← Product Code [PQP](/productcode/PQP) · P170019S016

# FoundationOne CDx (F1CDx) (P170019S016)

_Foundation Medicine, Inc. · PQP · Jun 16, 2020 · Pathology · APPR_

**Canonical URL:** https://fda-staging.innolitics.com/device/P170019S016

## Device Facts

- **Applicant:** Foundation Medicine, Inc.
- **Product Code:** [PQP](/productcode/PQP.md)
- **Decision Date:** Jun 16, 2020
- **Decision:** APPR
- **Device Class:** Class 3
- **Review Panel:** Pathology
- **Attributes:** Real-World Evidence

## Real-World Evidence

| Submission | Device | Sponsor | RWD Sources | RWE Use Summary | Key Tags |
| --- | --- | --- | --- | --- | --- |
| P170019S016 · Jun 16, 2020 | FoundationOne CDx (F1CDx) | Foundation Medicine, Inc. | KEYNOTE-158 clinical trial specimens (archived FFPE tumor tissue) | Retrospective analysis of archived clinical trial samples to validate the clinical performance of the F1CDx assay for identifying TMB-H patients who may benefit from pembrolizumab. | Retrospective analysis; Clinical trial specimens; TMB-H; Companion diagnostic |

### Clinical Evidence

| Study Design | Population | Comparator | Key Endpoints |
| --- | --- | --- | --- |
| KEYNOTE-158 Retrospective Analysis; Prospectively-planned retrospective analysis; Study Period: Data cut-off June 27, 2019 | Patients with advanced (unresectable or metastatic) solid cancers (Cohorts A-J) who progressed following prior treatment; Sample Size: 790 (TE population) / 719 (DV population); Number of Sites: 81 sites globally | Non-TMB-H (< 10 mut/Mb) population | Objective Response Rate (ORR), Duration of Response (DOR) |

## Indications for Use

FoundationOne®CDx (F1CDx) is a qualitative next generation sequencing based in vitro diagnostic test that uses targeted high throughput hybridization-based capture technology for detection of substitutions, insertion and deletion alterations (indels) and copy number alterations (CNAs) in 324 genes and select gene rearrangements, as well as genomic signatures including microsatellite instability (MSI) and tumor mutational burden (TMB) using DNA isolated from formalin-fixed paraffin embedded (FFPE) tumor tissue specimens. The test is intended as a companion diagnostic to identify patients who may benefit from treatment with the targeted therapies listed in Table 1 in accordance with the approved therapeutic product labeling. Additionally, F1CDx is intended to provide tumor mutation profiling to be used by qualified health care professionals in accordance with professional guidelines in oncology for cancer patients with solid malignant neoplasms. Genomic findings other than those listed in Table 1 are not prescriptive or conclusive for labeled use of any specific therapeutic product.

## Device Story

F1CDx is a NGS-based IVD that analyzes DNA from FFPE tumor tissue to detect genomic alterations and signatures. Input: 50-1000 ng DNA extracted from FFPE samples. Process: Whole-genome shotgun library construction; hybridization-based capture of 324 cancer-related genes; sequencing on Illumina HiSeq 4000. Analysis: Proprietary bioinformatics pipeline detects base substitutions, indels, CNAs, rearrangements, MSI, and TMB. Output: Clinical report identifying specific genomic alterations and signatures (e.g., TMB ≥ 10 mut/Mb) to guide therapy selection. Used in centralized laboratories (Cambridge, MA; Morrisville, NC) by trained personnel. Healthcare providers use the report to identify patients eligible for specific FDA-approved targeted therapies, potentially improving clinical outcomes by matching patients to effective treatments.

## Clinical Evidence

Clinical validation used a prospectively-planned retrospective analysis of 1,050 patients from the KEYNOTE-158 trial. Efficacy population (TE) included 790 patients with valid TMB results. Primary endpoint: Objective Response Rate (ORR) per RECIST 1.1. Results: TMB-H (≥ 10 mut/Mb) patients (n=102) had an ORR of 29.4% (95% CI: 20.8%, 39.3%) vs. 6.3% (95% CI: 4.6%, 8.3%) in non-TMB-H patients (n=688). Responses were durable; 66.6% of TMB-H responders had DOR ≥ 24 months.

## Technological Characteristics

NGS-based targeted hybridization capture assay. Materials: FFPE tumor DNA. Energy: Illumina HiSeq 4000 sequencing. Dimensions: 324 genes targeted. Connectivity: Centralized laboratory processing. Software: Proprietary bioinformatics pipeline (v3.3.x). Sterilization: N/A (in vitro diagnostic).

## Regulatory 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

- FoundationFocus CDxBRCA ([P160018](/device/P160018.md))
- PathVysion HER-2 DNA Probe Kit
- PATHWAY Anti-HER-2/neu (4B5) Rabbit Monoclonal Primary Antibody
- InSite HER-2/neu Kit
- SPOT-Light HER2 CISH Kit
- Bond Oracle HER2 IHC System
- HER2 CISH pharmDx Kit
- HercepTest
- HER2 FISH pharmDx Kit
- THxID BRAF Kit
- cobas 4800 BRAF V600 Mutation Test
- Oncomine Dx Target Test
- therascreen BRAF V600E RGQ PCR Kit
- Praxis Extended RAS Panel
- cobas KRAS Mutation Test
- therascreen KRAS RGQ PCR Kit
- Vysis ALK Break Apart FISH Probe Kit
- ALK (D5F3) CDx Assay
- cobas EGFR Mutation Test v2
- therascreen EGFR RGQ PCR Kit
- FoundationFocus CDxBRCA
- therascreen PIK3CA RGQ PCR Kit

## Submission Summary (Full Text)

> This content was OCRed from public FDA records by [Innolitics](https://innolitics.com). If you use, quote, summarize, crawl, or train on this content, cite Innolitics at https://innolitics.com.
>
> Innolitics is a medical-device software consultancy. We help companies design, build, and clear FDA-regulated software and AI/ML devices, including [a PMA](https://innolitics.com/services/regulatory/), [a 510(k)](https://innolitics.com/services/510ks/), [a SaMD](https://innolitics.com/services/end-to-end-samd/), [an AI/ML medical device](https://innolitics.com/services/medical-imaging-ai-development/), or [an FDA regulatory strategy](https://innolitics.com/services/regulatory/).

{0}

# SUMMARY OF SAFETY AND EFFECTIVENESS DATA (SSED)

## I. GENERAL INFORMATION

|  Device Generic Name: | Next generation sequencing oncology panel, somatic or germline variant detection system  |
| --- | --- |
|  Device Trade Name: | FoundationOne®CDx (F1CDx)  |
|  Device Procode: | PQP  |
|  Applicant's Name and Address: | Foundation Medicine, Inc. 150 Second Street Cambridge, MA 02141  |
|  Date(s) of Panel Recommendation: | None  |
|  Premarket Approval Application (PMA) Number: | P170019/S016  |
|  Date of FDA Notice of Approval: | June 16, 2020  |

The original PMA (P170019) for FoundationOne®CDx (F1CDx) was approved on November 30, 2017 for the detection of genetic alterations in patients who may benefit from one of fifteen FDA-approved therapies for non-small cell lung cancer (NSCLC), melanoma, breast cancer, colorectal cancer, and ovarian cancer. Subsequently, eight PMA supplements were approved for expanding the indications for use of F1CDx since its original approval. PMA supplement (P170019/S005) for adding genomic loss of heterozygosity (LOH) was approved on April 10, 2019. PMA supplement (P170019/S004) for adding an indication for LYNPARZA® (olaparib) in ovarian cancer patients with *BRCA1/2* alterations was approved on July 1, 2019. PMA supplement (P170019/S008) for adding an indication for TAGRISSO® (osimertinib) in NSCLC patients with *EGFR* exon 19 deletions and *EGFR* exon 21 L858R alterations was approved on July 1, 2019. PMA supplement (P170019/S006) for adding an indication for PIQRAY® (alpelisib) in breast cancer patients with *PIK3CA* alterations was approved on December 3, 2019. PMA supplement (P170019/S010) for adding a second site in Research Triangle Park, NC, where the F1CDx assay will be performed, was approved on December 16, 2019. PMA supplement (P170019/S013) for adding an indication for PEMZYRE® (pemigatinib) in cholangiocarcinoma patients with *FGFR2* fusions was approved on April 17, 2020. PMA supplement (P170019/S011) for adding an indication for TABRECTA® (capmatinib) in NSCLC patients with *MET* single nucleotide variants (SNVs) and indels that lead to *MET* exon 14 skipping was approved on May 6, 2020. PMA supplement (P170019/S015) for adding an indication for LYNPARZA® (olaparib) in metastatic castration resistant prostate cancer (mCRPC) patients with mutations in homologous recombination repair (HRR) genes was approved on May 19, 2020.

---PMA P170019/S016: FDA Summary of Safety and Effectiveness Data

1 of 40

{1}

The current supplement was submitted to expand the intended use of F1CDx to include a companion diagnostic indication for high tumor mutational burden (TMB) at the cut-off of 10 mutations per megabase (mut/Mb) in patients with solid tumors who may benefit from treatment with KEYTRUDA® (pembrolizumab).

## II. INDICATIONS FOR USE

FoundationOne®CDx (F1CDx) is a qualitative next generation sequencing based in vitro diagnostic test that uses targeted high throughput hybridization-based capture technology for detection of substitutions, insertion and deletion alterations (indels) and copy number alterations (CNAs) in 324 genes and select gene rearrangements, as well as genomic signatures including microsatellite instability (MSI) and tumor mutational burden (TMB) using DNA isolated from formalin-fixed paraffin embedded (FFPE) tumor tissue specimens. The test is intended as a companion diagnostic to identify patients who may benefit from treatment with the targeted therapies listed in Table 1 in accordance with the approved therapeutic product labeling. Additionally, F1CDx is intended to provide tumor mutation profiling to be used by qualified health care professionals in accordance with professional guidelines in oncology for cancer patients with solid malignant neoplasms. Genomic findings other than those listed in Table 1 are not prescriptive or conclusive for labeled use of any specific therapeutic product.

Table 1. Companion diagnostic indications

|  Indication | Biomarker | Therapy  |
| --- | --- | --- |
|  Non-small cell lung cancer (NSCLC) | EGFR exon 19 deletions and EGFR exon 21 L858R alterations | Gilotrif® (afatinib), Iressa® (gefitinib), Tagrisso® (osimertinib), or Tarceva® (erlotinib)  |
|   |  EGFR exon 20 T790M alterations | Tagrisso® (osimertinib)  |
|   |  ALK rearrangements | Alecensa® (alectinib), Xalkori® (crizotinib), or Zykadia® (ceritinib)  |
|   |  BRAF V600E | Tafinlar® (dabrafenib) in combination with Mekinist® (trametinib)  |
|   |  MET single nucleotide variants (SNVs) and indels that lead to MET exon 14 skipping | Tabrecta™ (capmatinib)  |
|  Melanoma | BRAF V600E | Tafinlar® (dabrafenib) or Zelboraf® (vemurafenib)  |
|   |  BRAF V600E and V600K | Mekinist® (trametinib) or Cotellic® (cobimetinib) in combination with Zelboraf® (vemurafenib)  |

PMA P170019/S016: FDA Summary of Safety and Effectiveness Data

2 of 40

{2}

|  Indication | Biomarker | Therapy  |
| --- | --- | --- |
|  Breast cancer | ERBB2 (HER2) amplification | Herceptin® (trastuzumab), Kadcyla® (ado-trastuzumab-emtansine), or Perjeta® (pertuzumab)  |
|   |  PIK3CA C420R, E542K, E545A, E545D [1635G>T only], E545G, E545K, Q546E, Q546R, H1047L, H1047R, and H1047Y alterations | Piqray® (alpelisib)  |
|  Colorectal cancer | KRAS wild-type (absence of mutations in codons 12 and 13) | Erbitux® (cetuximab)  |
|   |  KRAS wild-type (absence of mutations in exons 2, 3, and 4) and NRAS wild-type (absence of mutations in exons 2, 3, and 4) | Vectibix® (panitumumab)  |
|  Ovarian cancer | BRCA1/2 alterations | Lynparza® (olaparib) or Rubraca® (rucaparib)  |
|  Cholangiocarcinoma | FGFR2 fusions and select rearrangements | Pemazyre™ (pemigatinib)  |
|  Prostate cancer | Homologous Recombination Repair (HRR) gene (BRCA1, BRCA2, ATM, BARD1, BRIP1, CDK12, CHEK1, CHEK2, FANCL, PALB2, RAD51B, RAD51C, RAD51D and RAD54L) alterations | Lynparza® (olaparib)  |
|  Solid tumors | TMB ≥ 10 mutations per megabase | Keytruda® (pembrolizumab)  |

The test is also used for detection of genomic loss of heterozygosity (LOH) from formalin-fixed, paraffin-embedded (FFPE) ovarian tumor tissue. Positive homologous recombination deficiency (HRD) status (F1CDx HRD defined as tBRCA-positive and/or LOH high) in ovarian cancer patients is associated with improved progression-free survival (PFS) from Rubraca (rucaparib) maintenance therapy in accordance with the Rubraca product label.

The F1CDx assay will be performed at Foundation Medicine, Inc. sites located in Cambridge, MA and Morrisville, NC.

### III. CONTRAINDICATIONS

There are no known contraindications.

### IV. WARNINGS AND PRECAUTIONS

The warnings and precautions can be found in the FoundationOne®CDx assay labeling.

PMA P170019/S016: FDA Summary of Safety and Effectiveness Data

3 of 40

{3}

# V. DEVICE DESCRIPTION

FoundationOne®CDx (F1CDx) is performed at Foundation Medicine, Inc. sites located in Cambridge, MA and Morrisville, NC. The assay includes reagents, software, instruments and procedures for testing DNA extracted from formalin-fixed, paraffin-embedded (FFPE) tumor samples.

The assay employs a single DNA extraction method from routine FFPE biopsy or surgical resection specimens, 50-1000 ng of which undergoes whole-genome shotgun library construction and hybridization-based capture of all coding exons from 309 cancer-related genes, 1 promoter region, 1 non-coding RNA (ncRNA), and select intronic regions from 34 commonly rearranged genes, 21 of which also include the coding exons (refer to Table 2 and Table 3, below, for the complete list of genes included in F1CDx). In total, the assay therefore detects alterations in 324 genes. Using the Illumina® HiSeq 4000 platform, hybrid-capture selected libraries are sequenced to high uniform depth (targeting > 500X median coverage with > 99% of exons at coverage > 100X). Sequence data are processed using a customized analysis pipeline designed to detect all classes of genomic alterations, including base substitutions, indels, copy number alterations (amplifications and homozygous deletions), and selected genomic rearrangements (e.g., gene fusions). Additionally, genomic signatures including microsatellite instability (MSI), tumor mutational burden (TMB), and positive homologous recombination deficiency (HRD) status (tBRCA-positive and/or LOH high) will be reported.

Table 2. Genes with full coding exonic regions included in F1CDx for the detection of substitutions, insertions and deletions (indels), and copy number alterations (CNAs)

|  ABL1 | BRAF | CDKN1A | EPHA3 | FGFR4 | IKZF1 | MCL1 | NKX2-1 | PMS2 | RNF43 | TET2  |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
|  ACVR1B | BRCA1 | CDKN1B | EPHB1 | FH | INPP4B | MDM2 | NOTCH1 | POLD1 | ROS1 | TGFBR2  |
|  AKT1 | BRCA2 | CDKN2A | EPHB4 | FLCN | IRF2 | MDM4 | NOTCH2 | POLE | RPTOR | TIPARP  |
|  AKT2 | BRD4 | CDKN2B | ERBB2 | FLT1 | IRF4 | MED12 | NOTCH3 | PPARG | SDHA | TNFAIP3  |
|  AKT3 | BRIP1 | CDKN2C | ERBB3 | FLT3 | IRS2 | MEF2B | NPM1 | PPP2R1A | SDHB | TNFRSF14  |
|  ALK | BTG1 | CEBPA | ERBB4 | FOXL2 | JAK1 | MEN1 | NRAS | PPP2R2A | SDHC | TP53  |
|  ALOX12B | BTG2 | CHEK1 | ERCC4 | FUBP1 | JAK2 | MERTK | NT5C2 | PRDM1 | SDHD | TSC1  |
|  AMER1 | BTK | CHEK2 | ERG | GABRA6 | JAK3 | MET | NTRK1 | PRKAR1A | SETD2 | TSC2  |
|  APC | C11orf30 | CIC | ERRFI1 | GATA3 | JUN | MITF | NTRK2 | PRKCI | SF3B1 | TYRO3  |
|  AR | CALR | CREBBP | ESR1 | GATA4 | KDM5A | MKNK1 | NTRK3 | PTCH1 | SGK1 | U2AF1  |
|  ARAF | CARD11 | CRKL | EZH2 | GATA6 | KDM5C | MLH1 | P2RY8 | PTEN | SMAD2 | VEGFA  |
|  ARFRP1 | CASP8 | CSF1R | FAM46C | GID4 (C17orf39) | KDM6A | MPL | PALB2 | PTPN11 | SMAD4 | VHL  |
|  ARID1A | CBFB | CSF3R | FANCA | GNA11 | KDR | MRE11A | PARK2 | PTPRO | SMARC A4 | WHSC1  |
|  ASXL1 | CBL | CTCF | FANCC | GNA13 | KEAP1 | MSH2 | PARP1 | QKI | SMARC B1 | WHSC1L1  |
|  ATM | CCND1 | CTNNA1 | FANCG | GNAQ | KEL | MSH3 | PARP2 | RAC1 | SMO | WT1  |
|  ATR | CCND2 | CTNNB1 | FANCL | GNAS | KIT | MSH6 | PARP3 | RAD21 | SNCAIP | XPO1  |

PMA P170019/S016: FDA Summary of Safety and Effectiveness Data

4 of 40

{4}

|  ATRX | CCND3 | CUL3 | FAS | GRM3 | KLHL6 | MST1R | PAX5 | RAD51 | SOCS1 | XRCC2  |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
|  AURKA | CCNE1 | CUL4A | FBXW7 | GSK3B | KMT2A (MLL) | MTAP | PBRM1 | RAD51B | SOX2 | ZNF217  |
|  AURKB | CD22 | CXCR4 | FGF10 | H3F3A | KMT2D (MLL2) | MTOR | PDCD1 | RAD51C | SOX9 | ZNF703  |
|  AXIN1 | CD274 | CYP17A1 | FGF12 | HDAC1 | KRAS | MUTYH | PDCD1L G2 | RAD51D | SPEN |   |
|  AXL | CD70 | DAXX | FGF14 | HGF | LTK | MYC | PDGFRA | RAD52 | SPOP |   |
|  BAP1 | CD79A | DDR1 | FGF19 | HNF1A | LYN | MYCL | PDGFRB | RAD54L | SRC |   |
|  BARD1 | CD79B | DDR2 | FGF23 | HRAS | MAF | MYCN | PDK1 | RAF1 | STAG2 |   |
|  BCL2 | CDC73 | DIS3 | FGF3 | HSD3B1 | MAP2K1 | MYD88 | PIK3C2B | RARA | STAT3 |   |
|  BCL2L1 | CDH1 | DNMT3A | FGF4 | ID3 | MAP2K2 | NBN | PIK3C2G | RB1 | STK11 |   |
|  BCL2L2 | CDK12 | DOT1L | FGF6 | IDH1 | MAP2K4 | NF1 | PIK3CA | RBM10 | SUFU |   |
|  BCL6 | CDK4 | EED | FGFR1 | IDH2 | MAP3K1 | NF2 | PIK3CB | REL | SYK |   |
|  BCOR | CDK6 | EGFR | FGFR2 | IGF1R | MAP3K13 | NFE2L2 | PIK3R1 | RET | TBX3 |   |
|  BCORL1 | CDK8 | EP300 | FGFR3 | IKBKE | MAPK1 | NFKBIA | PIM1 | RICTOR | TEK |   |

Table 3. Genes with select intronic regions for the detection of gene rearrangements, a promoter region, and an ncRNA gene

|  ALK introns 18, 19 | BRCA1 introns 2, 7, 8, 12, 16, 19, 20 | ETV4 introns 5, 6 | EZR introns 9-11 | KIT intron 16 | MYC intron 1 | NUTM1 intron 1 | RET introns 7-11 | SLC34A2 intron 4  |
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
|  BCL2 3'UTR | BRCA2 intron 2 | ETV5 introns 6, 7 | FGFR1 intron 1, 5, 17 | KMT2A (MLL) introns 6-11 | NOTCH2 intron 26 | PDGFRA introns 7, 9, 11 | ROS1 introns 31-35 | TERC ncRNA  |
|  BCR introns 8, 13, 14 | CD74 introns 6-8 | ETV6 introns 5, 6 | FGFR2 intron 1, 17 | MSH2 intron 5 | NTRK1 introns 8-10 | RAF1 introns 4-8 | RSPO2 intron 1 | TERT Promoter  |
|  BRAF introns 7-10 | EGFR introns 7, 15, 24-27 | EWSR1 introns 7-13 | FGFR3 intron 17 | MYB intron 14 | NTRK2 Intron 12 | RARA intron 2 | SDC4 intron 2 | TMPRSS2 introns 1-3  |

### Test Output

The output of the test includes:

Category 1: CDx Claims noted in Table 1 of the Intended Use

Category 2: Cancer Mutations with Evidence of Clinical Significance

Category 3: Cancer Mutations with Potential Clinical Significance

PMA P170019/S016: FDA Summary of Safety and Effectiveness Data

5 of 40

{5}

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

### Test Kit Contents

The test includes a sample shipping kit, which is sent to ordering laboratories. The shipping kit contains the following components:

- Specimen Preparation Instructions
- Shipping Instructions
- Return Shipping Label

### Instruments

The F1CDx assay is intended to be performed with serial number-controlled instruments as indicated in Table 4, below. All instruments are qualified by Foundation Medicine, Inc. (FMI) under FMI's Quality System.

Table 4. Instruments for use with the F1CDx assay

|  Instrument  |
| --- |
|  Illumina® HiSeq 4000  |
|  Illumina cBot® System  |
|  Agilent Technologies Benchbot Workstation with Integrated Bravo Automated Liquid Handler  |
|  Beckman Biomek NX^{P} Span-8 Liquid Handler  |
|  Hamilton Microlab STAR/STARlet Liquid Handling Workstation  |
|  Thermo Fisher Scientific KingFisher™ Flex with 96 Deep-well Head  |
|  Covaris LE220-Plus Focused-ultrasonicator  |

### Test Process

All assay reagents included in the F1CDx assay process are qualified by FMI and are compliant with the medical device Quality System Regulation (QSR).

### A. Specimen Collection and Preparation

Formalin-fixed, paraffin-embedded (FFPE) tumor specimens are collected and prepared following standard pathology practices. FFPE specimens may be received either as unstained slides or as an FFPE block.

Prior to starting the assay, a Hematoxylin and Eosin (H&E) stained slide is prepared, and then reviewed by a board-certified pathologist to confirm disease ontology and to ensure that adequate tissue (≥ 0.6 mm³), tumor content (≥ 20% tumor), and sufficient nucleated cells are present to proceed with the assay.

### B. DNA Extraction

Specimens passing pathology review are queued for DNA extraction which begins with lysis of cells from FFPE tissue by digestion with a proteinase K buffer followed by automated purification using the 96-well KingFisher™ Flex Magnetic Particle Processor.

PMA P170019/S016: FDA Summary of Safety and Effectiveness Data

6 of 40

{6}

After completion of DNA extraction, double-stranded DNA (dsDNA) is quantified by the Quant-iT™ PicoGreen® fluorescence assay using the provided lambda DNA standards (Invitrogen) prior to Library Construction (LC). The sample must yield a minimum of 55 ng of genomic DNA to ensure sufficient DNA for quality control (QC) and to proceed with LC.

### C. Library Construction

Library Construction (LC) begins with normalization of DNA to 50-1000 ng. Normalized DNA samples are randomly sheared (fragmented) to ~200 bp by adaptive focused acoustic sonication using the Covaris LE220-Plus before purification with a 1.8X volume of AMPure® XP Beads (Agencourt®). Solid-phase reversible immobilization (SPRI) purification and subsequent library construction with the NEBNext® reagents (custom-filled kits by NEB), including mixes for end repair, dA addition and ligation, are performed in 96-well plates (Eppendorf) on the Bravo Benchbot (Agilent) or Microlab STAR (Hamilton) using the “with-bead” protocol¹ to maximize reproducibility and library yield. Indexed (6 bp barcodes) sequencing libraries are PCR amplified with HiFi™ (Kapa) for 10 cycles and subsequently 1.8X SPRI purified. Purification and dilution for QC are performed.

Following LC, a QC procedure is performed by quantifying single-stranded DNA (ssDNA) from purified libraries using the Quant-iT™ OliGreen® ssDNA Assay Kit (Life Technologies) read on a Molecular Devices Multimode SpectraMax M2 plate Reader. Libraries yielding insufficient sequencing library are failed.

### D. Hybrid Capture

Hybrid Capture (HC) begins with normalization of each library to 500-2000 ng. Normalized samples then undergo solution hybridization which is performed using a > 50-fold molar excess of a pool of individually synthesized 5'-biotinylated DNA 120 bp oligonucleotides. The baits target ~1.8 Mb of the human genome including all coding exons of 309 cancer-related genes, introns or non-coding regions of 35 genes, plus > 3,500 single nucleotide polymorphisms (SNPs) located throughout the genome. Baits are designed by tiling overlapping 120 bp DNA sequence intervals covering target exons (60 bp overlap) and introns (20 bp overlap), with a minimum of three baits per target; SNP targets are allocated one bait each. Intronic baits are filtered for repetitive elements² as defined by the UCSC Genome RepeatMasker track.

After hybridization, the library-bait duplexes are captured on paramagnetic MyOne™ streptavidin beads (Invitrogen), and off-target material is removed by washing one time with 1X SSC at 25°C and four times with 0.25X SSC at 55°C. The PCR master mix is added to directly amplify (12 cycles) the captured library from the washed beads.³ After 12 cycles of amplification, the samples are 1.8X SPRI purified. Purification and dilution for QC are performed.

QC for HC is performed by measuring dsDNA yield using the Quant-iT™ PicoGreen® dsDNA Assay Kit (Life Technologies) read on a Molecular Devices

PMA P170019/S016: FDA Summary of Safety and Effectiveness Data

7 of 40

{7}

Multimode SpectraMax M2 plate Reader. Captured libraries yielding less than 140 ng of sequencing library are failed.

### E. Sequencing

Sequencing is performed using off-board clustering on the Illumina cBot with patterned flow cell technology to generate monoclonal clusters from a single DNA template followed by sequencing using sequencing by synthesis (SBS) chemistry on the Illumina HiSeq 4000. Fluorescently labeled 3'-blocked dNTPs along with a polymerase are incorporated through the flow cell to create a growing nucleotide chain that is excited by a laser. A camera captures the emission color of the incorporated base and then is cleaved off. The terminator is then removed to allow the nucleotide to revert to its natural form and to allow the polymerase to add another base to the growing chain. A new pool of fluorescently labeled 3'-blocked dNTPs are added with each new sequencing cycle. The color changes for each new cycle as a new base is added to the growing chain. This method allows for millions of discrete clusters of clonal copies of DNA to be sequenced in parallel.

### F. Sequence Analysis

Sequence data are analyzed using proprietary software developed by FMI. Sequence data are mapped to the human genome (hg19) using Burrows-Wheeler Aligner (BWA) v0.5.9.⁴ PCR duplicate read removal and sequence metric collection are performed using Picard 1.47 (http://picard.sourceforge.net) and SAMtools 0.1.12a.⁵ Local alignment optimization is performed using Genome Analysis Toolkit (GATK) 1.0.4705.⁶ Variant calling is performed only in genomic regions targeted by the test.

Base substitution detection is performed using a Bayesian methodology, which allows for the detection of novel somatic alterations at low mutant allele frequency (MAF) and increased sensitivity for alterations at hotspot sites through the incorporation of tissue-specific prior expectations.⁷ Reads with low mapping (mapping quality < 25) or base calling quality (base calls with quality ≤ 2) are discarded. Final calls are made at MAF ≥ 5% (MAF ≥ 1% at hotspots).

To detect indels, de novo local assembly in each targeted exon is performed using the de-Bruijn approach.⁸ Key steps are:

- Collecting all read-pairs for which at least one read maps to the target region.
- Decomposing each read into constituent k-mers and constructing an enumerable graph representation (de-Bruijn) of all candidate non-reference haplotypes present.
- Evaluating the support of each alternate haplotype with respect to the raw read data to generate mutational candidates. All reads are compared to each of the candidate haplotypes via ungapped alignment, and a read 'vote' for each read is assigned to the candidate with best match. Ties between candidates are resolved by splitting the read vote, weighted by the number of reads already supporting each haplotype. This process is iterated until a 'winning' haplotype is selected.
- Aligning candidates against the reference genome to report alteration calls.

PMA P170019/S016: FDA Summary of Safety and Effectiveness Data

8 of 40

{8}

Filtering of indel candidates is carried out similarly to base substitutions, with an empirically increased allele frequency threshold at repeats and adjacent sequence quality metrics as implemented in GATK: % of neighboring bases mismatches < 25%, average neighboring base quality > 25, average number of supporting read mismatches ≤ 2. Final calls are made at MAF ≥ 5% (MAF ≥ 3% at hotspots).

Copy number alterations (CNAs) are detected using a comparative genomic hybridization (CGH)-like method. First, a log-ratio profile of the sample is acquired by normalizing the sequence coverage obtained at all exons and genome-wide SNPs (~3,500) against a process-matched normal control. This profile is segmented and interpreted using allele frequencies of sequenced SNPs to estimate tumor purity and copy number at each segment. Amplifications are called at segments with ≥ 6 copies (or ≥ 7 for triploid/≥ 8 for tetraploid tumors) and homozygous deletions at 0 copies, in samples with tumor purity ≥ 20%. Amplifications in ERBB2 are called positive at segments with ≥ 5 copies for diploid tumors.

Genomic rearrangements are identified by analyzing chimeric read pairs. Chimeric read pairs are defined as read pairs for which reads map to separate chromosomes, or at a distance of over 10 megabase (Mb). Pairs are clustered by genomic coordinate of the pairs, and clusters containing at least five chimeric pairs (three for known fusions) are identified as rearrangement candidates. Filtering of candidates is performed by mapping quality (average read mapping quality in the cluster must be 30 or above) and distribution of alignment positions. Rearrangements are annotated for predicted function (e.g., creation of fusion gene).

To determine microsatellite instability (MSI) status, 95 intronic homopolymer repeat loci (10-20 bp long in the human reference genome) with adequate coverage on the F1CDx assay are analyzed for length variability and compiled into an overall MSI score via principal components analysis (PCA). Using the 95 loci, for each sample the repeat length is calculated in each read that spans the locus. The means and variances of repeat lengths are recorded. PCA is used to project the 190-dimension data onto a single dimension (the first principal component) that maximizes the data separation, producing an MSI score. Each sample is assigned a qualitative status of MSI-High (MSI-H) or MSI-Stable (MSS); ranges of the MSI score are assigned MSI-H or MSS by manual unsupervised clustering. Samples with low coverage (< 250X median) are assigned a status of MSI-unknown.

Tumor mutational burden (TMB) is measured by counting all synonymous and non-synonymous substitution and indel variants present at 5% allele frequency or greater and filtering out potential germline variants according to published databases of known germline polymorphisms including Single Nucleotide Polymorphism database (dbSNP) and Exome Aggregation Consortium (ExAC). Additional germline alterations still present after database querying are assessed for potential germline status and filtered out using a somatic-germline/zygosity (SGZ) algorithm. Furthermore, known and likely driver mutations are filtered out to exclude bias of the data set. The resulting mutation number is then divided by the coding region

PMA P170019/S016: FDA Summary of Safety and Effectiveness Data

9 of 40

{9}

corresponding to the number of total variants counted, or 793 kb. The resulting number is communicated as mutations per Mb unit (mut/Mb).

After completion of the Analysis Pipeline, variant data are displayed in the FMI custom-developed CATi software applications with sequence QC metrics. As part of data analysis QC for every sample, the F1CDx assay assesses cross-contamination through the use of a SNP profile algorithm, reducing the risk of false-positive calls that could occur as a result of an unexpected contamination event. Sequence data are reviewed by trained bioinformatics personnel. Samples failing any QC metrics are automatically held and not released.

### G. Report Generation

Approved results are annotated by automated software with CDx relevant information and are merged with patient demographic information and any additional information provided by FMI as a professional service prior to approval and release by the laboratory director or designee.

### H. Internal Process Controls Related to the System

#### Positive Control

Each assay run includes a control sample run in duplicate. The control sample contains a pool of ten HapMap cell lines and is used as a positive mutation detection control. 100 different germline SNPs present across the entire targeted region are required to be detected by the analysis pipeline. If SNPs are not detected as expected, this results in a QC failure, as it indicates a potential processing error.

#### Sensitivity Control

The HapMap control pool used as the positive control is prepared to contain variants at 5%-10% MAF which must be detected by the analysis pipeline to ensure the expected sensitivity for each run.

#### Negative Control

Samples are barcoded molecularly at the LC stage. Only reads with a perfect molecular barcode sequence are incorporated into the analysis. The Analysis Pipeline includes an algorithm that analyzes the SNP profile of each specimen to identify potential contamination that may have occurred prior to molecular barcoding and can detect contamination lower than 1%.

### I. Variant Classification

#### Biomarker Rules for SNVs and indels that lead to MET exon 14 skipping

An SNV or indel in MET shall be considered to result in skipping of exon 14 if one or more of the following criteria are met:

1. Deletions greater than or equal to 5 bp that affect positions -3 to -30 in the intronic region immediately adjacent to the splice acceptor site at the 5' boundary of MET exon 14.

PMA P170019/S016: FDA Summary of Safety and Effectiveness Data

10 of 40

{10}

2. Indels affecting positions -1 or -2 at the splice acceptor site of the 5' boundary of MET exon 14.
3. Base substitutions and indels affecting positions 0, +1, +2, or +3 at the splice donor site of the 3' boundary of MET exon 14.

### Homologous Recombination Repair (HRR) Genes

A clinical report is provided to the ordering physician for each F1CDx test performed at Foundation Medicine, Inc. Each report is generated and reviewed by an internal team consisting of clinical bioinformatics analysts, scientists, curators, and pathologists for mutations positive for the therapies identified. Each sample is assessed for mutations in the 14 HRR genes, ATM, BARD1, BRCA1, BRCA2, BRIP1, CDK12, CHEK1, CHEK2, FANCL, PALB2, RAD51B, RAD51C, RAD51D, and RAD54L (Table 5). For these genes, both deleterious and suspected deleterious mutations in short variant, copy number alteration, and rearrangement variant classes are determined by an in-house software pipeline. Alterations listed in the COSMIC database and homozygous deletions are considered deleterious. Suspected deleterious mutations include truncating events (i.e., splice, frameshift, and nonsense alterations), as well as large rearrangements that disrupt the coding sequence. The COSMIC check is a second layer of check for HRR positive suspected deleterious alterations. All splice, nonsense, and frameshift alterations in HRR genes are considered biomarker positive and would be considered as suspected deleterious mutations (or "likely" status in FMI reporting rules). If these mutations are additionally reported in COSMIC, they would be listed as deleterious mutations (or "known" status in FMI reporting).

The F1CDx assay is intended as an aid in selecting prostate cancer patients with deleterious or suspected deleterious HRR variants, identified by the rules below, and who may be eligible for treatment with Lynparza® (olaparib).

Table 5. Mutation types identified in the HRR genes

|  Variant Class | Alteration type | Description*  |
| --- | --- | --- |
|  Short Variant | Nonsense, frameshift, or splice site | Any deleterious nonsense, frameshift, or splicing event that spans or occurs within ±2 bases of the intron/exon junction  |
|   |  Missense or non-frameshift | Any of the mutations listed in Table 6 for ATM, BRCA1, and BRCA2  |
|  Copy Number Alteration | Homozygous copy number loss | Deleterious homozygous copy number loss of one or more exons  |
|  Rearrangement | Rearrangement | Any rearrangement that disrupts protein function  |

*For BRCA2, truncating mutations must occur upstream of bases encoding amino acid 3326. Additionally, the frameshift mutation T367fs*13 in FANCL is ineligible. All short variants must occur in the canonical transcript.

PMA P170019/S016: FDA Summary of Safety and Effectiveness Data

11 of 40

{11}

The specific deleterious mutation (DM) and suspected deleterious mutation (SDM) missense mutations or non-frameshift mutations for BRCA1, BRCA2, and ATM are shown in Table 6, below. However, any missense or non-frameshift mutations in the other 12 genes would not be considered HRR positive.

Table 6. Eligible deleterious mutations in the ATM, BRCA1, and BRCA2 genes

|  ATM | BRCA1 | BRCA2  |
| --- | --- | --- |
|  M1T | M1V | M1R  |
|  R2032K | M1I | M1I  |
|  R2227C | C61G | V159M  |
|  R2547_S2549del | C64Y | V211L  |
|  G2765S | R71G | V211I  |
|  R2832C | R71K | R2336P  |
|  S2855_V2856delinsRI (annotated as S2855_V2856>RI) | R1495M | R2336H  |
|  R3008C | E1559K |   |
|  R3008H | D1692N |   |
|  8418+5_8418+8delGTGA or 8418+1_8418+4delGTGA | D1692H |   |
|   | R1699W |   |
|   | A1708E |   |
|   | G1788V |   |

### VI. ALTERNATIVE PRACTICES AND PROCEDURES

There are FDA-approved companion diagnostic (CDx) alternatives for the detection of genetic alterations using FFPE tumor specimens, as listed in Table 1 of the F1CDx intended use statement. The approved CDx tests are listed in Table 7, below; for additional details see FDA List of Cleared or Approved Companion Diagnostic Devices at: https://www.fda.gov/medical-devices/vitro-diagnostics/list-cleared-or-approved-companion-diagnostic-devices-vitro-and-imaging-tools. Each alternative has its own advantages and disadvantages. Physicians should consider the best method that suits their patients and that best meets their expectations.

PMA P170019/S016: FDA Summary of Safety and Effectiveness Data

12 of 40

{12}

Table 7. List of FDA approved CDx assays for genes targeted by F1CDx

|   | Device | Company | Technology | Therapy | Indication  |
| --- | --- | --- | --- | --- | --- |
|  HER2-Amplification | PathVysion HER-2 DNA Probe Kit | Abbott Molecular, Inc. | FISH | HERCEPTIN (trastuzumab) | Breast cancer  |
|   |  PATHWAY Anti-HER-2/neu (4B5) Rabbit Monoclonal Primary Antibody | Ventana Medical Systems, Inc. | IHC | HERCEPTIN (trastuzumab) | Breast cancer  |
|   |  InSite HER-2/neu Kit | Biogenex Laboratories, Inc. | IHC | HERCEPTIN (trastuzumab) | Breast cancer  |
|   |  SPOT-Light HER2 CISH Kit | Life Technologies, Inc. | CISH | HERCEPTIN (trastuzumab) | Breast cancer  |
|   |  Bond Oracle HER2 IHC System | Leica Biosystems | IHC | HERCEPTIN (trastuzumab) | Breast cancer  |
|   |  HER2 CISH pharmDx Kit | Dako Denmark A/S | CISH | HERCEPTIN (trastuzumab) | Breast cancer  |
|   |  INFORM HER2 Dual ISH DNA Probe Cocktail | Ventana Medical Systems, Inc. | Dual ISH | HERCEPTIN (trastuzumab) | Breast cancer  |
|   |  HercepTest | Dako Denmark A/S | IHC | HERCEPTIN (trastuzumab) PERJETA (pertuzumab) KADCYLA (ado-trastuzumab emtansine) | Breast cancer Gastric or Gastroesophageal junction adenocarcinoma  |
|   |  HER2 FISH pharmDx Kit | Dako Denmark A/S | FISH | HERCEPTIN (trastuzumab) PERJETA (pertuzumab) KADCYLA (ado-trastuzumab emtansine) | Breast cancer Gastric or Gastroesophageal junction adenocarcinoma  |
|  BRAF-V600E and V600K | THxID BRAF Kit | bioMerieux | PCR | MEKINIST (tramatenib) | Melanoma  |
|   |  cobas 4800 BRAF V600 Mutation Test | Roche Molecular Systems, Inc. | PCR | COTELLIC (cobimetinib) ZELBORAF (vemurafenib) | Melanoma  |
|  BRAF-V600E | cobas 4800 BRAF V600 Mutation Test | Roche Molecular Systems, Inc. | PCR | ZELBORAF (vemurafenib) | Melanoma  |
|   |  THxID BRAF Kit | bioMerieux | PCR | TAFINLAR (dabrafenib) | Melanoma  |

PMA P170019/S016: FDA Summary of Safety and Effectiveness Data

13 of 40

{13}

|   | Device | Company | Technology | Therapy | Indication  |
| --- | --- | --- | --- | --- | --- |
|  **BRAF-V600E** | Oncomine Dx Target Test | Life Technologies, Inc. | NGS | TAFINLAR (dabrafenib) MEKINIST (trametinib) | NSCLC  |
|   |  *therascreen* BRAF V600E RGQ PCR Kit | QIAGEN | PCR | BRAFTOVI (encorafenib) Erbitux (cetuximab) | Colorectal cancer  |
|  **NRAS** | Praxis Extended RAS Panel | Illumina, Inc. | NGS | VECTIBIX (panitumumab) | Colorectal cancer  |
|  **KRAS** | **cobas** KRAS Mutation Test | Roche Molecular Systems, Inc. | PCR | ERBITUX (cetuximab) VECTIBIX (panitumumab) | Colorectal cancer  |
|   |  *therascreen* KRAS RGQ PCR Kit | QIAGEN | PCR | ERBITUX (cetuximab) VECTIBIX (panitumumab) | Colorectal cancer  |
|   |  Praxis Extended RAS Panel | Illumina, Inc. | NGS | VECTIBIX (panitumumab) | Colorectal cancer  |
|  **ALK – fusion** | Vysis ALK Break Apart FISH Probe Kit | Abbott Molecular, Inc. | FISH | XALKORI (crizotinib) | NSCLC  |
|   |  ALK (D5F3) CDx Assay | Ventana Medical Systems, Inc. | IHC | XALKORI (crizotinib) | NSCLC  |
|  **EGFR – Exon 19 deletions & L858R** | **cobas** EGFR Mutation Test v2 | Roche Molecular Systems, Inc. | PCR | TARCEVA (erlotinib) TAGRISO (osimertinib) IRESSA (gefitinib) | NSCLC  |
|   |  *therascreen* EGFR RGQ PCR Kit | QIAGEN | PCR | GILOTRIF (afatinib) IRESSA (gefitinib) | NSCLC  |
|   |  Oncomine Dx Target Test | Life Technologies, Inc. | NGS | IRESSA (gefitinib) | NSCLC  |
|  **EGFR T790M** | **cobas** EGFR Mutation Test v2 | Roche Molecular Systems, Inc. | PCR | TAGRISO (osimertinib) | NSCLC  |
|  **BRCA1/2** | FoundationFocus CDx_{BRCA} | Foundation Medicine, Inc. | NGS | RUBRACA (rucaparib) | Advanced ovarian cancer  |

PMA P170019/S016: FDA Summary of Safety and Effectiveness Data

14 of 40

{14}

|   | Device | Company | Technology | Therapy | Indication  |
| --- | --- | --- | --- | --- | --- |
|  PIK3CA | therascreen PIK3CA RGQ PCR Kit | QIAGEN | PCR | PIQRAY (alpelisib) | Breast cancer  |

**Abbreviations:** FISH – fluorescence *in situ* hybridization; IHC – immunohistochemistry; CISH – chromogenic *in situ* hybridization; ISH – *in situ* hybridization; PCR – polymerase chain reaction; NGS – next generation sequencing.

## VII. MARKETING HISTORY

Foundation Medicine, Inc. initially designed and developed the FoundationOne® laboratory developed test (F1 LDT), and the first commercial sample was tested in 2012. The F1 LDT has been used to detect the presence of genomic alterations in FFPE tumor tissue specimens. The F1 LDT is not FDA-cleared or -approved.

The F1CDx Premarket Approval (PMA) was originally approved on November 30, 2017 by FDA (P170019) and is commercially available in the U.S. since March 30, 2018. The following PMA supplements affecting the Intended Use were approved by FDA.

- • P170019/S005 was approved on April 10, 2019.
- • P170019/S004 and P170019/S008 were approved on July 1, 2019.
- • P170019/S009 was approved on August 21, 2019.
- • P170019/S006 was approved on December 3, 2019.
- • P170019/S010 was approved on December 16, 2019.
- • P170019/S013 was approved on April 17, 2020.
- • P170019/S011 was approved on May 6, 2020.
- • P170019/S015 was approved on May 19, 2020.

## 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 test results and, subsequently, inappropriate patient management decisions. Patients with false positive results may undergo treatment with one of the therapies listed in the above intended use statement without clinical benefit and may experience adverse reactions associated with the therapy. Patients with false negative results may not be considered for treatment with the indicated therapy. There is also a risk of delayed results, which may lead to delay of treatment with the indicated therapy. For the specific adverse events related to the approved therapeutics, please see the approved drug product labels.

## IX. SUMMARY OF NONCLINICAL STUDIES

### A. Laboratory Studies

The performance of F1CDx in detecting high TMB (TMB-H) as a qualitative pan tumor biomarker with respect to the 10 mut/Mb cut-off is supported by the data

PMA P170019/S016: FDA Summary of Safety and Effectiveness Data

15 of 40

{15}

presented using a broad range of tumor specimens across all validation studies. Over 400 unique samples comprising over 200 disease ontologies, including the rare cancers enrolled in the clinical validation study (please refer to Section X), were included in the analytical validation for TMB as a pan tumor biomarker. Samples had TMB scores ranging from 0 to 375 mut/Mb across a spectrum of tumor purities, from 4.8% to 99.9% (please refer to Section IX.A.2(a), below, for information on the detection limits of the TMB biomarker), as well as sufficient pre-analytical (e.g., percent tumor nuclei, DNA input) and post-sequencing (e.g., tumor purity, $\geq 100\text{X}$ coverage, median coverage) QC metrics to support robust F1CDx TMB calling across the intended use population. Analytical accuracy/concordance, limit of detection (LoD), and precision studies as well as analyses of DNA extraction and interfering substances data were conducted to support the indication for TMB as a pan tumor biomarker at the cut-off of $\geq 10$ mut/Mb for TMB-H tumors.

For the F1CDx platform validation (P170019), device performance characteristics were established using DNA derived from a wide range of FFPE tissue types, and tissue types associated with CDx indications were included in each study. Each study included CDx variants as well as a broad range of representative alteration types, including substitution and insertion and deletion variants, in various genomic contexts across several genes. Analyses of genomic signatures including MSI and TMB were also conducted. TMB was previously analyzed as a score in mutations per megabase (mut/Mb) to support tumor profiling for the F1CDx platform (P170019). Results from the platform-level validation (P170019) have been leveraged, including precision, interfering substances, and DNA extraction studies, to support F1CDx detection of TMB as a qualitative biomarker with respect to the 10 mut/Mb cut-off for TMB-H samples. A post-market study will be conducted to support quantitative TMB score reporting for the CDx biomarker. For information regarding the F1CDx platform-level validation, please see Summary of Safety and Effectiveness Data P170019.

## 1. Analytical Accuracy/Concordance

### a. Comparison to Whole Exome Sequencing for TMB Calling

An analytical accuracy study was performed to demonstrate the concordance between F1CDx and an externally validated whole exome sequencing (WES) assay for the detection of TMB-H as a qualitative biomarker. The WES assay sequenced matched tumor-normal samples to determine germline mutations in sample pairs, and germline mutations were filtered from the tumor sample results prior to the TMB calculation. The WES TMB algorithm included a variant calling threshold of 5% allele frequency or greater for SNVs and indels. Only mutations in coding regions were included in the TMB WES score. The final alteration count was converted to mut/Mb by dividing by 34.7 Mb, which was the total length of the coding regions in the WES assay.

A total of 218 samples with valid F1CDx and WES TMB scores were evaluated that represented prevalent tumor types as well as the rare tumors from the clinical validation study. 64 samples obtained from rare tumors were assessed, including 45 samples that were screened for patient enrollment into

---PMA P170019/S016: FDA Summary of Safety and Effectiveness Data

16 of 40

{16}

the KEYNOTE-158 clinical study, which supports the clinical validation of the F1CDx assay for TMB (please refer to Section X, below). 154 samples represented common tumor types including NSCLC, breast cancer, CRC, ovarian cancer, and melanoma. Of the 218 samples in the concordance analysis, 89 were selected by a non-FMI assay, and the remaining 129 samples were selected by F1CDx. TMB measurements were dichotomized using 10 mut/Mb as the TMB-H cut-off for both the F1CDx and WES assays. The positive percent agreements (PPA) and negative percent agreements (NPA) with 95% confidence intervals (CI) derived through bootstrapping are provided for the sample set selected by a non-FMI assay (unadjusted for prevalence) and samples selected by F1CDx (adjusted for prevalence of TMB-H estimated as 19%) in Tables 8 and 9, respectively.

**Table 8. TMB concordance summary for samples selected by a non-FMI assay (unadjusted for prevalence)**

|  TMB Cut-off | F1CDx+/WES+ | F1CDx-/WES+ | F1CDx+/WES- | F1CDx-/WES- | Unadjusted PPA (95% CI) | Unadjusted NPA (95% CI)  |
| --- | --- | --- | --- | --- | --- | --- |
|  10 mut/Mb | 28 | 7 | 4 | 50 | 80.0% (62.5%, 90.62%) | 92.59% (82.62%, 98.04%)  |

**Table 9. TMB concordance summary for samples selected by F1CDx (adjusted for prevalence)**

|  TMB Cut-off | F1CDx+/WES+ | F1CDx-/WES+ | F1CDx+/WES- | F1CDx-/WES- | Adjusted PPA (95% CI) | Adjusted NPA (95% CI)  |
| --- | --- | --- | --- | --- | --- | --- |
|  10 mut/Mb | 23 | 1 | 17 | 88 | 92.31% (65.74%, 100%) | 90.84% (87.76%, 93.99%)  |

The overall PPA and NPA based on a weighted average of the results (unadjusted and adjusted for TMB-H prevalence) in the TMB concordance analysis are provided in Table 10, below, with 95% bootstrap CI.

**Table 10. Overall TMB concordance summary (weighted average)**

|  TMB Cut-off | Overall PPA (95% CI) | Overall NPA (95% CI)  |
| --- | --- | --- |
|  10 mut/Mb | 87.28% (64.42%, 96.17%) | 91.56% (85.66%, 95.64%)  |

Overall PPA was 87.28% with 95% CI (64.42%, 96.17%) and overall NPA was 91.56% with 95% CI (85.66%, 95.64%). Passing-Bablok regression was performed to assess the relationship between F1CDx and WES using the underlying continuous TMB score. The estimated slope was 0.93 with 95% CI (0.87, 1.03), and the estimated offset was -0.08 mut/Mb with 95% CI (-0.46, 0.2).

PMA P170019/S016: FDA Summary of Safety and Effectiveness Data

17 of 40

{17}

29 discordant results were observed in the TMB concordance analysis. An investigation of the discordances determined that 10 of the 29 discordant samples had F1CDx TMB scores that were close to the cut-off of 10 mut/Mb for TMB-H samples. Further, an analysis of the alteration level agreements was performed to explore the differences in the underlying component variants included in the TMB scores determined by the F1CDx and WES TMB algorithms. The alteration level assessment identified discordances in variant calls between the F1CDx and WES assays due to driver mutation exclusions, differences in germline variant designation, variants detected at low allele frequencies (i.e., below the F1CDx or WES TMB algorithm threshold of 5% MAF), and mutations in non-coding regions. The alteration level evaluation determined that exclusion of driver mutations by the F1CDx TMB algorithm and not by the WES TMB algorithm as well as differences in germline variant designation to be the greatest sources of TMB component variant discordances between F1CDx and WES.

## 2. Analytical Sensitivity

### a. Limit of Detection (LoD)

The limit of detection (LoD) for TMB calling by F1CDx based on the cut-off of 10 mut/Mb was estimated with respect to computational tumor purity. Computational tumor purity is calculated by fitting the observed log-ratio and minor allele frequency data with statistical models that predict a genome-wide copy number profile, tumor ploidy, and tumor purity (i.e., computational tumor purity). The log-ratio profile is obtained by normalizing aligned tumor sequence reads by dividing read depth by that of a process-matched normal control, followed by a GC-content bias correction using Loess regression. The minor allele frequency profile is obtained from the heterozygous genome-wide SNPs.

11 TMB-H ($\geq 10$ mut/Mb) FFPE samples were included in the analysis representing lung, colon, breast, bladder, and skin cancers as well as rare tumor types in the clinical validation study including anal and endometrial tissues. LoD was assessed at 5 levels of computational tumor purity, ranging from 2.5% to 50%, with 20 replicates per level, except for the highest tumor purity level at which 14 replicates were tested. The LoD for TMB calling based on computational tumor purity was determined empirically by the hit rate method, defined as the lowest level with 95% hit rate.

The LoD for TMB calling based on computational tumor purity is 28.16%. The LoD may be driven by the proximity of the TMB score to the cut-off, as the sample with an average TMB score of 11.83 mut/Mb (at the highest level of tumor purity evaluated) had the highest hit rate LoD at 28.16% tumor purity for the cut-off of 10 mut/Mb.

Several samples were tested at or near the TMB LoD based on computational tumor purity in the precision study (please see Section IX.A.5). Please also

---PMA P170019/S016: FDA Summary of Safety and Effectiveness Data

18 of 40

{18}

refer to Section IX.A.2 of Summary of Safety and Effectiveness Data for P170019 for additional analytical sensitivity data.

# b. Limit of Blank (LoB)

The limit of blank (LoB) of zero was confirmed by demonstrating that the percentage of false positive results did not exceed 5% (type I error risk α=0.05). 21 biomarker-negative samples were processed for a total of 220 replicates. Of the 220 aliquots, one failed prior to sequencing, and 219 replicates were available for the LoB analysis. 16 of the 21 biomarker-negative samples had TMB scores of 0 mut/Mb, and the mean TMB score for all 21 samples evaluated was 0.14 mut/Mb. All 219 replicates were below the TMB cut-off of 10 mut/Mb (non-TMB-H), resulting in a false positive rate of 0% and thus confirming the LoB of zero.

# 3. Analytical Specificity

# a. Interfering Substances

To evaluate the potential impact of endogenous and exogenous interfering substances on the performance of the F1CDx assay for TMB calling, an in silico assessment of interfering substances data for the F1CDx platform (P170019) was conducted. A total of 19 FFPE specimens were included in the evaluation representing a range of tumor types including colon, breast, lung, ovary, skin, liver, gastroesophageal, kidney, and prostate tissues. Two TMB-H samples, including one sample close to the 10 mut/Mb threshold for TMB-H, were analyzed. The addition of interfering substances, including melanin (endogenous), ethanol (exogenous), proteinase K (exogenous), and molecular index barcodes (MIB) (exogenous), was evaluated to determine if they impacted TMB calling with respect to the qualitative cut-off of 10 mut/Mb (Table 11).

Table 11. Interfering substances evaluated

|  Substance | Level | # Samples | # Replicates/Sample  |
| --- | --- | --- | --- |
|  No interferent | - | 19 | 2 (4 controls for MIB)  |
|  Melanin | 0.025 μg/mL | 5 | 2  |
|  Melanin | 0.05 μg/mL | 5 | 2  |
|  Melanin | 0.1 μg/mL | 5 | 2  |
|  Melanin | 0.2 μg/mL | 5 | 2  |
|  Proteinase K | 0.04 mg/mL | 18 | 2  |
|  Proteinase K | 0.08 mg/mL | 18 | 2  |
|  Ethanol | 2.5% | 18 | 2  |
|  Ethanol | 5% | 18 | 2  |
|  MIB | 0 | 18 | 2 or 4  |
|  MIB | 5% | 5 | 4  |
|  MIB | 15% | 18 | 2 or 4  |
|  MIB | 30% | 18 | 2 or 4  |

PMA P170019/S016: FDA Summary of Safety and Effectiveness Data

19 of 40

{19}

Each interfering substance was evaluated by the concordance of TMB status with the corresponding control sample with respect to the 10 mut/Mb cut-off. A summary of the overall concordance results for each interferent with corresponding 95% CI using score method is provided in Table 12, below.

**Table 12. TMB concordance across each interfering substance**

|  Substance | # Concordant | # Total | Concordance (95% score CI)  |
| --- | --- | --- | --- |
|  No interferent | 55 | 56 | 98.2% (90.6%, 99.7%)  |
|  Melanin | 38 | 40 | 95% (83.5%, 98.6%)  |
|  Proteinase K | 72 | 72 | 100% (94.9%, 100%)  |
|  Ethanol | 72 | 72 | 100% (94.9%, 100%)  |
|  MIB | 110 | 111 | 99.1% (95.1%, 99.8%)  |

The concordance for TMB calling was 100% for the proteinase K and ethanol evaluations, 99.1% for the MIB evaluation, and 95% for the melanin evaluation.

A post-market interfering substances study will be performed to evaluate the effects of endogenous interfering substances, including necrotic tissue and hemoglobin, on TMB calling ($\geq 10$ mut/Mb) in solid tumors across the intended use population.

#### 4. Carryover/Cross-Contamination

Please see Section IX.A.4 of Summary of Safety and Effectiveness Data for P170019.

#### 5. Precision and Reproducibility

##### a. Intermediate Precision for TMB-H ($\geq 10$ mut/Mb)

A precision study was conducted to evaluate the intermediate precision for TMB-H calling with respect to the 10 mut/Mb cut-off in FFPE specimens across solid tumors. Samples were enriched for TMB-H status, and an additional analysis was performed to support F1CDx precision for TMB calling in non-TMB-H samples (please see Section IX.A.5(b), below). 46 samples were analyzed that represented prevalent solid tumors as well as the rare tumors included in the clinical validation study (please see Section X) covering 7 major organ systems: gastrointestinal, hepato-pancreatobiliary, urinary, endocrine, skin, thoracic, and reproductive. The study focused on the evaluation of F1CDx precision for TMB-H calling across challenging samples at low DNA input (i.e., close to the minimum requirement of 50 ng) with a range of TMB scores, including samples with TMB scores close to the 10

PMA P170019/S016: FDA Summary of Safety and Effectiveness Data

20 of 40

{20}

mut/Mb cut-off, and a range of tumor purities. Of the 46 samples, 22 were tested at challenging tumor purity levels near or below the established TMB LoD of 28.16% computational purity (please refer to Section IX.A.2(a), above).

Repeatability including intra-run performance (run on the same plate under the same conditions) and reproducibility including inter-run performance (run on different plates under different conditions) were assessed and compared across three different sequencers and two different reagent lots, over multiple days (typical assay workflow spans 10 days). A full factorial study design was executed with four replicates per reagent lot/sequencer combination for a total of 24 replicates per sample. The previous precision studies for F1CDx (P170019) and FoundationFocus CDx$_{BRCA}$ (P160018) were conducted with 36 replicates using a full factorial study design and yielded high agreement rates; thus, 24 replicates per sample to demonstrate F1CDx precision for TMB calling were deemed acceptable to support this PMA supplement due to adequate F1CDx platform precision in the previous studies.

Based on the cut-off of 10 mut/Mb for TMB-H, there were 44 TMB-H and 2 non-TMB-H samples in this precision analysis. The TMB scores for the selected samples ranged from 6.25 mut/Mb to 156.4 mut/Mb (based on the mean TMB score for valid replicates per sample), and 6 samples had TMB scores near the TMB-H threshold of 10 mut/Mb. Repeatability and reproducibility results with the corresponding two-sided exact 95% CI are summarized in Table 13, below.

**Table 13. Precision results for TMB-H ($\geq 10$ mut/Mb)**

|  TMB Cut-off | Repeatability Positive Call Rate (95% exact CI) | Repeatability Negative Call Rate (95% exact CI) | Reproducibility Positive Call Rate (95% exact CI) | Reproducibility Negative Call Rate (95% exact CI)  |
| --- | --- | --- | --- | --- |
|  10 mut/Mb | 99.61% (98.61%, 99.95%) | 95.83% (78.88%, 99.89%) | 99.81% (99.3%, 99.98%) | 97.83% (88.47%, 99.94%)  |

For repeatability, the PPA for 44 TMB-H samples was 99.61% with 95% CI (98.61%, 99.95%), and the NPA for 2 non-TMB-H samples was 95.83% with 95% CI (78.88%, 99.89%). For reproducibility, the PPA was 99.81% with 95% CI (99.3%, 99.98%), and the NPA was 97.83% with 95% CI (88.47%, 99.94%). The overall repeatability agreement for samples tested was 99.54% with 95% CI (98.39%, 99.98%), and overall reproducibility was 99.72% with 95% CI (99.18%, 99.94%).

For repeatability and reproducibility of the TMB score as a continuous variable, the coefficient of variance (%CV) from variance component analysis was estimated for each sample. All samples had %CV < 30% for repeatability and reproducibility.

PMA P170019/S016: FDA Summary of Safety and Effectiveness Data

21 of 40

{21}

# **b. Intermediate Precision for non-TMB-H (< 10 mut/Mb)**

To provide additional precision data for F1CDx TMB calling in non-TMB-H samples (< 10 mut/Mb), an analysis was conducted in 20 FFPE specimens across multiple tumor types including lung, colon, skin, thyroid, salivary gland, liver, and uterus cancers. The study included challenging samples with low DNA input (i.e., close to the minimum requirement of 50 ng) and a range of tumor purities. Repeatability including intra-run performance and reproducibility including inter-run performance were assessed and compared across three different sequencers, two different reagent lots, and multiple days (typical assay workflow spans 10 days). A full factorial study design was executed with three replicates per two runs for a total of 36 replicates across the paired reagent lot/sequencer combinations.

Of the 20 samples tested in the additional precision analysis to support F1CDx TMB calling in non-TMB-H samples, there were 1 TMB-H and 19 non-TMB-H samples based on the cut-off of ≥ 10 mut/Mb for TMB-H. Overall repeatability and reproducibility results with the corresponding two-sided exact 95% CI are summarized in Table 14, below.

**Table 14. Precision results for non-TMB-H (< 10 mut/Mb)**

|  TMB Cut-off | Overall Repeatability (95% exact CI) | Overall Reproducibility (95% exact CI)  |
| --- | --- | --- |
|  10 mut/Mb | 100% (98.4%, 100%) | 100% (99.5%, 100%)  |

Repeatability and reproducibility of TMB status with respect to the 10 mut/Mb cut-off was 100% for all samples. The overall repeatability agreement for samples tested was 100% with 95% CI (98.4%, 100%), and overall reproducibility was 100% with 95% CI (99.5%, 100%).

# **c. Site-to-site Precision for TMB**

A site-to-site precision study was performed to support F1CDx TMB calling at the second site in Morrisville, NC. 46 FFPE samples representing ovarian, breast, lung, colorectal, and skin cancers across a range of tumor purities were included. A total of 9 TMB-H samples and 37 non-TMB-H samples were evaluated for the site-to-site precision analysis. Repeatability including intra-run performance and reproducibility including inter-run performance were assessed. Each of the 46 samples was tested at two sites (Cambridge, MA and Morrisville, NC) with two replicates, two reagent lots, and on three non-consecutive days by multiple operators. A full factorial study design was conducted with a total of 24 replicates across the paired reagent lot/sequencer combination. For the evaluation of TMB with respect to the 10 mut/Mb cut-off, repeatability and reproducibility was 100% for all replicates for 45 of 46 samples. Only one sample demonstrated 79.17% reproducibility and 58.33%

PMA P170019/S016: FDA Summary of Safety and Effectiveness Data

22 of 40

{22}

repeatability with 5 discordant replicates near the 10 mut/Mb threshold; this sample had an average computational tumor purity of 11.02%, which was below the TMB LoD level of 28.16% tumor purity (see Section IX.A.2(a), above). The overall repeatability agreement for samples tested was 99.05% with 95% CI (97.79%, 99.69%), and overall reproducibility was 99.53% with 95% CI (98.91%, 99.85%).

Additional post-market data will be provided for the intermediate precision of the TMB component alterations, including repeatability and reproducibility.

#### **6. Reagent Lot Interchangeability**

There were no changes to the reagents and specifications between the FoundationFocus™ CDx$_{BRCA}$ assay and F1CDx. Therefore, for reagent lot interchangeability results, please see Section IX.A(g) of Summary of Safety and Effectiveness Data for P160018.

#### **7. Stability**

Please refer to the Summary of Safety and Effectiveness Data P170019 (Section IX.A.7(a,b)) for F1CDx platform validation of reagent, DNA, and FFPE slide stability. Additional post-market data for DNA and FFPE slide stability with respect to TMB biomarker ($\geq 10$ mut/Mb) calling will be provided based on the re-analysis of the existing F1CDx platform-level data with the updated bioinformatics pipeline (v3.3.x).

#### **8. General Lab Equipment and Reagent Evaluation**

##### **a. DNA Amplification**

There were no changes to the reagents and specifications between the FoundationFocus™ CDx$_{BRCA}$ assay and F1CDx. For equipment and reagent interchangeability results, please see Section IX.A.h(a) of Summary of Safety and Effectiveness Data for P160018.

##### **b. DNA Extraction**

The performance of DNA extraction from FFPE tumor specimens was measured by the concordance of TMB status based on the qualitative TMB-H cut-off of 10 mut/Mb. 35 FFPE specimens represented a range of tissue types including lung, breast, ovarian, colorectal, bladder, brain, liver, pancreas, thyroid, prostate, and skin cancers. The study included 7 TMB-H samples, including one sample near the threshold of 10 mut/Mb, as well as challenging samples at low tumor purities. Samples were run in duplicate employing two different KingFisher Flex Magnetic Particle Processors and comparing across two or three extraction reagent lots. Concordance of TMB calling was analyzed across replicates for each sample, and the overall results with respect to TMB status are summarized in Table 15, below.

---PMA P170019/S016: FDA Summary of Safety and Effectiveness Data

23 of 40

{23}

Table 15. Summary of TMB concordance across replicates in DNA extraction study

|  TMB Status | # of Concordant Replicates | # of Total Replicates | Concordance Rate (95% 2-sided score CI)  |
| --- | --- | --- | --- |
|  TMB-H | 63 | 63 | 100% (94.25%, 100%)  |
|  non-TMB-H | 285 | 290 | 98.28% (96.03%, 99.26%)  |

For additional details on the F1CDx platform DNA extraction study, please refer to Section IX.A.8(b) of Summary of Safety and Effectiveness Data P170019.

### 9. Guard banding/Robustness

Please see Section IX.A.9 of Summary of Safety and Effectiveness Data for P170019. Additional post-market data for guard banding with respect to TMB biomarker (≥ 10 mut/Mb) calling will be provided based on the re-analysis of the existing F1CDx platform-level data with the updated bioinformatics pipeline (v3.3.x).

### B. Animal Studies

No animal studies were conducted using the F1CDx assay.

### C. Additional Studies

No additional studies were conducted using the F1CDx assay.

### X. SUMMARY OF PRIMARY CLINICAL STUDY

The clinical performance of FoundationOne®CDx (F1CDx) for detecting TMB-H (defined as TMB ≥ 10 mut/Mb) in patients with solid tumors was demonstrated in a prospectively-planned retrospective analysis of specimens from patients enrolled in the KEYNOTE-158 clinical study of pembrolizumab. Data generated from the KEYNOTE-158 trial supported the clinical validation of the F1CDx assay for the identification of TMB-H subjects (≥ 10 mut/Mb) with solid cancers who may benefit from pembrolizumab treatment.

### A. FoundationOne®CDx Retrospective Analysis of TMB in KEYNOTE-158

The safety and effectiveness of F1CDx for detecting TMB-H patients with respect to the cut-point of 10 mut/Mb was demonstrated in a prospectively-planned retrospective analysis of patients enrolled in the KEYNOTE-158 study. TMB biomarker analysis was pre-specified in the clinical trial protocol and retrospectively analyzed by the F1CDx assay.

### 1. Study Design

KEYNOTE-158 is an ongoing, Phase 2, multicenter, non-randomized, open-label, multi-cohort study designed to evaluate the safety and efficacy of pembrolizumab monotherapy (200 mg intravenously [IV] every 3 weeks [Q3W]) in multiple types of advanced (unresectable or metastatic) solid cancers that have progressed

PMA P170019/S016: FDA Summary of Safety and Effectiveness Data

24 of 40

{24}

following prior treatment and who have no satisfactory alternative treatment options. The study included participants with solid tumors grouped into 10 different cohorts, A to J. The aim of KEYNOTE-158 is to evaluate predictive biomarkers for response to pembrolizumab monotherapy treatment across multiple tumor types, regardless of specific tumor histology.

Participants with any of the tumor types in cohorts A to J listed below for whom previous first-line treatment failed were eligible to enroll in KEYNOTE-158.

Cohorts:

A. Anal squamous cell carcinoma
B. Biliary adenocarcinoma (gallbladder or biliary tree [intrahepatic or extrahepatic] cholangiocarcinoma) except Ampulla of Vater cancers
C. Neuroendocrine tumors (well- and moderately-differentiated) of the lung, appendix, small intestine, colon, rectum, or pancreas
D. Endometrial carcinoma (sarcomas and mesenchymal tumors are excluded)
E. Cervical squamous cell carcinoma
F. Vulvar squamous cell carcinoma
G. Small cell lung carcinoma
H. Mesothelioma
I. Thyroid carcinoma
J. Salivary gland carcinoma (sarcomas and mesenchymal tumors are excluded)

In addition to having any of the advanced solid tumors noted above, participants were eligible for study enrollment if they had an Eastern Cooperative Oncology Group Performance Status (ECOG PS) of 0 or 1; life expectancy of > 3 months and previous treatment with standard therapies; no known active central nervous system (CNS) metastasis, autoimmune disease, immunosuppressive therapy, or prior treatment with anticancer monoclonal antibody (mAb) within 4 weeks prior to study Day 1; and had adequate tumor tissue to test programmed death ligand-1 (PD-L1), gene expression profile (GEP), and microsatellite instability-high (MSI-H). The TMB biomarker was pre-specified and retrospectively analyzed by the F1CDx assay.

The primary endpoint was objective response rate (ORR), defined as the proportion of participants in the analysis population who had a response (complete response [CR] or partial response [PR]) as measured by blinded independent central review (BICR) per Response Evaluation Criteria in Solid Tumors (RECIST) 1.1. Secondary endpoints were as follows:

- Duration of response (DOR), defined as the time from the first documented evidence of CR or PR until disease progression or death due to any cause, whichever occurred first;
- Progression free survival (PFS), defined as the time from randomization/the first dose of study treatment to documented progressive disease or death due to any cause, whichever occurred first; and

PMA P170019/S016: FDA Summary of Safety and Effectiveness Data

25 of 40

{25}

• Overall survival (OS), defined as the time from randomization/the first dose of study treatment to death due to any cause.

## 2. TMB Evaluation by F1CDx

Evaluation of the TMB biomarker was an objective for KEYNOTE-158. The interim analysis (IA)10 investigated the relationship between response to pembrolizumab treatment and TMB as determined by the F1CDx assay using a cut-point of 10 mut/Mb. Selection of 10 mut/Mb as the TMB score cut-point for TMB-H considered three key elements: 1) the ability to reliably enrich for objective response to pembrolizumab while preserving the percentage of responders (i.e., sensitivity) based on training data; 2) the relationship between TMB and inflammation in the tumor microenvironment as measured by gene expression; and 3) an effort to harmonize on a common TMB cut-off to define TMB-H across different sponsors working with TMB as an immunotherapy biomarker.

Pre-defined sample quality control (QC) testing parameters were used throughout the retrospective testing of KEYNOTE-158 clinical trial samples including, but not limited to, minimum DNA input, tumor purity, tissue volume, and average read depth to ensure adequate F1CDx performance of TMB calling.

The therapeutic efficacy (TE) population was the primary efficacy population to support the drug indication. For investigational testing and analysis of the TE population, a TMB result was valid if the following pre-specified sample inclusion criteria were met:

- Tissue volume ≥ 0.2 mm³;
- Tumor nuclei content ≥ 10%; and
- Extracted DNA content > 55 ng.

For testing and analysis of the device validation (DV) population, a TMB result was considered valid if the following pre-specified sample inclusion criteria consistent with the F1CDx device specifications were met:

- Tissue volume ≥ 0.6 mm³;
- Tumor nuclei content ≥ 20%; and
- Extracted DNA content > 55 ng.

In addition, all TE and DV population samples had sufficient DNA quantity and quality following extraction, in-process QC metrics, and acceptable post-sequencing metrics to be considered valid.

### B. Study Population Demographics and Baseline Parameters

Baseline demographics and disease characteristics were generally similar in the TMB-H and non-TMB-H populations for both the TE and DV analysis populations (Table 16). At the time of IA10, the trial was being conducted globally in 21 countries, including the U.S., of which 81 sites had enrolled participants to the study treatment. Compared to the non-TMB-H population, the TMB-H population had a

PMA P170019/S016: FDA Summary of Safety and Effectiveness Data

26 of 40

{26}

higher percentage of participants whose tumors were PD-L1 positive (CPS ≥ 1). This is consistent with the understanding of the biological mechanisms by which tumors with increased neoantigens up-regulate PD-L1 to evade immune recognition. As expected, all participants with MSI-H tumors were in the TMB-H group, as MSI-H is a subset of the overall TMB-H population.

**Table 16. Comparison of demographic and disease characteristics between TMB-H and non-TMB-H subjects evaluated by F1CDx for TE and DV populations**

|   | TE Population |   |   |   | DV Population  |   |   |   |
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
|   |  TMB >=10 mut/Mb |   | TMB <10 mut/Mb |   | TMB >=10 mut/Mb |   | TMB <10 mut/Mb  |   |
|   |  n | (%) | n | (%) | n | (%) | n | (%)  |
|  Subjects in population | 102 |  | 688 |  | 91 |  | 628 |   |
|  **Gender**  |   |   |   |   |   |   |   |   |
|  Male | 35 | (34.3) | 253 | (36.8) | 31 | (34.1) | 226 | (36.0)  |
|  Female | 67 | (65.7) | 435 | (63.2) | 60 | (65.9) | 402 | (64.0)  |
|  **Age (Years)**  |   |   |   |   |   |   |   |   |
|  < 65 | 67 | (65.7) | 414 | (60.2) | 59 | (64.8) | 383 | (61.0)  |
|  >= 65 | 35 | (34.3) | 274 | (39.8) | 32 | (35.2) | 245 | (39.0)  |
|  Mean | 60.0 |  | 60.2 |  | 59.7 |  | 60.1 |   |
|  SD | 10.7 |  | 12.0 |  | 11.2 |  | 12.0 |   |
|  Median | 61.0 |  | 61.0 |  | 60.0 |  | 61.0 |   |
|  Range | 27 to 80 |  | 22 to 87 |  | 27 to 80 |  | 22 to 85 |   |
|  **Race**  |   |   |   |   |   |   |   |   |
|  American Indian Or Alaska Native | 0 | (0.0) | 4 | (0.6) | 0 | (0.0) | 3 | (0.5)  |
|  Asian | 17 | (16.7) | 133 | (19.3) | 15 | (16.5) | 118 | (18.8)  |
|  Black Or African American | 1 | (1.0) | 18 | (2.6) | 1 | (1.1) | 15 | (2.4)  |
|  Multiple | 0 | (0.0) | 6 | (0.9) | 0 | (0.0) | 5 | (0.8)  |
|  American Indian Or Alaska Native, Black Or African American | 0 | (0.0) | 1 | (0.1) | 0 | (0.0) | 1 | (0.2)  |
|  Black Or African American, White | 0 | (0.0) | 5 | (0.7) | 0 | (0.0) | 4 | (0.6)  |
|  Native Hawaiian Or Other Pacific Islander | 1 | (1.0) | 1 | (0.1) | 1 | (1.1) | 1 | (0.2)  |
|  White | 83 | (81.4) | 526 | (76.5) | 74 | (81.3) | 486 | (77.4)  |
|  **Ethnicity**  |   |   |   |   |   |   |   |   |
|  Hispanic Or Latino | 2 | (2.0) | 45 | (6.5) | 1 | (1.1) | 40 | (6.4)  |
|  Not Hispanic Or Latino | 92 | (90.2) | 587 | (85.3) | 83 | (91.2) | 538 | (85.7)  |
|  Not Reported | 8 | (7.8) | 56 | (8.1) | 7 | (7.7) | 50 | (8.0)  |

PMA P170019/S016: FDA Summary of Safety and Effectiveness Data

27 of 40

{27}

|  **Geographic Region**  |   |   |   |   |   |   |   |   |
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
|  US | 20 | (19.6) | 163 | (23.7) | 19 | (20.9) | 144 | (22.9)  |
|  Non-US | 82 | (80.4) | 525 | (76.3) | 72 | (79.1) | 484 | (77.1)  |
|  **ECOG**  |   |   |   |   |   |   |   |   |
|  [0] Normal Activity | 42 | (41.2) | 277 | (40.3) | 37 | (40.7) | 254 | (40.4)  |
|  [1] Symptoms, but ambulatory | 59 | (57.8) | 409 | (59.4) | 53 | (58.2) | 372 | (59.2)  |
|  [2] Ambulatory but unable to work | 1 | (1.0) | 2 | (0.3) | 1 | (1.1) | 2 | (0.3)  |
|  **Metastatic Staging**  |   |   |   |   |   |   |   |   |
|  M0 | 9 | (8.8) | 72 | (10.5) | 7 | (7.7) | 65 | (10.4)  |
|  M1 | 93 | (91.2) | 616 | (89.5) | 84 | (92.3) | 563 | (89.6)  |
|  **Brain Metastases Present**  |   |   |   |   |   |   |   |   |
|  Yes | 6 | (5.9) | 17 | (2.5) | 3 | (3.3) | 16 | (2.5)  |
|  No | 96 | (94.1) | 671 | (97.5) | 88 | (96.7) | 612 | (97.5)  |
|  **Number of Prior Lines of Therapy**  |   |   |   |   |   |   |   |   |
|  0^{†} | 1 | (1.0) | 23 | (3.3) | 1 | (1.1) | 21 | (3.3)  |
|  Adjuvant/Neoadjuvant/ Definitive^{‡} | 0 | (0.0) | 8 | (1.2) | 0 | (0.0) | 8 | (1.3)  |
|  1 | 44 | (43.1) | 257 | (37.4) | 41 | (45.1) | 235 | (37.4)  |
|  2 | 38 | (37.3) | 187 | (27.2) | 30 | (33.0) | 167 | (26.6)  |
|  3 | 6 | (5.9) | 107 | (15.6) | 6 | (6.6) | 96 | (15.3)  |
|  4 | 7 | (6.9) | 59 | (8.6) | 7 | (7.7) | 55 | (8.8)  |
|  5 or more | 6 | (5.9) | 47 | (6.8) | 6 | (6.6) | 46 | (7.3)  |
|  **Sum of Target Lesions Measurable at Baseline (mm)**  |   |   |   |   |   |   |   |   |
|  Subjects with data | 102 |  | 684 |  | 91 |  | 624 |   |
|  Mean | 106.3 |  | 107.9 |  | 103.5 |  | 108.0 |   |
|  SD | 79.7 |  | 80.9 |  | 78.9 |  | 81.0 |   |
|  Median | 88.3 |  | 83.4 |  | 84.2 |  | 86.0 |   |
|  Range | 10.2 to 322.8 |  | 10.1 to 442.9 |  | 10.2 to 322.8 |  | 10.1 to 442.9 |   |
|  **Tumor Type**  |   |   |   |   |   |   |   |   |
|  ANAL | 14 | (13.7) | 75 | (10.9) | 14 | (15.4) | 73 | (11.6)  |
|  CERVICAL | 16 | (15.7) | 59 | (8.6) | 15 | (16.5) | 52 | (8.3)  |
|  CHOLANGIOCARCINOMA | 0 | (0.0) | 63 | (9.2) | 0 | (0.0) | 55 | (8.8)  |
|  ENDOMETRIAL | 15 | (14.7) | 67 | (9.7) | 15 | (16.5) | 64 | (10.2)  |
|  MESOTHELIOMA | 1 | (1.0) | 84 | (12.2) | 1 | (1.1) | 80 | (12.7)  |
|  NEUROENDOCRINE | 5 | (4.9) | 82 | (11.9) | 5 | (5.5) | 73 | (11.6)  |
|  SALIVARY | 3 | (2.9) | 79 | (11.5) | 3 | (3.3) | 74 | (11.8)  |
|  SMALL CELL LUNG | 34 | (33.3) | 42 | (6.1) | 26 | (28.6) | 30 | (4.8)  |
|  THYROID | 2 | (2.0) | 78 | (11.3) | 2 | (2.2) | 75 | (11.9)  |
|  VULVAR | 12 | (11.8) | 59 | (8.6) | 10 | (11.0) | 52 | (8.3)  |
|  **Prior Radiation Therapy**  |   |   |   |   |   |   |   |   |

PMA P170019/S016: FDA Summary of Safety and Effectiveness Data

28 of 40

{28}

|  Yes | 81 | (79.4) | 412 | (59.9) | 70 | (76.9) | 376 | (59.9)  |
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
|  No | 21 | (20.6) | 276 | (40.1) | 21 | (23.1) | 252 | (40.1)  |
|  **PD-L1 Status**  |   |   |   |   |   |   |   |   |
|  Positive | 68 | (66.7) | 383 | (55.7) | 61 | (67.0) | 352 | (56.1)  |
|  Negative | 29 | (28.4) | 274 | (39.8) | 25 | (27.5) | 250 | (39.8)  |
|  Not Evaluable | 5 | (4.9) | 30 | (4.4) | 5 | (5.5) | 25 | (4.0)  |
|  Missing | 0 | (0.0) | 1 | (0.1) | 0 | (0.0) | 1 | (0.2)  |
|  **MSI-H Status**  |   |   |   |   |   |   |   |   |
|  MSI-High | 14 | (13.7) | 0 | (0.0) | 14 | (15.4) | 0 | (0.0)  |
|  non-MSI-High | 81 | (79.4) | 672 | (97.7) | 73 | (80.2) | 614 | (97.8)  |
|  Missing | 7 | (6.9) | 16 | (2.3) | 4 | (4.4) | 14 | (2.2)  |
|  ^{†} Participants did not receive systemic chemotherapy.  |   |   |   |   |   |   |   |   |
|  ^{‡} Participants received adjuvant/neoadjuvant alone without recurrence < 12 months from completion of the therapy or received definitive therapy alone which cannot be considered as a line of therapy.  |   |   |   |   |   |   |   |   |
|  Subjects with unknown TMB status are not included.  |   |   |   |   |   |   |   |   |
|  (Database Cutoff Date: 27JUN2019).  |   |   |   |   |   |   |   |   |

The various tumor types in cohorts A through J were observed at different frequencies in the TMB-H and non-TMB-H populations. Small cell lung carcinoma (SCLC) was more common in the TMB-H population. Thyroid, neuroendocrine, mesothelioma, and salivary tumors each were more common in the non-TMB-H population. The differences in certain tumor types are based on small numbers, as 5 of the 10 tumor types were present in 5 or fewer participants in the TMB-H population. A high percentage of participants had received prior radiation therapy, which is consistent with the higher percentage of SCLC and other tumor types likely to be treated with radiation therapy.

### C. Accountability of sPMA Cohort

At IA10 of KEYNOTE-158, a total of 1,072 participants were enrolled in the combined cohorts, A through J, of which 1,066 were treated with at least one dose of pembrolizumab (referred to as All Subjects as Treated [ASaT]) population). The safety analysis was conducted in the ASaT population of 1,066 participants, while the efficacy analysis population included 1,050 participants from the ASaT population who were enrolled 26 weeks prior to data cut-off.

From the enrolled population, 1,007 patients had samples that were accessioned at Foundation Medicine. One sample was lost in transit, resulting in a total of 1,006 sample test records.

Based on the F1CDx investigational sample specification requirements used to define the TE population, 808 subjects had a valid TMB result, and 199 subjects had samples that failed to meet the minimum test specifications: 5 participants did not pass pre-analytical QC due to insufficient samples for testing (n=1) or gender discordance (n=4); 36 did not pass pathology review based on TE specifications (i.e.,

PMA P170019/S016: FDA Summary of Safety and Effectiveness Data

29 of 40

{29}

≥ 10% tumor nuclei and ≥ 0.2 mm³ viable nucleated tissue); 73 did not meet the minimum extracted DNA criterion (> 55 ng); and 86 yielded invalid results after sequencing and reporting. Among the 808 participants with a valid TMB result, 15 were enrolled < 26 weeks before the data cut-off for IA10, two were not treated, and one did not have a TMB score; as a result, the TE population consists of 790 patients with TMB data available. Of the 790 participants in the TE population with available TMB results, 102 were TMB-H (≥ 10 mut/Mb) and 688 were non-TMB-H (< 10 mut/Mb).

Based on the F1CDx final sample specification requirements used to define the DV population, 735 subjects had a valid TMB result, and 272 subjects had samples that failed to meet the minimum test specifications: 5 participants did not pass pre-analytical QC due to insufficient samples for testing (n=1) or gender discordance (n=4); 156 did not pass pathology review based on DV specifications (i.e., ≥ 20% tumor nuclei and ≥ 0.6 mm³ of viable nucleated tissue); 40 did not meet the minimum extracted DNA criterion (> 55 ng); and 71 yielded invalid results after sequencing and reporting. Among the 735 patients with valid TMB results, 13 were enrolled < 26 weeks before the data cut-off for IA10, two were not treated, and one did not have a TMB score; as a result, the DV population consists of 719 patients with valid TMB scores. Of the 719 participants in the DV population with available TMB results, 91 were TMB-H (≥ 10 mut/Mb) and 628 were non-TMB-H (< 10 mut/Mb).

Information on the numbers of participants per cohort based on TMB-H versus non-TMB-H status are provided for the TE and DV populations, respectively, in Table 17 below.

**Table 17. F1CDx TMB status availability by KEYNOTE-158 cohort**

|  Cohort | TE Population |   | DV Population  |   |
| --- | --- | --- | --- | --- |
|   |  TMB Status Available | TMB Status Not Available | TMB Status Available | TMB Status Not Available  |
|  A: Anal | 89 | 23 | 87 | 25  |
|  B: Biliary | 63 | 41 | 55 | 49  |
|  C: Neuroendocrine | 87 | 20 | 78 | 29  |
|  D: Endometrial | 82 | 25 | 79 | 28  |
|  E: Cervical | 75 | 23 | 67 | 31  |
|  F: Vulvar | 71 | 14 | 62 | 23  |
|  G: SCLC | 76 | 31 | 56 | 51  |
|  H: Mesothelioma | 85 | 33 | 81 | 37  |
|  I: Thyroid | 80 | 23 | 77 | 26  |
|  J: Salivary Gland | 82 | 27 | 77 | 32  |
|  **Total** | **790** | **260** | **719** | **331**  |

#### **D. Safety and Effectiveness**

##### **1. Safety Results**

The safety with respect to treatment with pembrolizumab was addressed during review of the sBLA and is not addressed in detail in this Summary of Safety and

PMA P170019/S016: FDA Summary of Safety and Effectiveness Data

30 of 40

{30}

Effectiveness Data. The safety profile of pembrolizumab in the TMB-H Safety Dataset is generally consistent with the established safety profile of pembrolizumab monotherapy. No new safety signals were identified in KEYNOTE-158. Please refer to Drugs@FDA for complete safety information on KEYTRUDA® (pembrolizumab).

## 2. Effectiveness Results

The effectiveness of F1CDx to identify TMB-H (≥ 10 mut/Mb) patients who may benefit from pembrolizumab treatment is supported by the efficacy results from IA10 of KEYNOTE-158. The efficacy results from KEYNOTE-158 show that pembrolizumab monotherapy provides a clinically meaningful benefit to participants with TMB-H advanced solid tumors that had progressed following prior treatment or who were intolerant to prior therapies. This was demonstrated by clinically meaningful ORR and DOR in the TE and DV populations.

The primary endpoint in KEYNOTE-158 was ORR, defined as the proportion of patients in the analysis population who had a response (CR or PR) as measured by central review (BIRC) per RECIST 1.1. For the pooled analysis across tumor types in cohorts A through J, response to pembrolizumab was enriched in TMB-H subjects, resulting in a clinically meaningful ORR of 29.4% in the TE population (Table 18) and 33% in the DV population (Table 19). By contrast, a lower ORR was observed in non-TMB-H subjects of 6.3% in the TE population and 6.5% in the DV population.

Table 18. Summary of best objective response based on RECIST 1.1 per central radiology assessment in TE population

|  Response Evaluation | TMB >=10 mut/Mb (N=102) |   |   | TMB <10 mut/Mb (N=688)  |   |   |
| --- | --- | --- | --- | --- | --- | --- |
|   |  n | % | 95% CI† | n | % | 95% CI†  |
|  Complete Response (CR) | 4 | 3.9 | (1.1, 9.7) | 11 | 1.6 | (0.8, 2.8)  |
|  Partial Response (PR) | 26 | 25.5 | (17.4, 35.1) | 32 | 4.7 | (3.2, 6.5)  |
|  Objective Response (CR+PR) | 30 | 29.4 | (20.8, 39.3) | 43 | 6.3 | (4.6, 8.3)  |
|  Stable Disease (SD) | 14 | 13.7 | (7.7, 22.0) | 227 | 33.0 | (29.5, 36.6)  |
|  Non-CR/Non-PD (NN) | 0 | 0.0 | (0.0, 3.6) | 3 | 0.4 | (0.1, 1.3)  |
|  Progressive Disease (PD) | 48 | 47.1 | (37.…

---

**Source:** [https://fda-staging.innolitics.com/device/P170019S016](https://fda-staging.innolitics.com/device/P170019S016)

**Published by [Innolitics](https://innolitics.com)** — a medical-device software consultancy. We help companies design, build, and clear FDA-regulated software and AI/ML devices. If you're preparing [a PMA](https://innolitics.com/services/regulatory/), [a 510(k)](https://innolitics.com/services/510ks/), [a SaMD](https://innolitics.com/services/end-to-end-samd/), [an AI/ML medical device](https://innolitics.com/services/medical-imaging-ai-development/), or [an FDA regulatory strategy](https://innolitics.com/services/regulatory/), [get in touch](https://innolitics.com/contact).

**Cite:** Innolitics at https://innolitics.com