects of Necrotic Tissue

The ability of the $\mathrm{THxID^{TM}}$ BRAF Kit to perform correctly when samples have high necrotic tissue content was evaluated. A total of 21 melanoma FFPE specimens (skin and lymph node; V600E, V600K or wild-type) with necrotic tissue concentrations ranging from $15\%$ to $60\%$ were tested with the $\mathrm{THxID^{TM}}$ BRAF assay. Necrotic tissue content was determined by pathology review. Each sample was tested in 3 replicates (from 3 extractions) with one lot of $\mathrm{THxID^{TM}}$ BRAF Kit. Melanin content ranged from 0 to $90\%$ . For 20 of the 21 samples with necrotic tissue, the initial $\mathrm{THxID^{TM}}$ BRAF results agreed with the sequencing result. For one sample (V600K with $50\%$ necrotic tissue and $80\%$ melanin), the original $\mathrm{THxID^{TM}}$ BRAF result was invalid for one of three replicates. In accordance with the IFU, the

PMA P120014: FDA Summary of Safety and Effectiveness Data

{18}

residual eluates of the three replicates were diluted 1:4. The re-test results agreed with expected result for three of three replicates. The study demonstrates that necrotic tissue does not interfere with the assay.

## 10. Interference – Endogenous Substances (Hemoglobin and Triglycerides)

To evaluate the impact of high concentrations of hemoglobin and triglycerides (i.e., mimic grossly hemolytic or fatty specimens), two concentrations of hemoglobin (2 mg/mL and 4 mg/mL) or triglycerides (37mM and 74mM) were added to samples during the lysis step (i.e., into lysis buffer prior following deparaffinization and prior to extraction). These concentrations reflect 1x and 2x the CLSI recommended high concentration. Eleven (11) FFPE skin and lymph melanoma tissues were tested with each interferent. Specimens were BRAF V600E, V600K and wild-type and the ΔCt spanned the range to include specimens with low mutant content. Each sample was tested in 3 replicates (from 3 extractions). The results with interferent were compared to the results without interferent. All observed results agreed with the expected result determined by Sanger sequencing across all samples and replicates. There was also no significant change to the ΔCts observed between samples with interferent and without interferent.

## 11. Precision – Reproducibility

An external study was performed to assess the reproducibility of the THxID™ BRAF assay across 3 external testing sites (2 operators per site), 2 runs per operator and 3 non-consecutive testing days across a 3-week period, un duplicate with two types of samples: (1) a panel consisting of 15 samples using prepared DNA eluates, and (2) FFPE melanoma tissues to evaluate the impact of the extraction method across multiple users on the results. DNA eluates were prepared by blending DNA extracted from FFPE skin or lymph specimens to achieve mutant DNA concentrations near the limit of detection (~5%) or 3x the LoD (~15%-50%) representing V600 wild-type, V600E or V600K. Additionally, three levels of total DNA input were included in the evaluation [low (20-30ng), high (600-700ng/reaction), and medium (approximately 60ng – 400ng/reaction)]. Samples were blinded to users. Three samples had high melanin content. Three reagent lots and three PCR thermocyclers were included in this evaluation. The results of the overall agreement were 100% across all samples in the panel except the sample that was derived from skin with a V600E mutant DNA content blended to be very close to the LoD. Further analysis of this sample revealed the mutant content to be below the LoD of 5%. Overall agreement is shown below.

Overall Agreement Between Three Sites for Reproducibility Panel

| Panel Member | Specimen type | DNA input | Percent mutant* | Dilution Step | No. of valid tests all three sites/ Total number of tests (95% CI) |
| --- | --- | --- | --- | --- | --- |
| Wild-Type | Skin | Low | n/a | n/a | 72/72 (94.9% - 100%) |
| Wild-Type | Skin | High | n/a | n/a | 72/72 (94.9% - 100%) |
| Wild-Type | Lymph node | Low | n/a | n/a | 72/72 (94.9% - 100%) |
| Wild-Type | Lymph node | High | n/a | n/a | 72/72 (94.9% - 100%) |
| Wild-Type | Skin – | Med | n/a | Diluted | 72/72 (94.9% - 100%) |

PMA P120014: FDA Summary of Safety and Effectiveness Data

{19}

| | high melanin | | | | |
| --- | --- | --- | --- | --- | --- |
| V600E | Skin – high melanin | Med | - | Diluted | 72/72 (94.9% - 100%) |
| V600E | Skin | Low | Close to LoD | n/a | 60/72 (73.1% -90.2%) |
| V600E | Lymph node | High | Close to LoD | n/a | 72/72 (94.9% - 100%) |
| V600E | Skin | Med | Med- high | - | 72/72 (94.9% - 100%) |
| V600E | Lymph node | Med | Med- high | - | 72/72 (94.9% - 100%) |
| V600K | Lymph node – high melanin | Med | - | Diluted | 72/72 (94.9% - 100%) |
| V600K | Skin | Low | Close to LoD | - | 72/72 (94.9% - 100%) |
| V600K | Lymph node | High | Close to LoD | - | 72/72 (94.9% - 100%) |
| V600K | Skin | Med | Med-high | - | 72/72 (94.9% - 100%) |
| V600K | Lymph node | Med | Med-high | - | 72/72 (94.9% - 100%) |

An estimate of the within-run precision, between-run (operators), between-days, between lots, between sites/instruments, and the total precision using restricted maximum likelihood method (REML) was conducted. The standard deviation and $\% \mathrm{CV}$ for Ct and $\Delta \mathrm{Ct}$ results for V600E samples, V600K samples, internal control (IC) Ct values for the WT samples, and positive and negative controls, were investigated as a measure of the variability of the assay. For the Wild-type panel samples, the internal control Ct for the V600E multiplexes ranged from 24.4 to 28.7 9 with $\% \mathrm{CV}$ range $0 - 3.3\%$ and the V600K multiplexes IC ranged from 24.2 to 28.5 with $\% \mathrm{CV}$ range $0 - 2.8\%$ . The mean $\Delta \mathrm{Ct}$ ranged from 1.9 to 6.1 for the V600E mutation positive samples with associated $\% \mathrm{CV}$ values ranging from 0 to $16.7\%$ . The mean $\Delta \mathrm{Ct}$ ranged from 1.2 to 4.9 for the V600K mutation positive samples with associated $\% \mathrm{CV}$ values ranging from 0 to $25.6\%$ . The higher imprecision was associated with the high melanin content sample. For the V600E positive control the mean $\Delta \mathrm{Ct}$ was 3.7 and the $\% \mathrm{CV}$ values ranged from 0 to $12.8\%$ . The V600K positive control mean $\Delta \mathrm{Ct}$ value was 3.5 with associated $\% \mathrm{CV}$ ranging from 0 to $15.1\%$ .

# 12. Precision - Repeatability

The goal of this study was to determine repeatability (within laboratory) precision of the entire $\mathrm{THxID^{TM}}$ BRAF assay (extraction and amplification) by evaluating the repeatability between runs, operators, instruments, days and lots at one testing site. The precision panel was comprised of 8 panel members prepared from FFPE melanoma specimens of skin and lymph node representing the different genotypes (V600E, V600K, and wild-type), DNA input (low, moderate, high), and melanin content (range 0 to $100\%$ ) and mutant DNA content (based on $\Delta C_t$ values). Macrodissection was not necessary. Supplemental testing of an additional V600K samples was conducted for one run per day due to the one failed specimen results bringing total to 9. The panel members were evaluated in house with 2 $\mathrm{THxID^{TM}}$ BRAF Kit lots, 4 days per lot, 2 runs per day, 2 instruments (2 days per instrument), 2 operators (each performing 1 run per day) and 2 replicates for each condition in each run. Results were assessed qualitatively for percentage of correct calls across the 32 sections per specimen, and an estimation of variance components for the analysis of quantitative variables was performed using the Restricted Maximum

PMA P120014: FDA Summary of Safety and Effectiveness Data

{20}

Likelihood (REML) method. The results demonstrated 100% agreement after retesting with the exception of the V600K specimen with high DNA input and mutant DNA content intended to be close to the LoD. However, the ΔCt value was at the cut-off and the results reflect the fluctuation in mutation content that can occur between sections of the same FFPE specimen. Because the V600K panel member at high DNA content close the claimed LoD of 5% was correct only half the time, an additional V600K panel member was obtained and tested at the same high DNA content (317 ng/μL) at the claimed LoD. Precision estimates for all Ct values and ΔCt values were similar. The results demonstrated 100% agreement after retesting.

Correct Call Rate for Precision Testing Panel

| Specimen Type | Genotype | Mean DNA input ng/uL /range | Mean ΔCt (Mutant content) | Number of miscalls/ invalids | Correct results | % correct (95% CI) |
| --- | --- | --- | --- | --- | --- | --- |
| Skin* | Wild-type | 12.0 (low) (3.8 – 26.4) | (not applicable) | 0/0 | 32/32 | 100 (89.3 – 100) |
| Lymph-node | Wild-type | 174.5 (mid-range) (109 – 303.9) | (not applicable) | 0/0 | 32/32 | 100 (89.3 – 100) |
| Skin | V600E | 42.5 (low) 12.7 – 65.8 | 3.17 (low) | 0/0 | 32/32 | 100 (89.3 – 100) |
| Lymph-node | V600E | 578.7 (high) (328.4- 831.6) | 5.0 (low) | 0/0 | 32/32 | 100 (89.3 – 100) |
| Skin (high melanin) | V600E | 140.4 (mid-range) 97.1- 252.78 | 0.84 (high) | 0/0 | 32/32 (after dilution 1:4) | 100 (89.3 – 100) |
| Lymph-node | V600K | 44.2 (low) (24.8 – 99.12) | 1.74 (high) | 0/1 | 32/32 | 100 (89.3 – 100) |
| Lymph-node | V600K | 84.8 (mid-range) (51.7- 135.5) | 2.55 (mid-range) | 0/0 | 32/32 | 100 (89.3 – 100) |
| Lymph-node | V600K | 302.83 (high) (44.83 to 491.30) | 7.02 (very low) | 15 (false negatives) /0 | 17/32 | 53.1 (36.4 –69.1) |
| Lymph-node | V600K | 317.4 (high) (173- 427) | 4.56 (low) | 0/0 | 16/16 | 100 (89.3 – 100) |

## 13. Sample Handling Variability

The purpose of the FFPE melanoma tissues was to evaluate the impact of the extraction method across multiple users on the results. For the FFPE specimens, curls cut directly from blocks were placed in microtubes and alternating sections of each specimen were forwarded to 3 labs so that each lab received 3 sections per specimen. Each of the 9 tumor specimens was then prepared according to the IFU. Two operators were used to prepare the extractions from each specimen. The samples evaluated were as follows:

## 14. Lot-to-Lot Reproducibility

PMA P120014: FDA Summary of Safety and Effectiveness Data

{21}

An estimate of between-lot reproducibility was obtained from the three site reproducibility using the restricted maximum likelihood method (REML). Each site utilized kits from two different reagent lots as follows: Site one used lots 4 and 5, Site 2 used lots 5 and 6, and Site 3 used lots 4 and 6. For each specimen the SD and %CV were summarized for each mean Ct value obtained for the internal control for both V600E and V600K reactions, the ΔCt values and the positive controls. The total imprecision (as a function of %CV) between lots was less than 3.10% for internal control Ct values for each reaction, less than 8.7% and 13.4% for the ΔCt values for the positive controls and specimens, respectively.

## 15. Guard Band Studies

The objective of the guard band studies was to establish the robustness of the PCR conditions for the THxID™ BRAF assay. The following parameters were assessed in three different studies: (1) varying thermal cycling profile (alterations to cycling times and temperatures), and (2) varying volumes (±25%) of the Reagent Mix (Primers and probes V600K and V600E solutions and Master Mix) and (3) varying Proteinase K digestion times. An analysis of variance was carried out to evaluate the impact of the studied parameters, as well as their interaction, on the Ct and Delta Ct values. For the first two conditions, 3 V600E replicates at 15% mutant content and 3 WT replicates (V600E PCR), and 3 V600K replicates at 15% mutant content and 3 WT replicates (V600K PCR) (generated using genomic DNA extracted from melanoma cell lines) were tested at both 20 and 700 ng DNA input per reaction, three replicates per run. For the evaluation of proteinase K digestion, a lymph node V600E specimen, a skin V600K specimen and a lymph node wild-type specimen were assessed at three incubation times (30 minutes, one hour and 2 hours). The results were satisfactory to within 0.7°C, ±1 second, ±14.5% master mix volume, ±18.5% Reagent sphere volume.

## 16. Cross-Contamination

A study to evaluate the incidence of false positive results related to cross-contamination of specimens during the entire work flow of the THxID™ BRAF assay was conducted by alternating BRAF mutation negative (WT) specimens and BRAF mutation positive (V600E and V600K) having high mutation content. Specimens were selected to increase the risk of carryover, i.e., low DNA content for WT specimens (35 ng input), and high mutant DNA input (606 ng input) and concentration for mutation-positive specimens (one V600E and one V600K). WT FFPE clinical specimens were processed in alternating order with V600E and V600K FFPE clinical specimens. A total of five PCR runs were performed generating 115 WT sample amplification results for evaluation of carryover with each V600E and V600K PCR reaction. No unexpected amplification or false positive results were observed for wild-type samples during the study, demonstrating that carry-over contamination does not occur under tested conditions.

## 17. Comparison of Amplification if Plate vs. Strip Formats

The ABI 7500Fast dx Real-time PCR Instrument allows a user to perform the test with either plates or strips. The goal of this study was to confirm that results of the

PMA P120014: FDA Summary of Safety and Effectiveness Data
Page 22

{22}

THxID™ BRAF assay generated using plates for PCR amplification are comparable to results obtained when using strips for PCR amplification. Mutant V600E or V600K DNA extracted from cell lines were diluted with wild-type (WT) DNA to obtain a 3% mutant to WT ratio and a total DNA concentration of 10 ng/μl for WT, V600E and V600K samples. There were a total of 12 runs; 6 with plates and 6 with strips. Forty-four (44) replicates of WT DNA were tested for both V600E reaction and V600K reaction; 92 replicates of mutant DNA were tested for the corresponding PCR mutation reaction (i.e. V600E or V600K). The study was performed with one lot of assay reagent, one instrument and one operator. Each DNA solution was tested in both strips and plates. Mean values of Ct and ΔCt values were obtained for each sample type. Results showed an absolute difference of less than 0.6 Ct and less than 0.3 ΔCt between results obtained for the plate and strips. The study confirms that results generated using plates for PCR amplification are comparable to those obtained with strips.

## 18. Stability – Specimen

a) Clinical Specimen (blocks, slide-mounted, and sections) The stability of FFPE specimens (skin and lymph) used to perform the THxID™ BRAF assay was evaluated for FFPE blocks, slides and sections under a variety of temperatures and time points up to 24 months. The studies assessed specimen stability using released lots of the ThxID™ BRAF products, and were designed to evaluate regular storage conditions, transport storage conditions, prolonged freezing, and freeze-thaw (3 cycles). Melanin content of specimens ranged from 0 to 100% and represented WT, V600E and V600K mutations. Data was available for three months. The acceptance criteria were that all results passed and there was no degradation in Ct or ΔCt values when compared to the baseline time point. The data demonstrated that:

- The clinical skin and lymph node FFPE blocks stored at 18-25°C are stable 3 months.
- The clinical FFPE 10μm thick sections in tube stored at 18-25°C or at 2-8°C or at -31°C/ -19°C are stable 3 months.
- The clinical FFPE 10μm thick sections on slides (after mimicking a shipment condition at 18-25°C and 3 cycles of freeze-thaw) are stable 3 months at 18-25°C
- The clinical FFPE 10μm thick section on slide stored at 18-25°C or at 2-8°C or at -31°C/ -19°C are stable 3 months.

b) Extracted Clinical Specimen DNA The stability of DNA extract eluate generated from twelve different FFPE skin (2 WT, 2 V600E, and 2 V600K) and lymph node (2 WT, 2 V600E, and 2 V600K) specimens was evaluated. Storage and freeze thaw conditions were evaluated to mimic potential sample conditions as follows: storage at room temperature (18-25°C) for 2 hours, storage at 2-8°C for 24 and 48 hours, storage at -31°C to -19°C including 4 cycles of freezing/ thawing, for 1, 2, 6 and 7 months, and storage at ≤ -60°C including 4 cycles of freezing/ thawing, for 3, 6 and 7 months. A

PMA P120014: FDA Summary of Safety and Effectiveness Data
Page 23

{23}

trend analysis of the Ct Internal Control (IC) and $\Delta \mathrm{Ct}$ values were compared over time. The data met the predefined acceptance criteria in that any difference from baseline was always below 1 Ct of $1\Delta \mathrm{Ct}$ values, and no change in the qualitative result occurred. The data supports stability to 7 months at $-31^{\circ}\mathrm{C}$ to $-19^{\circ}\mathrm{C}$

## 19. Stability- Reagents

**Real-Time stability and shipping:** The shelf life of the THxID™ BRAF Kit was determined by real-time stability studies performed on 3 lots of THxID™ BRAF (upright and inverted) over a period of 24 months under two conditions (1) storage at $2 - 8^{\circ}\mathrm{C}$ to mimic the storage in the warehouse, and during shipment; and (2) customer storage conductions in which 2 lots have undergone thermal shocks to mimic transport conditions. The customer storage conditions for THxID™ BRAF PUR reagents (except THxID™ BRAF columns) are stored at $18 - 25^{\circ}\mathrm{C}$, and THxID-BRAF column from the THxID™ BRAF PUR and THxID™ BRAF AMP reagents stored at $2 - 8^{\circ}\mathrm{C}$. The procedure to perform stability studies consists of 4 positive controls (PC), 4 negative controls (NC), 4 WT samples (2 extractions x 2 replicates of amplification), 8 V600E samples and 8 V600K samples (4 extractions x 2 replicates). To date the stability testing supports expiration dating of 6 months when stored at 2-8°C, and at 6 months for the THxID BRAF PUR kit at $18 - 25^{\circ}\mathrm{C}$ except for the kit columns which are stored at $2 - 8^{\circ}\mathrm{C}$.

## 20. Open Vial

**Multiple-Use Stability of Reagent Solutions:** A multiple-use stability study was evaluated at 3 different time points: Reagents, Master Mix, and reconstituted extraction buffers were stored 1 month at -31°C to -19°C and included 2 freeze/thawing cycles. Result obtained on solution prepared with reagents coming from 2 lots stored 2 months at $2 - 8^{\circ}\mathrm{C}$ allow support the following multiple-use stability claims in the package insert:

i. V600E Primers solution (PRM dil + PRM V600E) can be use twice in 1 month when stored at $-31^{\circ}\mathrm{C}$ to $-19^{\circ}\mathrm{C}$,
ii. V600K Primers solution (PRM dil + PRM V600K) can be use twice in 1 month when stored at $-31^{\circ}\mathrm{C}$ to $-19^{\circ}\mathrm{C}$,
iii. Positive Control Solution (CONT+dil + CONT+) can be use twice in 1 month when stored at $-31^{\circ}\mathrm{C}$ to $-19^{\circ}\mathrm{C}$,
iv. Master Mix can be use twice in 1 month when stored at $2 - 8^{\circ}\mathrm{C}$.

**In-Use Stability of Reagent Solutions:** An in-use stability study was conducted in order to evaluate the stability of reagents, master mix and reconstituted extraction buffers after 1 hour storage at room temperature ($18 - 25^{\circ}\mathrm{C}$):

The stability testing was done at 2 time points on two lots stored under customer conditions ($2 - 8^{\circ}\mathrm{C}$). The procedure to perform stability point is based on the QC release test. Results obtained on solution prepared with reagents coming from 2 lots stored 2 months at $2 - 8^{\circ}\mathrm{C}$ allow the following in-use stability claim in the package insert:

i. V600E Primers solution (PRM dil + PRM V600E) can be stored 30 minutes on the bench at $18 - 25^{\circ}\mathrm{C}$,

PMA P120014: FDA Summary of Safety and Effectiveness Data
Page 24

{24}

ii. V600K Primers solution (PRM dil + PRM V600K) can be stored 30 minutes on the bench at 18-25°C,
iii. Reagent Mix solution (Master Mix + V600E or V600K solution) can be stored 30 minutes on the bench at 18-25°C. Remaining reagent Mix solution cannot be re-used.
iv. Positive Control Solution (CONT+dil + CONT+) can be stored 30 minutes on the bench at 18-25°C
v. Master Mix can be stored 30 minutes on the bench at 18-25°C.

## B. Animal Studies

None

## C. Additional Studies

None

## X. SUMMARY OF PRIMARY CLINICAL STUDIES

The diagnostic manufacturer bioMérieux (BMX) performed retrospective studies to establish a reasonable assurance of safety and effectiveness of the THxID™ BRAF kit to select patients who may benefit from treatment with dabrafenib (Tafinlar®) on the basis of a V600E positive result with their melanoma tissue; and to select patients who may benefit from treatment with tramatenib (Mekinist™) on the basis of a V600E or V600K positive test result with their melanoma specimen. Data from these studies were the basis for the PMA approval decision.

GlaxoSmithKline (GSK) is the manufacturer of dabrafenib (Tafinlar®) which is a selective inhibitor of BRAF kinase activity in melanoma patients whose tumors encode the BRAF T1799A (V600E) mutation; and tramatenib (Mekinist™) which is a kinase that inhibits the mitogen activated protein kinase (MEK) kinase activity in melanoma patients whose tumors encode either the V600E mutation or the BRAF mutation GT1798-1799AA (V600K)¹.

B-RAF is a serine/threonine kinase that belongs to a phosphorylation cascade called the MAPK pathway (Mitogen Activated Protein Kinase). This pathway is activated by extracellular signals such as growth factors, and modulates key cellular processes such as cell division. In this cascade, RAS (a membrane-anchored GTPase) phosphorylates and activates B-RAF, which then phosphorylates and activates MEK1/2 kinase. Approximately 40-60% of melanomas have BRAF mutations, and 80-90% of these mutations consist of the T to A substitution at base 1799 in exon 15 of the BRAF oncogene resulting in a glutamate amino acid at codon 600 rather than a valine amino acid (i.e., V600E mutations). An additional 10-15% of these mutations result in a lysine substitution (i.e., V600K mutations) due to GT to AA base substitutions at bases 1798 and 1799. The THxID™ BRAF test is designed to detect the V600E and V600K mutations which represent approximately 97% of the total mutations at this codon.

¹ Refer to drug label available on the web at DRUGS@fda for specific drug indications.

PMA P120014: FDA Summary of Safety and Effectiveness Data
Page 25

{25}

A summary of the studies conducted to support a dabrafenib (Tafinlar®) selection claim is presented in Section (1) and a summary of the studies conducted to support a tramatenib (Mekinist™) selection claim are presented in Section (2) below.

## (1) Summary of Primary Clinical Studies - Dabrafenib (TAFINLAR®)

A summary of the clinical studies for Dabrafenib are presented below.

### A. Study Design - Dabrafenib

A phase 3 [BREAK-3 (BRF11683); NCT01227889] clinical trial was conducted to determine the safety and efficacy of a new investigational drug, dabrafenib (Tafinlar®). BREAK-3 was an international two-arm, open-label, randomized (3:1) Phase 3 study comparing the efficacy and safety of dabrafenib to dacarbazine in patients with advanced (Stage III) or metastatic (Stage IV) melanoma whose tumor tissue harbors a V600E mutation. The BREAK-3 study was sponsored by GlaxoSmithKline (GSK) and the data was submitted to the FDA in New Drug Application (NDA 202806). Patients were screened for enrollment into this trial using an investigational clinical trial assay (CTA) designed to detect the V600E and V600K mutations. Enrollment commenced in January 2011 completed September 2011. A total of 734 patients were screened for eligibility, and 250 were enrolled into either the dabrafenib treatment arm (n=187) or to dacarbazine (n=63). Randomization was stratified for disease stage at baseline. Treatment continued until disease progression, death or withdrawal. The main efficacy measure was progression free survival (PFS) as assessed by the investigator. Specimens from patients were banked and retested in retrospective studies designed to establish the analytical and clinical concordance between the THxID -BRAF assay and the CTA (i.e., bridging studies) to clinically validate the test as safe and effective for selecting patients who may benefit with dabrafenib (Tafinlar®).

### 1. Inclusion and Exclusion Criteria for Specimen Testing - Dabrafenib

Formalin fixed, paraffin embedded (FFPE), tissue blocks from a biopsy in the metastatic setting were required for the assessment of BRAF mutation status to determine trial eligibility. Each block contained 5-10 mm² of tumor tissue submitted in standard 4×3 cm cassettes. If sites were unable to send a tissue block, tissue slides were submitted from a single 5 micron thick section. If archived biopsy tissue was not available, an FFPE core biopsy from a metastatic site was required. One section was stained with hematoxylin and eosin (H&amp;E stain) to determine tumor presence. Tumors were required to be macrodissected if tumor content was below 80% of the section. For retesting with the THxID™ BRAF Kit, tissue sections from the same block were used first. In the rare absence of the availability of tissue sections, archived DNA eluate from the original extraction was used. All specimens from patients whose eligibility for the trial was based on their tumor mutations status were retested (i.e., includes mutation positives, mutation negatives, and invalids). Reasons for missing samples were accounted. For all studies, repeat testing was performed according to the trouble shooting section of the

PMA P120014: FDA Summary of Safety and Effectiveness Data
Page 26

{26}

test labeling [i.e., Instructions for Use (IFU)]. No more than two repeat tests were performed for a sample when attempting to obtain a valid result.

2. Clinical Inclusion and Exclusion Criteria for Patient Enrollment-Dabrafenib Inclusion Criteria - Subjects eligible for enrollment must meet all the following criteria

a. Has provided signed informed consent.
b. Histologically confirmed advanced (unresectable Stage III) or metastatic melanoma (Stage IV) and BRAF mutation-positive (V600 E) melanoma as determined via central testing with a BRAF mutation assay.
c. Are treatment naïve for advanced (unresectable) or metastatic melanoma, with the exception of IL-2.
d. Measurable disease according to Response Evaluation Criteria in Solid Tumors (RECIST 1.1).
e. Age ≥18 years of age
f. Able to swallow and retain oral medication.
g. Women with child-bearing potential and men with reproductive potential must be willing to practice acceptable methods of birth control during the study. Additionally, women of childbearing potential must have a negative serum pregnancy test within 14 days prior to the first dose of study treatment.
h. Eastern Cooperative Oncology Group (ECOG) Performance Status of 0-1[Oken, 1982].
i. Must have adequate organ function as defined by a set of screening values described in the clinical protocol.

Exclusion Criteria – Subjects meeting any of the following criteria must not be enrolled in the study

a. Previous treatment for metastatic melanoma, including treatment with a BRAF or MEK inhibitor.
b. Known ocular or primary mucosal melanoma.
c. Currently receiving cancer therapy (chemotherapy, radiation therapy, immunotherapy, biologic therapy, or surgery).
d. Use of any investigational anti-cancer or other drug within 28 days or 5 half-lives, whichever is longer, preceding the first dose of dabrafenib.
e. Current use of a prohibited medication or is expected to require any of these medications during treatment with dabrafenib.
f. Any major surgery, radiotherapy, or immunotherapy within the last 4 weeks.
g. Presence of active gastrointestinal disease or other condition that will interfere significantly with the absorption of drugs. If clarification is needed as to whether a condition will significantly affect absorption of drugs, contact the GSK medical monitor for permission to enroll the subject.
h. A history of Human Immunodeficiency Virus (HIV) infection
i. A history of glucose-6-phosphate dehydrogenase (G6PD) deficiency.

PMA P120014: FDA Summary of Safety and Effectiveness Data
Page 27

{27}

j. A history of other malignancy. Subjects who have been disease-free for 5 years, or subjects with a history of completely resected non-melanoma skin cancer or successfully treated in situ carcinoma are eligible.

k. Evidence of active CNS disease (radiographically unstable, symptomatic lesions). However prior treatment with stereotactic radiosurgery (SRS) or surgical resection is allowed if the subject remains without evidence of disease progression in the brain ≥3 months, and has been off corticosteroids for ≥3 weeks. Whole brain radiotherapy is not allowed except in those subjects who have had definitive resection or SRS of all radiographically detectable parenchymal lesions.

l. History of alcohol or drug abuse within 6 months prior to Screening.

m. Psychological, familial, sociological, or geographical conditions that do not permit compliance with the protocol, or unwillingness or inability to follow the procedures required in the protocol.

n. Cardiac abnormalities as defined in the clinical protocol.

## 3. Follow-up Schedule – Dabrafenib

### Patients were followed for efficacy:

Patients in each treatment arm underwent scheduled clinical and tumor assessments at baseline and then every 3 weeks thereafter until confirmation of disease progression. Patients who were removed from treatment were assessed thereafter every 12 weeks until death or study completion. All subjects who permanently discontinue study treatment were followed for survival and additional anti-cancer therapies [including radiotherapy] every 12 weeks until death or study completion. In addition, those subjects who permanently discontinued study treatment without progressive disease were to have radiographic disease assessments performed on the same assessment schedule noted protocol every 9 weeks until Week 27 and then every 12 weeks thereafter until disease progression, start of new anti-cancer therapy, or death.

### Patients were followed for safety:

Information related to the adverse events experienced by patients treated with dabrafenib in the BREAK-3 trial were collected until disease progression or unacceptable toxicity for at least 6 months. Adverse events (AEs) were collected from the time the first dose of study treatment was administered until 28 days after discontinuation of study treatment. Serious adverse events (SAEs) were collected over the same time period. Abnormal laboratory and safety assessments reported as abnormal were outlined and evaluated according to schedule. After discontinuation of treatment, the investigator monitored all AEs and SAEs that were ongoing until resolution or stabilization of the event or until the subject was lost to follow-up. The investigator and site staff were responsible for detecting, documenting and reporting events that meet the definition of an adverse event (AE) or serious adverse event (SAE) as described in the study protocol.

PMA P120014: FDA Summary of Safety and Effectiveness Data
Page 28

{28}…