Signatera™ CDx

{0} # **SUMMARY OF SAFETY AND EFFECTIVENESS DATA (SSED)** # **I. GENERAL INFORMATION** Device Generic Name: Circulating Tumor DNA, Molecular Residual Disease, Somatic Variant Detection System, Solid Tumors Device Trade Name: Signatera™ CDx Device Procode: PQP Applicant's Name and Address: Natera, Inc. 201 Industrial Road, Suite 410 San Carlos, CA 94070 Date(s) of Panel Recommendation: None Premarket Approval Application (PMA) Number: P260004 Date of FDA Notice of Approval: May 15, 2026 Breakthrough Device: Granted breakthrough device status on December 6, 2022. # **II. INDICATIONS FOR USE** Signatera™ CDx is a personalized, tumor-informed, multiplex-PCR and next-generation sequencing (NGS)-based assay that detects circulating tumor DNA (ctDNA) molecular residual disease (MRD) from plasma of peripheral whole blood collected in Streck Cell-Free DNA Blood Collection Tubes (BCTs), using bespoke assays designed to track somatic variants identified from sequencing of a patient's formalin-fixed, paraffin-embedded (FFPE) tumor specimen. Signatera™ CDx is intended to identify patients with ctDNA MRD positive status who may benefit from treatment listed in Table 1, in accordance with the approved therapeutic product labeling. Table 1: Signatera™ CDx Indicated Use and Associated Therapy | Biomarker | Indication | Therapy | | --- | --- | --- | | ctDNA MRD | Muscle Invasive Bladder Cancer (MIBC) | TECENTRIQ® (atezolizumab), TECENTRIQ HYBREZA® (atezolizumab and hyaluronidase-tqjs) | # **III. CONTRAINDICATIONS** Testing cannot be performed on patients who: PMA P260004: FDA Summary of Safety and Effectiveness Data Page 1 {1} 1. Have a history of allogenic bone marrow transplant. # IV. WARNINGS AND PRECAUTIONS 1. Only Blood Collection Tubes (BCTs) listed must be used. Do not freeze the contents of the BCTs. Do not transfer blood collected with other anticoagulants into the Signatera CDx BCTs, as this may inhibit the performance of the assay. Heparin collection tubes should not be used for blood collection since heparin can interfere with assay results. Do not reuse the BCTs. 2. Treat all samples as biohazardous, using appropriate handling and disposal procedures in accordance with local safety regulations. 3. This assay requires both tissue and blood samples. 4. These instructions are intended only for trained medical professionals. 5. Use only recommended materials, equipment, and procedures specified in the IFU. 6. Always wear appropriate personal protective equipment (PPE) such as gloves, lab coats, and eye protection. # LIMITATIONS 1. For in vitro diagnostic use. 2. For prescription use only. 3. Signatera CDx should be performed at least 6 weeks after cystectomy in muscle invasive bladder cancer (MIBC) patients. 4. A negative result does not preclude the presence of ctDNA MRD. Patients with negative results should continue with serial testing until a positive result or completion of the recommended 12-month testing window. 5. Signatera CDx is intended to detect ctDNA MRD from MIBC. 6. The assay has decreased ability to detect ctDNA MRD in samples with low levels of ctDNA (below 0.1 mean tumor molecules (MTM)/mL). 7. Signatera CDx is not intended to be used for standalone diagnostic purposes. 8. Signatera CDx does not report germline variants and does not infer hereditary cancer risk. 9. Decisions on patient care and treatment should be based on the independent medical judgment of the treating physician, taking into account all clinicopathological factors, in accordance with the standard of care. 10. The potential impact on the assay results of excessive anticoagulant from the Streck Cell-Free DNA BCTs or underfilling of the Streck Cell-Free DNA BCTs with less than 10 mL of blood has not been evaluated. 11. Incomplete or overmixing of the Streck Cell-Free DNA BCTs impacts assay performance. 12. The assay is intended to be performed on specific serial number-controlled instruments at Natera, Inc. clinical laboratories. PMA P260004: FDA Summary of Safety and Effectiveness Data Page 2 {2} # V. DEVICE DESCRIPTION Signatera CDx is a personalized, tumor-informed, multiplex-PCR and next-generation sequencing (NGS)-based assay using tumor-specific single nucleotide variants (SNVs) identified in a patient's formalin-fixed paraffin-embedded (FFPE) tumor tissue and detected in plasma isolated from anticoagulated peripheral whole blood from patients previously diagnosed with muscle invasive bladder cancer (MIBC). Signatera CDx is intended to be used as an in vitro diagnostic device to detect circulating tumor DNA (ctDNA) molecular residual disease (MRD) after curative-intent treatment (cystectomy) in patients with resectable MIBC to aid physician decision-making of adjuvant treatment in conjunction with other clinicopathological factors. The device includes two workflows: (1) tumor variant identification and selection using whole exome sequencing (WES) of a FFPE tumor tissue sample and patient-matched whole blood sample to identify and select 16 bespoke tumor-specific SNVs, and (2) tumor variant detection using multiplex PCR and ultra-deep next-generation sequencing for detection of the tumor-associated SNVs in a patient's plasma-derived cell-free DNA (cfDNA) for determination of ctDNA MRD status. Serial testing with the plasma workflow can be performed starting at least 6 weeks after cystectomy, every 6 weeks for 9 months with a final test at one year. A patient's plasma sample is considered ctDNA MRD-positive when at least two (2) SNVs out of 16 are detected. Otherwise, the sample is considered ctDNA MRD-negative. Tumor variant identification and selection is performed once on FFPE tissue samples, and tumor variant detection in the patient's plasma is performed at multiple longitudinal time points. Briefly, DNA isolated from FFPE tumor tissue and matched whole blood collected in K2EDTA blood collection tubes (BCTs) are processed for tumor variant identification using WES. Through the tumor variant identification and selection process, tumor-specific somatic variants are identified following elimination of mutations from the germline and clonal hematopoiesis of indeterminate potential (CHIP) by subtraction of the patient-matched whole blood-derived genomic sequence. From these candidate variants, a panel of 16 bespoke tumor SNVs is selected based primarily on tumor variant allele frequency (VAF) as a proxy for clonality, predicted noise profile in the plasma, and multiplex PCR compatibility between primers. A patient-specific 16-plex primer set is manufactured to be used to detect ctDNA MRD in cfDNA isolated from the plasma component of whole blood samples collected in Streck BCTs. # Assay Output Signatera CDx is a qualitative assay that reports “Signatera CDx Positive” when at least two (2) of the 16 bespoke tumor SNVs are detected and reports “Signatera CDx Negative” when fewer than two (2) bespoke tumor SNVs are detected. # Instruments PMA P260004: FDA Summary of Safety and Effectiveness Data Page 3 {3} Signatera CDx testing is performed with serial number-controlled instruments as indicated below in Table 2. All instruments are qualified by Natera, Inc. Table 2. Signatera CDx Instrument List (Instruments Qualified by Natera) | Instrument | Manufacturer | Function | Workflow Component (WES or Plasma) | | --- | --- | --- | --- | | NovaSeq 6000 | Illumina | WES library sequencing platform | WES | | NovaSeq 6000 | Illumina | Plasma library sequencing platform | Plasma | | Biomek i7 Automated Workstation | Beckman Coulter | WES Library Preparation | WES | | Veriti 96-well thermal cycler | ThermoFisher | Library Preparation | Plasma | | QIAsymphony SP | QIAGEN | Whole Blood gDNA Extraction and cfDNA extraction | WES and Plasma | ### Reagents All assay reagents included in the Signatera CDx are purchased from Natera qualified vendors, manufactured and/or qualified by Natera, Inc. Reagents, materials, and equipment needed to perform the test are used exclusively in Natera clinical laboratories. The reagents required to perform the assay are listed below. The following reagents are manufactured by Natera: - WES Workflow - Positive run controls used in patient-matched whole blood sequencing - Oligonucleotide adapters and primers - Plasma Workflow - Process and run controls - Library preparation ligation buffer - Oligonucleotide primers The following reagents are purchased from qualified vendors and are qualified by Natera under its Quality System: - WES Workflow - DNA extraction reagents - DNA purification reagents - DNA quantitation dye - Library preparation enzyme mix and buffers - Library enrichment probes PMA P260004: FDA Summary of Safety and Effectiveness Data Page 4 {4} ○ Positive and negative run control used in FFPE tissue sequencing ○ Negative run control used in patient-matched whole blood sequencing ○ Next generation sequencing kit ● Plasma Workflow ○ cfDNA extraction reagents ○ cfDNA purification reagents ○ cfDNA quantitation dye ○ DNA purification reagents ○ Library preparation enzyme mix and buffers ○ Library enrichment reagents ○ No Template Control ○ Next generation sequencing kit The materials required for specimen collection, but not provided, include five slide boxes, one Becton Dickinson (BD) IVD K2EDTA BCT (#367863), and two Streck IVD cell-free DNA BCTs (#230471). ### Assay Process #### A. Signatera CDx WES Workflow The WES workflow is performed at Natera, Inc. clinical laboratory located in CA, USA. #### Specimen Collection and Preparation The tumor tissue samples are shipped at ambient temperature as FFPE blocks (with a surface area >25 mm²) or FFPE slides (10-20 slides with 5μm thickness or 5-10 slides with 10μm thickness). FFPE tumor tissue blocks are sectioned into 6 slides of 10μm thickness. 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 confirm tumor content is sufficient (≥ 20% nucleated tumor cells) to proceed with the assay. Whole blood from the same patient is collected in one K2EDTA BCT (6 mL) and used for germline/CHIP subtraction. #### DNA Extraction Tumor DNA is extracted from pathology-confirmed, macro-dissected (if applicable), unstained FFPE tissue slides. DNA extraction from tissue is performed on Biomek liquid handlers using Omega Bio-tek Mag-Bind® FFPE DNA/RNA Kit qualified by Natera. The patient-matched whole blood DNA is extracted from peripheral blood, using an automated workflow on the QIAsymphony platform with QIAsymphony qualified by Natera. #### DNA Purification & Quantification PMA P260004: FDA Summary of Safety and Effectiveness Data Page 5 {5} Each tissue DNA and patient-matched whole blood (matched normal) genomic DNA sample is quantified using a fluorescence-based assay detecting double-stranded DNA following the manufacturer's instructions. Samples yielding ≥50 ng of gDNA mass for the tissue and ≥100 ng for patient-matched whole blood are used in library preparation for the WES workflow. ## Library Construction The normalized tissue DNA and matched normal gDNA are sheared and subjected to end-repair, A-tailing, and adapter ligation. The ligated products are then purified, PCR amplified using barcoded primers and further purified to obtain high-quality libraries. Both tissue and matched normal libraries are quantified using a fluorescence-based assay detecting double-stranded DNA. Samples meeting quality control requirements proceed to hybrid capture. ## Hybrid Capture Hybrid capture is performed using amplified libraries from tissue and matched normal samples. Libraries from the same sample type are combined, purified, and eluted in a hybridization mix containing target-specific capture probes. Following hybridization, captured libraries are purified, enriched through PCR amplification, and subjected to additional purification steps to obtain high-quality sequencing pools. The enriched pools are then quantified, normalized to appropriate loading concentration, and denatured prior to sequencing. ## Next Generation Sequencing Libraries are sequenced on the NovaSeq 6000 platform, qualified by Natera, at 2x150bp to achieve the on-target coverage, as specified in the assay protocol. ## Sequence Analysis Raw sequencing data are converted to FASTQ format and reads are mapped to the hg19 reference genome resulting in two .bam files, one for the tumor sample and one for the patient-matched blood. The mapped sequencing reads go through a quality control (QC) process to flag samples that did not meet pre-specified QC metrics. Reads are aligned to the hg19 human reference genome. Sample contamination and/or sample swaps are detected by assessing genotype concordance between tumor and matched normal samples. Somatic SNVs are identified by comparing tumor and patient-matched whole blood data, excluding known population polymorphisms, variants in repetitive regions and mutations arising from CHIP. High-confidence somatic variants are prioritized primarily based on VAF and background error rate modeling. Top-ranked variants are selected for custom assay design. Primer pairs targeting tumor-specific variants are generated using automated primer design algorithms, evaluated for specificity and primer-primer interaction potential, and optimized to produce a final set of non-overlapping, high-quality amplicons for downstream analysis. PMA P260004: FDA Summary of Safety and Effectiveness Data Page 6 {6} WES workflow sample-level sequencing control metrics are summarized in Table 3 below. Table 3. Sample-Level Sequencing Control Metrics for WES Workflow | Metric | Purpose / Threshold | | --- | --- | | Data Quality | Ensures the sequencing run generates sufficient high-quality sequence reads / ≥ 80% Reads with Base Q30. | | NGS Coverage | Ensures adequate read depth within a run*** | | Contamination and sample-swap QC | Ensures sample contamination and sample-swaps are identified*** | | Designability | Ensures a 16-plex primer set can be designed / Ability to design a 16-plex assay (identification of at least 16 somatic variants meeting all criteria to be included in the plasma panel) | *** Pre-specified threshold values. Data not shown. ### B. Signatera CDx Plasma Workflow The plasma workflow is performed at Natera, Inc. clinical laboratory located in TX, USA. ### Specimen Collection and Preparation Whole blood samples are collected in two 10 mL peripheral whole blood collection tubes (Streck Cell-Free BCT) and shipped at ambient temperature. ### Plasma Isolation Upon receipt at the testing laboratory, blood samples are accessioned, centrifuged, and plasma is isolated using an automated process within a defined time frame following collection. ### DNA Extraction Plasma samples (10 mL) are processed for cfDNA extraction using an automated workflow, and the purified cfDNA is eluted into a DNA suspension buffer. ### DNA Purification & Quantification Each cfDNA sample is purified using a magnetic bead–based cleanup method and quantified with a fluorescence-based assay detecting double-stranded DNA. Samples yielding ≥10 ng of cfDNA are used in library preparation. ### Library Construction PMA P260004: FDA Summary of Safety and Effectiveness Data Page 7 {7} cfDNA meeting minimum input requirements is used as input into library preparation. The cfDNA is end-repaired, A-tailed, and ligated with custom adapters. The purified ligation product is amplified and purified prior to further use. ## Multiplex PCR An aliquot of each cfDNA library is used as input for a multiplex PCR reaction targeting tumor-specific variants. Amplified libraries are labeled with sample-specific index barcodes and pooled with appropriate positive and negative controls and further purified to enrich for target fragments, then quantified using a fluorescence-based assay detecting double-stranded DNA. Prior to sequencing, the pooled libraries are normalized to appropriate loading concentration and denatured. ## Next Generation Sequencing Paired-end sequencing is performed using the Illumina NovaSeq 6000, qualified by Natera. ## Sequence Analysis Raw data files are converted to FASTQ format and demultiplexed based on sample-specific index barcodes. Paired reads are merged, and low-quality bases or mismatches between read pairs are filtered to minimize sequencing errors. The processed reads are aligned to the hg19 reference genome. Mapped reads undergo QC to remove off-target or low-confidence sequences, and high-quality reads are assigned to their corresponding amplicons and samples. Only on-target reads with high mapping Q30 reads are included for further analysis. For each of the 16 SNV targets, the variant is considered detected if supporting variant reads are above the pre-specified confidence threshold. The sample-level sequencing control metrics are documented in Table 4. Table 4. Sample Level Sequencing Control Metrics for Plasma Workflow | Metric | Purpose / Threshold | | --- | --- | | Total Filtered Target Number of Reads | Ensures adequate number of sequencing reads are obtained for each sample*** | | Median Amplicon Depth of Reads | Ensures adequate read depth across all 16-plex panel targets for reliable variant detection*** | | Working Targets | Ensures that a minimum number of 16-plex panel targets are successfully amplified and sequenced / Fail below 11 | PMA P260004: FDA Summary of Safety and Effectiveness Data Page 8 {8} | Median Sequencing Error Rates | Ensures sequencing quality by measuring the frequency of transition and transversion errors*** | | --- | --- | | Contamination and sample-swap QC | Ensures sample contamination and sample-swaps are identified*** | *** Pre-specified threshold values. Data not shown. ### C. Report Generation A patient’s plasma sample is considered ctDNA MRD-positive when at least two (2) SNVs out of 16 are detected. Otherwise, the sample is considered ctDNA MRD-negative. Signatera CDx is a qualitative test, reporting the presence or absence of ctDNA MRD, as “Signatera CDx-Positive” or “Signatera CDx-Negative”. Samples failing any QC metric are automatically held and no report is released. A patient sample that has failed a QC metric may be rerun by using stored intermediate products. A “Test Not Performed” (TNP) result is issued for samples unable to pass QC metrics. ### D. Internal Process Controls #### WES workflow Positive Controls: A positive control sample is included for both the tissue and patient-matched normal whole blood (MN) workflows. The tissue control contains a FFPE reference standard and is introduced at the gDNA extraction stage. The MN control contains a genomic DNA reference standard and is added during sample normalization. Quality control metrics for the positive controls are evaluated at the batch level to ensure consistency and performance across runs. Negative Controls: A no-template control is included in both the tissue and matched normal workflows at the sample normalization stage to monitor for potential contamination at the batch level. Negative control samples are processed alongside clinical samples through library preparation and quantification steps. Batches in which negative controls meet predefined quality controls are cleared to proceed to hybridization and subsequent sequencing analysis. #### Plasma workflow The workflow includes batch-level run controls and sample-level endogenous controls to ensure run and sample analysis integrity. Each sample batch includes a well-characterized reference cell line as a ctDNA MRD-negative sample and a well-characterized cell line blend as ctDNA MRD-positive samples, one blended at a high VAF and one at a low VAF. A no-template control is included to assess run-to-run or carryover contamination. Batches in which positive and negative cell-line controls and no-template control meet predefined quality controls proceed to subsequent sample-level analysis. PMA P260004: FDA Summary of Safety and Effectiveness Data Page 9 {9} ### E. Software System The Signatera CDx software system consists of data analysis platforms and supporting laboratory management systems that integrate end-to-end laboratory and analytical workflows. It connects laboratory operations, data management, and computational analysis through a centralized information management platform and cloud-based services, ensuring standardized, traceable, and compliant processing of laboratory and analytical data. ### VI. ALTERNATIVE PRACTICES AND PROCEDURES There are no FDA-approved or -cleared companion diagnostic (CDx) alternatives for the detection of ctDNA MRD in this intended use population. ### VII. MARKETING HISTORY Signatera CDx 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 test results, and subsequently, inappropriate assignment of treatment with TECENTRIQ (atezolizumab) or TECENTRIQ HYBREZA (atezolizumab and hyaluronidase-tqjs). Patients with false positive results may receive treatment with TECENTRIQ (atezolizumab) or TECENTRIQ HYBREZA (atezolizumab and hyaluronidase-tqjs) without clinical benefit and may experience adverse reactions and exposure to potential toxicity associated with the therapy. Patients with false negative results may not be considered for treatment with TECENTRIQ (atezolizumab) or TECENTRIQ HYBREZA (atezolizumab and hyaluronidase-tqjs). There is also a risk of delayed test results, which may lead to delays of treatment. For the specific adverse events related to the approved therapeutics, please see the FDA approved package insert for TECENTRIQ and TECENTRIQ HYBREZA which is available at Drugs@FDA. ### IX. SUMMARY OF NONCLINICAL STUDIES #### A. Device Cutoff Determination ##### 1. Sample-level Device Cutoff A patient's plasma sample is considered ctDNA MRD-positive when at least two (2) SNVs out of 16 patient-specific SNVs are detected. Otherwise, the sample is considered ctDNA MRD-negative. This device cutoff was set through an in-silico analysis that balanced specificity and sensitivity, while prioritizing specificity during initial or longitudinal testing. This in-silico PMA P260004: FDA Summary of Safety and Effectiveness Data Page 10 {10} analysis modeled a varying number of SNVs detectable in plasma and the detection of two SNVs yielded 99.7% specificity and 90% sensitivity at a sample-level VAF of 0.01%. Consequently, the device cutoff for a positive ctDNA MRD result was set to the threshold of at least two (2) SNVs detected out of 16 patient-specific SNVs. ## 2. Plasma Workflow Variant-level Calling Threshold Signatera CDx variant-level calling confidence thresholds of the plasma workflow were established using 1,368 ctDNA MRD-negative clinical samples from various cancer types. A proprietary background error model was employed to evaluate and select variant-specific calling confidence thresholds by calculating false positive rates for each base substitution class. The calling thresholds were chosen to achieve ≥99.5% sample-level specificity with a minimum of two (2) positive variants required for a sample-level positive call. ### B. Laboratory Studies Signatera CDx performance characteristics were established using clinical samples from patients with MIBC. The clinical samples consisted of FFPE tumor tissue, patient-matched whole blood, and plasma extracted from peripheral whole blood. Studies included patient-specific SNVs representative of the intended use population with various genomic contexts across the device's reportable range. Due to limitations in clinical sample availability and the requirement for three different specimen types (i.e., FFPE tumor tissue, whole blood and plasma) from each patient to perform the assay, contrived samples were utilized in some non-clinical studies. These contrived samples were comprised of MIBC tumor-derived fragmented FFPE DNA spiked into healthy donor plasma samples. A contrived sample functional characterization (CSFC) study was conducted to demonstrate comparable performance between contrived samples and clinical cfDNA samples prior to their use in nonclinical studies. Since Signatera CDx includes both upfront tumor tissue characterization through WES to design a patient-specific 16-plex primer set and plasma cfDNA sequencing to detect ctDNA MRD, several studies were performed to assess the performance of each workflow component in addition to end-to-end workflow assessments. #### 1. Analytical Accuracy / Concordance with an Orthogonal Method Analytical accuracy of Signatera CDx was determined through concordance with an externally validated orthogonal assay. The study used 66 sample sets, each consisting of MIBC tumor tissue, patient-matched whole blood, and plasma from the same patient. The specimens were split into two equal amounts to be tested on both assays for assessment of positive and negative agreement. The positive percent agreement (PPA) and negative percent agreement (NPA) values were evaluated and presented in Table 5. Table 5. Sample-level Agreement between Signatera CDx and the Comparator Assay PMA P260004: FDA Summary of Safety and Effectiveness Data Page 11 {11} | Signatera CDx | Comparator | | Total | | --- | --- | --- | --- | | | Positive | Negative | | | Positive | 31 | 6^{1} | 37 | | Negative | 2 | 13 | 15 | | No Call | 9 | 5 | 14 | | Total | 42 | 24 | 66 | | Point Estimate (n/N) [95% CI] (Excluding No Calls) | PPA = 93.9% (31/33) [80.4%, 98.3%], NPA = 68.4% (13/19) [46.0%, 84.6%], | | | | Point Estimate (n/N) [95% CI] (Including No Calls as Discordance) | PPA = 73.8% (31/42) [58.9%, 84.7%], NPA = 54.2% (13/24) [35.1%, 72.1%], | | | CI: Wilson Score Confidence Interval $^{1}$ Upon retesting to determine whether the discordance was repeatable, two initially comparator negative samples returned positive calls when rerun with the comparator assay. Discordance results may be attributed to variability in calls around the LoD of the assays. There are two (2) samples with a positive result by the comparator assay and a negative result by Signatera CDx, and six (6) samples with a negative result by the comparator assay and a positive result by Signatera CDx. These discordant samples all have low ctDNA levels. Variant-level agreement analyses were performed to calculate the overlap of the 16-plex variants between Signatera CDx and the orthogonal assay, and the concordance (PPA and NPA) for these overlapping variants. For the 52 samples with valid ctDNA MRD results from both assays, 53.5% (445/832) of the 16-plex variants overlapped between the two assays, with a median of 9 overlapping targets across samples (Table 6). Out of the 445 overlapping 16-plex variants, 45 variants yielded no result on the orthogonal assay. The point estimates for PPA and NPA for the 400 overlapping 16-plex variants with a valid result on the orthogonal assay were 88.8% (143/161) and 87% (208/239), respectively (Table 7). Lower variant-level agreement is expected relative to patient-level MRD determination, as the test is designed to detect the overall presence of ctDNA rather than any single specific mutation. **Table 6. Overlap of Designed 16-plex Between Signatera CDx and the Comparator Assay** | | Min | 1^{st} Quantile | Median | 3^{rd} Quantile | Max | Mean | | --- | --- | --- | --- | --- | --- | --- | | Overlap of 16-plex | 4 | 7 | 9 | 10 | 12 | 8.6 | **Table 7. Variant-level Agreement between Signatera CDx and the Comparator Assay** PMA P260004: FDA Summary of Safety and Effectiveness Data Page 12 {12} | Signatera CDx | Comparator | | | Total | | --- | --- | --- | --- | --- | | | Positive | Negative | No Result | | | **Positive** | 143 | 30 | 29 | 202 | | **Negative** | 15 | 208 | 15 | 238 | | **No Result** | 3 | 1 | 1 | 5 | | **Total** | 161 | 239 | 45 | 445 | | **Point Estimate (n/N) [95% CI]** | **PPA = 88.8% (143/161) [83.0%, 92.8%]** **NPA = 87.0% (208/239) [82.2%, 90.7%]** | | | | A post market study is planned (listed in Conditions of Approval in Section XV below) to supplement the analytical accuracy of Signatera CDx in low positive clinical specimens. ## 2. Analytical Sensitivity ### a. Limit of Blank (LoB) Studies #### WES Workflow LoB The limit of blank for SNV calls in non-cancerous bladder tissues was evaluated using a tissue replicate pairing approach, in which each non-cancerous bladder FFPE tissue replicate was analyzed against sequencing data from the germline DNA replicate from the same individual acting as the matched whole blood comparator within the standard WES workflow. Across 180 WES pipeline runs (30 comparisons per sample across six samples), zero (0/180) runs generated a 16-plex panel, demonstrating high specificity for 16-plex assay design. The average number of variants identified as potential 16-plex targets was 1.7, corresponding to a WES variant specificity of >99.99% across the 45 MB region assessed. An additional study was conducted to evaluate the risk of detecting false variants by the Signatera CDx WES workflow in MIBC tumor tissues, using a cohort of 391 clinical MIBC FFPE tissue specimens from the intended use population. A validated orthogonal WES assay was performed to characterize the somatic SNVs in these specimens. The Signatera CDx WES workflow identified a total of 138,868 somatic SNVs, of which 122,621 (88.3%) were also detected by the orthogonal WES assay. Out of the 16,247 SNVs detected exclusively by Signatera CDx WES workflow, 176 SNVs were selected as 16-plex targets, representing 0.13% of the 138,868 SNVs detected by Signatera CDx and 2.8% of the 6,256 (391×16) Signatera CDx 16-plex targets generated for these patients. #### Plasma Workflow LoB PMA P260004: FDA Summary of Safety and Effectiveness Data Page 13 {13} The LoB of the plasma workflow was established by evaluating whole blood samples from healthy (cancer-free) donors across two sets of operators, reagent lots, and instruments. Bespoke 16-plex primer sets were generated from 103 unique clinical MIBC FFPE tumor tissue and patient-matched whole blood sample pairs. Three (3) different cfDNA samples were derived from pooled plasma samples from a unique set of healthy donors. A total of 69 healthy donor samples were utilized in this study. Each of the 103 primer pools was tested with the three (3) cfDNA samples for a total of 309 independent assessments. Samples were tested using 45-66 ng cfDNA input, at the challenging, high end of the cfDNA input range for Signatera CDx. All combinations were called ctDNA MRD-negative by Signatera CDx, resulting in a sample-level false positive rate of 0% (0/309). Seven (7) positive SNV target calls were identified in the target-level analysis, for a target-level false positive rate of 0.14% (7/4928). # **b. Limit of Detection (LoD) Studies** # **WES Workflow LoD** The WES LoD was established using four (4) clinical MIBC FFPE tumor tissue and patient-matched whole blood sample pairs diluted to six target levels using gDNA from non-cancerous bladder FFPE tissue samples. The study processed seven (7) replicates per sample at each dilution level for each of three (3) reagent lots, for a total of 21 replicates per sample per dilution level, run at the assay's lowest DNA input of 50 ng. A representative set of SNVs were selected across the four unique samples, taking consideration of variant prevalence, genomic context and base substitution class. For each pre-selected somatic SNV, the LoD was estimated using probit regression analysis. If more than one dilution level had a detection rate < 10% or > 90% on either end of the dilution series, a probit model could not be fit and the LoD for that SNV was established as the lowest dilution level with ≥ 95% empirical detection rate. The summary of LoD estimates for each SNV and SNV category for the WES workflow is presented in **Table 8**. The range of SNV-level LoD estimates was 3.8% - 11.3% VAF, with a median of 5.3% VAF. **Table 8. Summary Results of WES LoD Establishment for Each SNV and SNV Category** | SNV Category | Target ID | LoD Estimates per SNV | Range of LoD Estimate per SNV Category | | --- | --- | --- | --- | | Commonly Targeted MIBC Genes^{1} | chr4_1804642_C_G (Gene FGFR3) | 5.9% VAF | 3.9-9.0% VAF | | | chr4_1806089_G_T (Gene FGFR3) | 5.0% VAF | | PMA P260004: FDA Summary of Safety and Effectiveness Data Page 14 {14} | SNV Category | Target ID | LoD Estimates per SNV | Range of LoD Estimate per SNV Category | | --- | --- | --- | --- | | | chr17_7579716_G_A (Gene TP53) | 9.0% VAF | | | | chr4_1803568_C_G (Gene FGFR3) | 3.9% VAF | | | Selected into 16-plex | chr7_138602109_G_C | 5.3% VAF | 4.5-6.9% VAF | | | chr2_21225180_G_C | 5.7% VAF | | | | chr13_30351259_A_T | 4.5% VAF | | | | chr1_183085681_C_G | 6.9% VAF | | | Difficult to Sequence^{2} | chr6_5086204_C_G | 4.8% VAF | 4.8-6.7% VAF | | | chr2_42396803_C_A | 5.2% VAF | | | | chr12_53454502_C_T | 6.7% VAF | | | | chr2_170493815_G_T | 5.3% VAF | | | Not Difficult to Sequence | chr3_39374376_C_G | 3.8% VAF | 3.8-7.4% VAF | | | chr2_201473694_G_A | 5.2% VAF | | | | chr7_20826418_C_T | 7.4% VAF | | | | chr7_138602109_G_C | 5.3% VAF | | | GC Rich (> 65% GC Content) | chr6_5086204_C_G | 4.8% VAF | 4.8-7.7% VAF | | | chr2_42396803_C_A | 5.2% VAF | | | | chr12_53454502_C_T | 6.7% VAF | | PMA P260004: FDA Summary of Safety and Effectiveness Data Page 15 {15} | SNV Category | | Target ID | LoD Estimates per SNV | Range of LoD Estimate per SNV Category | | --- | --- | --- | --- | --- | | | | chr3_48629139_C_T | 7.7% VAF | | | GC Depleted (< 25% GC Content) | | chr2_26068393_C_A | 11.3% VAF | 4.6-11.3% VAF | | | | chr5_38972013_A_G | 5.3% VAF | | | | | chr2_175245466_C_G | 4.6% VAF | | | | | chr12_81751874_C_A | 5.9% VAF | | | High Coverage^{3} | | chr3_39374376_C_G | 3.8% VAF | 3.8-7.3 % VAF | | | | chr1_111781401_G_C | 5.4% VAF | | | | | chr7_98988657_G_C | 7.3% VAF | | | | | chr14_69077745_C_T | 5.0% VAF | | | Low Coverage^{3} | | chr1_53333244_C_T | 5.2% VAF | 5.2-5.7 % VAF | | | | chr11_103092850_C_G | 5.4% VAF | | | | | chr2_21225180_G_C | 5.7% VAF | | | | | chr10_122649395_C_G | 5.2% VAF | | | Base Substitution Class | A>C / T>G | chr16_27246550_A_C | 5.4% VAF | | | | A>G / T>C | chr13_43181019_T_C | 5.7% VAF | | | | A>T / T>A | chr4_55573411_A_T | 5.8% VAF | | | | C>A / G>T | chr2_42396803_C_A | 5.2% VAF | | PMA P260004: FDA Summary of Safety and Effectiveness Data Page 16 {16} | SNV Category | | Target ID | LoD Estimates per SNV | Range of LoD Estimate per SNV Category | | --- | --- | --- | --- | --- | | | C>G / G>C | chr3_39374376_C_G | 3.8% VAF | | | | C>T / G>A | chr2_201473694_G_A | 5.2% VAF | | | **Overall Assay LoD Estimate** | | | 3.8-11.3 % VAF | | ¹ SNV targets for the IMvigor011 study population most commonly occurred on five genes: TP53, FGFR3, KDM6A, DMD, or HUWE1. However, because the 16-plex are highly specific for each patient, SNVs within these genes only represented 0.3% of all potential 16-plex targets among the study subjects in the clinical validation study. No eligible SNVs from KDM6A, DMD and HUWE1 were found in the samples for this study. ² SNVs identified in the difficult to sequence whole genome sequencing (WGS) panel generated by the consortium Genome in a Bottle (NIST, Genome in a Bottle Project, 2024). ³ High Coverage: ≥ 75th percentile of tumor total Depth of Read (DOR) per sample; Low Coverage: ≤ 25th percentile of tumor total DOR per sample LoD of the WES workflow was confirmed in the WES precision study using 10 undiluted clinical MIBC samples representative of the intended use population at the minimum DNA input (i.e., 50 ng tissue DNA, 100 ng matched normal DNA). The samples were tested in triplicate across three (3) operators and three (3) instruments/reagent lot combinations, for a total of 27 replicates per sample. For each SNV, the mean observed VAF was calculated as the average of its VAF values across the samples that passed QC. For each SNV with mean VAF in the 1-1.5x LoD range (5.3-7.95% VAF), a SNV-specific hit rate was calculated and then aggregated into a single overall SNV hit rate estimate. The LoD for the Signatera CDx WES workflow was confirmed with an overall hit rate of 97.74% (95% CI: [97.58%, 97.89%]) across a total of 1323 unique SNVs (refer to Precision of WES Workflow and Table 20). ### Plasma Workflow LoD Establishment LoD of the Signatera CDx plasma workflow was established as sample-level mean VAF of the 16-plex targets, which is correlated with the number of targets detected in each sample. The study evaluated ten (10) cfDNA MRD-positive, MIBC clinical samples representative of the intended use population and covering different SNV categories. Clinical cfDNA samples were diluted with cfDNA from healthy donor plasma to six target sample-level mean VAF levels (0.066, 0.033, 0.017, 0.008, 0.004, and 0.002%, equivalent to 0.2, 0.1, 0.05, 0.025, 0.013, and 0.0063 mean tumor molecules (MTM)/mL, respectively). Each sample was tested at the minimum cfDNA input mass of 10 ng with the associated bespoke 16-plex primer sets designed from that patient's matched tumor tissue and PMA P260004: FDA Summary of Safety and Effectiveness Data Page 17 {17} whole blood samples. The LoD estimate based on probit models was 0.0214% VAF. An LoD of 0.043% sample-level mean VAF was conservatively established based on a consistent $\geq 95\%$ hit rate across reagent lots, which is equivalent to 0.13 MTM/mL at cfDNA input of 10ng extracted from 10mL of plasma (Table 9). **Table 9. Summary of Plasma Workflow LoD Establishment Results** | Intended VAF % | | 0.066 | 0.033 | 0.017 | 0.008 | 0.004 | 0.002 | 0 | | --- | --- | --- | --- | --- | --- | --- | --- | --- | | Intended MTM/mL^{1} | | 0.2 | 0.1 | 0.05 | 0.025 | 0.013 | 0.0063 | 0 | | **Mean Observed VAF %** | Lot 1 | 0.0717 | 0.0383 | 0.0168 | 0.008 | 0.0028 | 0.0018 | 0 | | | Lot 2 | 0.0801 | 0.0426 | 0.0164 | 0.0078 | 0.0029 | 0.0004 | 0 | | | Lot 3 | 0.0769 | 0.0388 | 0.0169 | 0.0087 | 0.0019 | 0.0019 | 0 | | **Mean Observed MTM/mL** | Lot 1 | 0.2174 | 0.116 | 0.051 | 0.0243 | 0.0084 | 0.0054 | 0 | | | Lot 2 | 0.2426 | 0.1292 | 0.0497 | 0.0238 | 0.0088 | 0.0011 | 0 | | | Lot 3 | 0.2329 | 0.1176 | 0.0512 | 0.0264 | 0.0057 | 0.0057 | 0 | | **Hit Rate; (n/N)** | Lot 1 | 100.0% (20/20) | 100.0% (20/20) | 85.0% (17/20) | 65.0% (13/20) | 25.0% (5/20) | 20.0% (4/20) | 0.0% (0/20) | | | Lot 2 | 100.0% (20/20) | 100.0% (20/20) | 80.0% (16/20) | 55.0% (11/20) | 30.0% (6/20) | 5.0% (1/20) | 0.0% (0/20) | | | Lot 3 | 100.0% (20/20) | 100.0% (20/20) | 85.0% (17/20) | 65.0% (13/20) | 20.0% (4/20) | 20.0% (4/20) | 0.0% (0/20) | | **Mean # Positive Targets** | Lot 1 | 9.20 | 6.50 | 3.50 | 2.20 | 0.90 | 0.90 | 0 | | | Lot 2 | 10.05 | 6.50 | 3.40 | 1.70 | 1.00 | 0.55 | 0.1 | | | Lot 3 | 10.45 | 6.70 | 3.60 | 2.15 | 1.00 | 0.65 | 0.15 | $^{1}$MTM/mL = (sample-level mean VAF x cfDNA mass (ng) x 1000 (pg/ng)) / (3.3 (pg) x plasma volume (mL)) Exploratory variant level analysis supports that the assay's sensitivity of detecting individual variants is around 0.33% VAF. A post market study is planned (listed in Conditions of Approval in Section XV below) to supplement the LoD of Signatera CDx plasma workflow for individual variants. PMA P260004: FDA Summary of Safety and Effectiveness Data Page 18 {18} # **Plasma Workflow LoD Confirmation** Signatera CDx plasma workflow LoD was confirmed at the established LoD using three ctDNA MRD-positive clinical MIBC samples representative of the intended use population tested at levels spanning the assay's cfDNA input range of 10 - 66 ng. Twenty replicates were tested per sample at 0.1 MTM/ml at each cfDNA input level, corresponding to a sample-level mean VAF of 0.033% at 10 ng DNA input, 0.010% at 33 ng DNA input and 0.005% at 66 ng DNA input, using a single reagent lot. All three cfDNA input levels tested achieved ≥ 95% hit rates, confirming the assay LoD at different DNA input levels (Table 10). Table 10. Summary of Plasma Workflow LoD Confirmation Results at Different Input Levels | cfDNA input (ng) | Intended VAF (%) | Mean Observed VAF % | Hit Rate (n/N) | 95% CI | Equivalent MTM/mL | Mean # Positive Targets | | --- | --- | --- | --- | --- | --- | --- | | 10 | 0.033 | 0.0442% | 100% (60/60) | [94.0%, 100%] | 0.1340 | 6.48 | | 33 | 0.010 | 0.0122% | 96.6% (57/59) | [88.5%, 99.1%] | 0.1224 | 5.31 | | 66 | 0.005 | 0.0057% | 96.7% (58/60) | [88.6%, 99.1%] | 0.1141 | 4.63 | ### 3. Analytical Specificity #### a. Interfering Substances The impact of potentially interfering substances on the performance of Signatera CDx was evaluated using four (4) ctDNA MRD-positive clinical MIBC cfDNA samples diluted with healthy donor plasma to 2-3 x LoD. Samples were processed at the minimum cfDNA input of 10 ng, with 3 replicates each per condition. Interference was evaluated separately for each condition based on sample-level PPA. The number of detected positive targets from the 16-plex and sample-level mean VAF were also evaluated. All nine potentially interfering substances demonstrated PPA of 100% (95% CI: [75.75%, 100%]), and there were no significant differences in the number of detected positive targets or sample-level mean VAF values between conditions (Table 11). Table 11. Summary Results for Potentially Interfering Substances PMA P260004: FDA Summary of Safety and Effectiveness Data Page 19 {19} | Interferent | Test Concentration | PPA (%) (n/N) | 95% CI | Mean # Positive Targets | Mean Sample-level Mean VAF | | --- | --- | --- | --- | --- | --- | | Nominal | N/A | N/A | N/A | 10.9 | 0.093 | | Albumin | 60 g/L | 100 (12/12) | [75.75%, 100%] | 10.8 | 0.096 | | Bilirubin Conjugated | 475 μmol/L (40 mg/dL) | 100 (12/12) | [75.75%, 100%] | 11.2 | 0.106 | | Bilirubin Unconjugated | 684 μmol/L (40 mg/dL) | 100 (12/12) | [75.75%, 100%] | 10.8 | 0.085 | | Ethanol | 6.00E+02 mg/dL (1.30E+05 μmol/L) | 100 (12/12) | [75.75%, 100%] | 10.3 | 0.091 | | Hemoglobin | 10 g/L (1000 mg/dL) | 100 (12/12) | [75.75%, 100%] | 11.2 | 0.090 | | Index Barcodes | Standard +30% | 100 (12/12) | [75.75%, 100%] | 10.2 | 0.096 | | Proteinase K | 2x Standard Amount | 100 (12/12) | [75.75%, 100%] | 10.2 | 0.093 | | Staphylococcus epidermidis | 1x10^6 CFU/mL | 100 (12/12) | [75.75%, 100%] | 10.1 | 0.088 | | Triglycerides | 16.94 mol/L (1,500 mg/dL) | 100 (12/12) | [75.75%, 100%] | 10.5 | 0.087 | ### b. In silico Specificity Analysis #### Hybrid Capture Bait Specificity: Hybrid capture bait specificity was assessed using sequencing data from WES processing of 77 clinical MIBC FFPE tumor tissue and patient-matched whole blood samples. For the reads in regions where a variant was called, the reads that also mapped to a probe region were measured and reported as “on-variant, on-probe” reads. The proportion of high-quality reads that were on-variant, on-probe reads was calculated to evaluate hybrid capture probe specificity. All 77 tumor samples produced valid outputs. The average proportion of high-quality reads that were on-variant, on-probe among all high-quality reads was > 99.99%. #### Multiplex PCR Primer Specificity: Primer specificity was assessed using plasma sequencing data from the full workflow testing of 77 clinical MIBC patient samples representative of the intended use population. In this study, BAM files from plasma samples were processed to extract on-variant and off-variant reads. The PMA P260004: FDA Summary of Safety and Effectiveness Data Page 20 {20} proportion of on-variant reads was calculated to evaluate multiplex PCR primer pool specificity. All 77 plasma BAM files completed analysis with no QC or computational failures. The mean proportion of high-quality plasma reads aligning to the intended amplicon regions was >99%. #### 4. Contrived Sample Functional Characterization (CSFC) Comparable performance between clinical cfDNA samples and contrived samples was confirmed by demonstrating physical and functional equivalency between the contrived and clinical samples. Ten unique clinical specimens from patients with MIBC were used to create contrived cfDNA-like clinical specimens. This contrived material was serially diluted with cfDNA extracted from plasma of healthy donors to target six levels consistent with the LoD establishment study design and tested at 10 ng DNA input using three reagent lots, each lot with two replicates per sample per dilution level for a total of 360 data points. Physical equivalency was evaluated through electropherogram profiles of clinical and contrived DNA fragment size distributions. The analytical sensitivity in contrived samples was established by fitting a probit regression model. The contrived sample results were compared to the LoD establishment results for clinical cfDNA samples at the lowest assay cfDNA input mass of 10 ng to demonstrate functional equivalency. For each sample type, the regression model coefficients for intercept and slope, and the predicted estimates for C5, C25, C50, C75, and C95 were calculated. Electropherogram profiles of clinical and contrived DNA showed similar peak and size distribution demonstrating physical equivalency. The probit regression yielded closely aligned models for clinical and contrived samples across the range of tested VAF levels. None of the calculated absolute differences between regression model estimates for clinical and contrived samples were statistically significant, supporting comparability between the clinical and contrived samples. #### 5. Precision Studies The precision of Signatera CDx was evaluated through three different precision studies encompassing the end-to-end device workflow, as well as the WES workflow and the plasma workflow, respectively, as detailed below. ##### a. Precision of Signatera CDx End-to-End Workflow The end-to-end precision of the entire Signatera CDx workflow was demonstrated using clinical patient sample trios of tumor FFPE tissue, whole blood, and plasma. For each patient, the tumor tissue and whole blood pair was run in triplicate using three reagent lot combinations and three sequencing runs, yielding three 16-plex primer designs. At least one WES replicate was run at the minimum DNA input. Each primer design PMA P260004: FDA Summary of Safety and Effectiveness Data Page 21 {21} was then used in the plasma workflow with a corresponding plasma sample that was run in triplicate using three reagent lot combinations and three sequencing runs. All plasma sample replicates were run at the 10ng minimum cfDNA input level. A total of 37 unique sample trios were evaluated in this study. Precision of the WES workflow was assessed for 35 unique samples. Two samples were excluded due to QC failures leaving only one evaluable replicate. A total of 25,014 unique variants were identified by the WES workflow, with an overall concordance of 91.6% (95% CI: [91.4%, 91.8%]). The number of unique variants identified in each sample ranged from 43 to 2,332 with the variant positive call rate aggregated within a sample ranging from 54.5% - 96.8%. The specimen with the lowest variant positive call rate had 14 targets consistently overlapping across replicates, supporting that the SNVs with low positive call rates in this specimen were not prioritized for selection as 16-plex targets. Across replicates, the average number of 16-plex target overlap ranged from 9 - 15.3 (Table 12). **Table 12. Summary of Precision Results per Sample—WES Workflow Output** | Specimen | # Valid WES Outputs | Total Unique Variants Detected Across WES Replicates | Variant Positive Call Rate (n/N) [two-sided 95% CI] | Average Overlapping 16-plex [Range] | | --- | --- | --- | --- | --- | | 1 | 2 | 1,185 | 88.9% (2,106/2,370) [87.5%, 90.1%] | 13.0 [13, 13] | | 2 | 3 | 627 | 83.4% (1,568/1,881) [81.6%, 85.0%] | 14.3 [14, 15] | | 3 | 2 | 1,393 | 96.7% (2,694/2,786) [96.0%, 97.3%] | 13.0 [13, 13] | | 5 | 3 | 570 | 90.7% (1,551/1,710) [89.2%, 92.0%] | 11.3 [11, 12] | | 7 | 3 | 276 | 91.8% (760/828) [89.7%, 93.5%] | 10.3 [9, 12] | | 8 | 3 | 196 | 93.2% (548/588) [90.9%, 95.0%] | 9.7 [9, 11] | | 9 | 3 | 446 | 54.5% (729/1,338) [51.8%, 57.1%] | 14.0 [14, 14] | | 10 | 3 | 260 | 87.4% (682/780) [84.9%, 89.6%] | 15.0 [15, 15] | | 11 | 3 | 197 | 87.5% (517/591) [84.6%, 89.9%] | 13.3 [12, 15] | | 12 | 3 | 959 | 73.8% (2,124/2,877) [72.2%, 75.4%] | 12.3 [11, 13] | PMA P260004: FDA Summary of Safety and Effectiveness Data Page 22 {22} | Specimen | # Valid WES Outputs | Total Unique Variants Detected Across WES Replicates | Variant Positive Call Rate (n/N) [two-sided 95% CI] | Average Overlapping 16-plex [Range] | | --- | --- | --- | --- | --- | | 13 | 3 | 2,332 | 84.7% (5,923/6,996) [83.8%, 85.5%] | 13.3 [12, 15] | | 14 | 3 | 303 | 74.9% (681/909) [72.0%, 77.6%] | 15.3 [15, 16] | | 15 | 3 | 1,688 | 96.6% (4,890/5,064) [96.0%, 97.0%] | 10.3 [9, 12] | | 16 | 3 | 418 | 74.2% (931/1,254) [71.8%, 76.6%] | 13.3 [12, 15] | | 17 | 3 | 198 | 66.8% (397/594) [63.0%, 70.5%] | 14.7 [14, 15] | | 18 | 3 | 694 | 89.9% (1,872/2,082) [88.5%, 91.1%] | 15.0 [15, 15] | | 19 | 2 | 638 | 88.9% (1,135/1,276) [87.1%, 90.6%] | 12.0 [12, 12] | | 20 | 3 | 278 | 85.6% (714/834) [83.1%, 87.8%] | 12.3 [12, 13] | | 21 | 3 | 584 | 74.7% (1,309/1,752) [72.6%, 76.7%] | 13.0 [12, 14] | | 22 | 3 | 317 | 93.1% (885/951) [91.3%, 94.5%] | 11.0 [10, 12] | | 23 | 3 | 192 | 92.9% (535/576) [90.5%, 94.7%] | 14.3 [14, 15] | | 24 | 3 | 214 | 84.6% (543/642) [81.6%, 87.2%] | 11.0 [10, 12] | | 25 | 3 | 427 | 64.6% (828/1,281) [62.0%, 67.2%] | 13.0 [13, 13] | | 26 | 3 | 694 | 96.2% (2,002/2,082) [95.2%, 96.9%] | 14.3 [14, 15] | | 27 | 3 | 937 | 93.0% (2,615/2,811) [92.0%, 93.9%] | 9.0 [8, 10] | | 28 | 3 | 714 | 94.2% (2,018/2,142) [93.1%, 95.1%] | 14.3 [14, 15] | | 29 | 3 | 1,481 | 91.4% (4,063/4,443) [90.6%, 92.2%] | 12.0 [12, 12] | | 30 | 3 | 43 | 74.4% (96/129) [66.3%, 81.2%] | 14.7 [14, 15] | PMA P260004: FDA Summary of Safety and Effectiveness Data Page 23 {23} | Specimen | # Valid WES Outputs | Total Unique Variants Detected Across WES Replicates | Variant Positive Call Rate (n/N) [two-sided 95% CI] | Average Overlapping 16-plex [Range] | | --- | --- | --- | --- | --- | | 31 | 3 | 660 | 91.8% (1,817/1,980) [90.5%, 92.9%] | 10.3 [10, 11] | | 32 | 3 | 599 | 86.8% (1,560/1,797) [85.2%, 88.3%] | 13.7 [13, 14] | | 33 | 3 | 1,570 | 96.8% (4,557/4,710) [96.2%, 97.2%] | 13.7 [13, 14] | | 34 | 3 | 699 | 95.9% (2,011/2,097) [95.0%, 96.7%] | 13.3 [12, 14] | | 35 | 2 | 740 | 92.4% (1,368/1,480) [91.0%, 93.7%] | 10.0 [10, 10] | | 36 | 3 | 1,357 | 60.3% (2,454/4,071) [58.8%, 61.8%] | 11.0 [10, 12] | | 37 | 3 | 1,128 | 85.1% (2,879/3,384) [83.8%, 86.2%] | 11.7 [11, 13] | Thirty-four (34) sample trios (254 replicates) produced sufficient WES and Plasma results and were included in the analysis of ctDNA MRD concordance. Three samples were excluded due to failed QC. Sample-level ctDNA MRD concordance was 96.9% (246/254), and the overall variant-level concordance was 91.6% (3,671/4,007) (Table 13). Lower variant-level precision is expected relative to patient-level MRD determination due to stochastic sampling variability, as the test is designed to detect the overall presence of ctDNA rather than any single specific mutation. Five WES and plasma reagent lot sets were included in the study with each sample evaluated using three unique lot sets. Sample-level within-lot precision ranges from 95.7% to 100% for WES workflow reagents and 97.2% to 100% for plasma workflow reagents (Table 14). **Table 13. Summary of End-to-End Workflow Precision—Plasma Workflow Output** | Samples | # of samples | Total # of replicates | Sample-Level Call Concordance (n/N) [95% CI] | Variant-Level Call Concordance (n/N) [95% CI] | | --- | --- | --- | --- | --- | | All | 34 | 254 | 96.9% (246/254) [93.9%, 98.4%] | 91.6% (3671/4007) [90.7%, 92.4%] | PMA P260004: FDA Summary of Safety and Effectiveness Data Page 24 {24} | **Positive above LoD** | 14 | 94 | 100% (94/94) [96.1%, 100%] | 89.1% (1323/1485) [87.4%, 90.6%] | | --- | --- | --- | --- | --- | | **Positive below LoD** | 5 | 40 | 90.0% (36/40) [76.9%, 96.0%] | 76.0% (479/630) [72.5%, 79.2%] | | **Negative with 1 Positive Target** | 2 | 18 | 83.3% (15/18) [60.8%, 94.2%] | 96.8% (276/285) [94.1%, 98.3%] | | **Negative with 0 Positive Targets** | 13 | 102 | 99.0% (101/102) [94.7%, 99.8%] | 99.1% (1593/1607) [98.5%, 99.5%] | **Table 14. Summary of Precision Results by Reagent Lot** | Reagent Lot | Within-Lot Precision | | | --- | --- | --- | | | % (n/N) [95% CI] | % Per Specimen^{1} (Min, Max) | | WES Lot 1 | 100% (34/34) [89.8%, 100%] | (100%, 100%) | | WES Lot 2 | 100% (33/33) [89.6%, 100%] | (100%, 100%) | | WES Lot 3 | 96.4% (81/84) [90.0%, 98.8%] | (50%, 100%) | | WES Lot 4 | 96.0% (48/50) [86.5%, 98.9%] | (67%, 100%) | | WES Lot 5 | 95.7% (45/47) [85.8%, 98.8%] | (67%, 100%) | | Plasma Lot 1 | 100% (42/42) [91.6%, 100%] | (100%, 100%) | | Plasma Lot 2 | 97.7% (85/87) [92.0%, 99.4%] | (50%, 100%) | | Plasma Lot 3 | 100% (46/46) [92.3%, 100%] | (100%, 100%) | | Plasma Lot 4 | 97.2% (35/36) [85.8%, 99.5%] | (67%, 100%) | | Plasma Lot 5 | 97.2% (35/36) [85.8%, 99.5%] | (67%, 100%) | $^{1}$ Each lot was tested across a minimum of 12 specimens that yielded 2-3 evaluable plasma replicates. The majority (30/34) of samples were 100% ctDNA MRD concordant across replicates. For the remaining four samples which were primarily PMA P260004: FDA Summary of Safety and Effectiveness Data Page 25 {25} challenging samples below LoD or borderline negative with one positive target detected on average, the concordance ranges from 66.7% - 88.9%. Variant-level concordance aggregated within a sample ranged from 68.1% - 100%, with the lowest concordance observed in an ctDNA MRD positive sample with 100% sample-level concordance (Table 15). **Table 15. Summary of Precision Results per Sample—Plasma Workflow Output** | Specimen | Valid Plasma Outputs | Modal Call Status^{1} | Sample-level Call Concordance (n/N) [95% CI] | # Positive Targets | | Sample-level Mean VAF (%) | | Variant-level Call Concordance (n/N) [95% CI] | | --- | --- | --- | --- | --- | --- | --- | --- | --- | | | | | | Mean (SD) | CV % | Mean (SD) | CV % | | | 1 | 6/6 | Positive Above LoD | 100% (6/6) [61.0%, 100%] | 15.67 (0.52) | 3.3 | 9.7599 (0.3324) | 3.41 | 100% (94/94) [96.1%, 100%] | | 2 | 9/9 | Positive Above LoD | 100% (9/9) [70.1%, 100%] | 15.89 (0.33) | 2.1 | 6.4776 (0.1573) | 2.43 | 100% (143/143) [97.4%, 100%] | | 3 | 6/6 | Positive Above LoD | 100% (6/6) [61.0%, 100%] | 16 (0) | 0 | 3.2779 (0.1372) | 4.18 | 100% (96/96) [96.2%, 100%] | | 4 | 3/3 | Positive Above LoD | 100% (3/3) [43.9%, 100%] | 16 (0) | 0 | 1.5413 (0.0397) | 2.58 | 100% (48/48) [92.6%, 100%] | | 5 | 6/9 | Positive Above LoD | 100% (6/6) [61.0%, 100%] | 15.33 (0.82) | 5.32 | 1.1013 (0.0985) | 8.94 | 100% (92/92) [96.0%, 100%] | | 6 | 3/3 | Positive Above LoD | 100% (3/3) [43.9%, 100%] | 16 (0) | 0 | 1.096 (0.1324) | 12.08 | 100% (48/48) [92.6%, 100%] | | 7 | 6/9 | Positive Above LoD | 100% (6/6) [61.0%, 100%] | 16 (0) | 0 | 0.814 (0.054) | 6.63 | 100% (96/96) [96.2%, 100%] | | 8 | 3/9 | Positive Above LoD | 100% (3/3) [43.9%, 100%] | 16 (0) | 0 | 0.4827 (0.0468) | 9.7 | 100% (37/37) [90.6%, 100%] | | 9 | 9/9 | Positive Above LoD | 100% (9/9) [70.1%, 100%] | 15.78 (0.44) | 2.79 | 0.4607 (0.0417) | 9.04 | 99.3% (142/143) [96.1%, 99.9%] | | 10 | 9/9 | Positive Above LoD | 100% (9/9) [70.1%, 100%] | 12.33 (0.5) | 4.05 | 0.0829 (0.0102) | 12.36 | 78.5% (113/144) [71.1%, 84.4%] | | 11 | 9/9 | Positive Above LoD | 100% (9/9) [70.1%, 100%] | 11.11 (1.36) | 12.28 | 0.0821 (0.0115) | 14.04 | 79.9% (115/144) [72.6%, 85.6%] | | 12 | 7/9 | Positive Above LoD | 100% (7/7) [64.6%, 100%] | 10 (1.29) | 12.91 | 0.0699 (0.0111) | 15.9 | 78.6% (88/112) [70.1%, 85.2%] | | 13 | 9/9 | Positive Above LoD | 100% (9/9) [70.1%, 100%] | 8.89 (1.96) | 22.11 | 0.0424 (0.0106) | 25.05 | 73.6% (106/144) [65.9%, 80.1%] | PMA P260004: FDA Summary of Safety and Effectiveness Data Page 26 {26} | Specimen | Valid Plasma Outputs | Modal Call Status^{1} | Sample-level Call Concordance (n/N) [95% CI] | # Positive Targets | | Sample-level Mean VAF (%) | | Variant-level Call Concordance (n/N) [95% CI] | | --- | --- | --- | --- | --- | --- | --- | --- | --- | | | | | | Mean (SD) | CV % | Mean (SD) | CV % | | | 14 | 9/9 | Positive Above LoD | 100% (9/9) [70.1%, 100%] | 6.33 (1.73) | 27.35 | 0.0389 (0.0118) | 30.47 | 72.9% (105/144) [65.1%, 79.5%] | | 15 | 9/9 | Positive Below LoD | 100% (9/9) [70.1%, 100%] | 9 (2.65) | 29.4 | 0.0335 (0.0119) | 35.68 | 68.1% (98/144) [60.1%, 75.1%] | | 16 | 9/9 | Positive Below LoD | 100% (9/9) [70.1%, 100%] | 5.89 (1.17) | 19.81 | 0.0266 (0.0044) | 16.7 | 71.5% (103/144) [63.7%, 78.3%] | | 17 | 9/9 | Positive Below LoD | 100% (9/9) [70.1%, 100%] | 4.78 (1.48) | 31.01 | 0.0172 (0.0054) | 31.08 | 75.0% (108/144) [67.3%, 81.4%] | | 18 | 9/9 | Positive Below LoD | 66.7% (6/9) [35.4%, 87.9%] | 2.67 (2.18) | 81.73 | 0.0152 (0.0117) | 77.27 | 84.4% (114/135) [77.4%, 89.6%] | | 19 | 4/6 | Positive Below LoD | 75.0% (3/4) [30.1%, 95.4%] | 1.75 (0.5) | 28.57 | 0.0063 (0.0064) | 101.13 | 88.9% (56/63) [78.8%, 94.5%] | | 20 | 9/9 | Negative 1 Positive Target | 66.7% (6/9) [35.4%, 87.9%] | 1.22 (0.67) | 54.55 | 0.0025 (0.0038) | 150.52 | 97.9% (138/141) [93.9%, 99.3%] | | 21 | 9/9 | Negative 1 Positive Target | 100% (9/9) [70.1%, 100%] | 0.67 (0.5) | 75 | 0^{2} (0) | NE^{3} | 95.8% (138/144) [91.2%, 98.1%] | | 22 | 9/9 | Negative 0 Positive Targets | 100% (9/9) [70.1%, 100%] | 0.33 (0.5) | 150 | 0 (0) | NE | 97.9% (141/144) [94.1%, 99.3%] | | 23 | 9/9 | Negative 0 Positive Targets | 100% (9/9) [70.1%, 100%] | 0.33 (0.5) | 150 | 0 (0) | NE | 97.9% (141/144) [94.1%, 99.3%] | | 24 | 9/9 | Negative 0 Positive Targets | 100% (9/9) [70.1%, 100%] | 0.33 (0.5) | 150 | 0 (0) | NE | 97.9% (141/144) [94.1%, 99.3%] | | 25 | 9/9 | Negative 0 Positive Targets | 88.9% (8/9) [56.5%, 98.0%] | 0.33 (0.71) | 212.13 | 0.0005 (0.0016) | 300 | 97.8% (132/135) [93.7%, 99.2%] | | 26 | 9/9 | Negative 0 Positive Targets | 100% (9/9) [70.1%, 100%] | 0.22 (0.44) | 198.43 | 0 (0) | NE | 98.6% (142/144) [95.1%, 99.6%] | | 27 | 3/9 | Negative | 100% (3/3) [43.9%, 100%] | 0 (0) | NE | 0 (0) | NE | 100% (36/36) [90.4%, 100%] | PMA P260004: FDA Summary of Safety and Effectiveness Data Page 27 {27} | Specimen | Valid Plasma Outputs | Modal Call Status^{1} | Sample-level Call Concordance (n/N) [95% CI] | # Positive Targets | | Sample-level Mean VAF (%) | | Variant-level Call Concordance (n/N) [95% CI] | | --- | --- | --- | --- | --- | --- | --- | --- | --- | | | | | | Mean (SD) | CV % | Mean (SD) | CV % | | | | | 0 Positive Targets | | | | | | | | 28 | 9/9 | Negative 0 Positive Targets | 100% (9/9) [70.1%, 100%] | 0 (0) | NE | 0 (0) | NE | 100% (143/143) [97.4%, 100%] | | 29 | 9/9 | Negative 0 Positive Targets | 100% (9/9) [70.1%, 100%] | 0 (0) | NE | 0 (0) | NE | 100% (144/144) [97.4%, 100%] | | 30 | 6/9 | Negative 0 Positive Targets | 100% (6/6) [61.0%, 100%] | 0 (0) | NE | 0 (0) | NE | 100% (95/95) [96.1%, 100%] | | 31 | 9/9 | Negative 0 Positive Targets | 100% (9/9) [70.1%, 100%] | 0 (0) | NE | 0 (0) | NE | 100% (144/144) [97.4%, 100%] | | 32 | 6/9 | Negative 0 Positive Targets | 100% (6/6) [61.0%, 100%] | 0 (0) | NE | 0 (0) | NE | 100% (96/96) [96.2%, 100%] | | 33 | 9/9 | Negative 0 Positive Targets | 100% (9/9) [70.1%, 100%] | 0 (0) | NE | 0 (0) | NE | 100% (144/144) [97.4%, 100%] | | 34 | 6/9 | Negative 0 Positive Targets | 100% (6/6) [61.0%, 100%] | 0 (0) | NE | 0 (0) | NE | 100% (94/94) [96.1%, 100%] | $^{1}$ Modal Call Status of a specimen is the Signatera CDx results for majority of the replicates. $^{2}$ Sample-level mean VAF of ctDNA MRD-Negative results is set to 0%. $^{3}$ NE: Not estimable. ### b. Precision of WES Workflow This study evaluated the Signatera CDx WES workflow within-run and between-run / within-laboratory precision for identifying tumor-derived somatic variants under varied conditions including different reagent lots, instruments, and operators across multiple days. Ten clinical MIBC tumor FFPE samples representative of the intended use population were tested at the minimum WES DNA input (50 ng tissue DNA, 100 ng patient-matched whole blood DNA). Twenty-seven replicates were tested per sample for a total of 270 data points across three operators and three instruments/reagent lot combinations. PMA P260004: FDA Summary of Safety and Effectiveness Data Page 28 {28} For each specimen, all variants identified at least once across replicates (range: 301 -3,118) with an average VAF across replicates of at least 2.5% were evaluated. Within-run precision was defined as agreement with the within-run majority call while between-run / within-laboratory precision was defined as agreement with the majority call among all the replicates of a given SNV. The overall estimates were taken as the average across all samples and variants, resulting in a within-run precision of 98.54% and between-run / within-laboratory precision of 98.25% (Table 16). Across individual variants, the mean within-run precision is 96.3%, ranging from 66.7% to 100%, and the mean between-run / within-laboratory precision is 95.2%, ranging from 51.9% to 100% (Table 17). Within-run precision of SNV calls was also calculated for each of the nine batches. The analysis of within-run precision across instrument sets and reagent lots is presented in Table 18. **Table 16. Summary of Precision of Signatera CDx WES Workflow** | Number of Unique SNVs in Range^{1} | Within-run Precision [95% CI] | Between-run / Within-laboratory Precision [95% CI] | | --- | --- | --- | | 11,403 | 98.54% [97.47%, 99.07%] | 98.25% [96.95%, 98.90%] | $^{1}$Analysis includes all SNVs identified at least once across replicates with an average VAF above 2.5%. **Table 17. Distribution of WES Precision across Variants$^{1}$** | | Min | Q1 | Median | Mean | Q3 | Max | | --- | --- | --- | --- | --- | --- | --- | | **Within-run Precision** | 66.7 % | 100% | 100% | 96.3% | 100% | 100% | | **Between-run / Within-laboratory Precision** | 51.9% | 96.3% | 100% | 95.2% | 100% | 100% | $^{1}$Analysis includes all SNVs identified at least once across replicates with an average VAF above 2.5%. **Table 18. Precision of SNV Agreement by Instrument Set and Reagent Lot$^{1}$** | Reagent | Precision [95% CI] | | | | | --- | --- | --- | --- | --- | | | Instrument Set 1 | Instrument Set 2 | Instrument Set 3 | Overall | | **Reagent Lot 1** | 98.59% [97.53%, 99.11%] | 98.45% [97.25%, 99.06%] | 98.73% [97.80%, 99.24%] | 98.59% [97.53%, 99.13%] | PMA P260004: FDA Summary of Safety and Effectiveness Data Page 29 {29} | **Reagent Lot 2** | 98.43% [97.07%, 99.02%] | 98.63% [97.72%, 99.09%] | 98.29% [96.99%, 98.91%] | 98.45% [97.27%, 99.00%] | | --- | --- | --- | --- | --- | | **Reagent Lot 3** | 98.47% [97.42%, 99.03%] | 98.62% [97.58%, 99.10%] | 98.64% [97.77%, 99.14%] | 98.58% [97.60%, 99.08%] | | **Overall** | 98.50% [97.35%, 99.05%] | 98.57% [97.52%, 99.08%] | 98.55% [97.53%, 99.09%] | 98.54% [97.47%, 99.07%] | $^{1}$Analysis includes all SNVs identified at least once across replicates with an average VAF above 2.5%. Variant level positive call rates were calculated as the detection rate in the unique variants across all sample replicates, which ranged from 90.5% to 99.3% across the samples. Among the 27 16-plex assays per specimen, an average of 9.6-14.3 targets overlapped between replicate pairs. The highest rate of target overlap was observed in the specimen with the lowest overall SNV positive call rate, as the SNVs with low positive call rates in this specimen were not prioritized for selection as 16-plex targets (Table 19). Table 20 summarizes the positive call rates stratified by SNV category and VAF. **Table 19. Summary Results of WES Precision Per Sample** | Specimen | # Replicates | Total Unique SNVs^{1} | Variant Positive Call Rate (n/N) [95% CI] | Average Overlapping 16-plex [range] | | --- | --- | --- | --- | --- | | 1 | 27 | 446 | 98.6% (11,870/12,042) [98.3%, 98.8%] | 13.7 [12, 16] | | 2 | 27 | 334 | 98.6% (8,893/9,018) [98.4%, 98.8%] | 12.7 [10, 15] | | 3 | 27 | 303 | 94.9% (7,763/8,181) [94.4%, 95.3%] | 13.4 [10, 15] | | 4 | 27 | 485 | 93.5% (12,246/13,095) [93.1%, 93.9%] | 13.9 [12, 16] | | 5 | 27 | 1711 | 99.3% (45,865/46,197) [99.2%, 99.4%] | 10.9 [7, 14] | PMA P260004: FDA Summary of Safety and Effectiveness Data Page 30 {30} | 6 | 27 | 3118 | 98.0% (82,544/84,186) [98.0%, 98.1%] | 9.6 [7, 13] | | --- | --- | --- | --- | --- | | 7 | 27 | 301 | 90.5% (7,351/8,127) [89.8%, 91.1%] | 14.3 [12, 16] | | 8 | 27 | 1031 | 99.1% (27,574/27,837) [98.9%, 99.2%] | 12.9 [10, 16] | | 9 | 27 | 2567 | 99.2% (68,735/69,309) [99.1%, 99.2%] | 12.3 [10, 16] | | 10 | 27 | 1107 | 98.1% (29,315/29,889) [97.9%, 98.2%] | 14.0 [12, 16] | $^{1}$Analysis includes all SNVs identified at least once across replicates with an average VAF above 2.5%. **Table 20. Summary Results of WES Precision by SNV Category and VAF Levels** | SNV Category | VAF % Bin (Range)^{4} | # Unique SNVs | Mean VAF % | Variant Positive Call Rate (n/N) [95% CI] | | --- | --- | --- | --- | --- | | Overall | [2.5, 5.3) | 937 | 3.7233 | 85.28% (21575/25299) [84.84%, 85.71%] | | | [5.3, 7.95) | 1323 | 6.9540 | 97.74% (34913/35721) [97.58%, 97.89%] | | | [7.95, 11.925) | 2239 | 9.8830 | 99.23% (59989/60453) [99.16%, 99.30%] | | | [11.925, 17.8875) | 2663 | 14.6386 | 99.46% (71514/71901) [99.41%, 99.51%] | | | [17.8875, 100) | 4241 | 30.8147 | 99.70% (114158/114507) [99.66%, 99.73%] | | Commonly Targeted MIBC Genes^{1} | [2.5, 5.3) | 0 | NE | NE | | | [5.3, 7.95) | 1 | 6.3528 | 96.30% (26/27) [81.72%, 99.34%] | | | [7.95, 11.925) | 0 | NE | NE | | | [11.925, 17.8875) | 2 | 12.7047 | 100% (54/54) [93.36%, 100%] | | | [17.8875, 100) | 19 | 48.4895 | 99.22% (509/513) [98.01%, 99.70%] | | Difficult to Sequence^{2} | [2.5, 5.3) | 319 | 3.7646 | 84.99% (7320/8613) [84.22%, 85.73%] | | | [5.3, 7.95) | 418 | 6.8842 | 97.34% (10986/11286) [97.03%, 97.62%] | | | [7.95, 11.925) | 677 | 9.9350 | 99.00% (18096/18279) [98.84%, 99.13%] | | | [11.925, 17.8875) | 883 | 14.5776 | 99.38% (23694/23841) [99.28%, 99.48%] | PMA P260004: FDA Summary of Safety and Effectiveness Data Page 31 {31} | | [17.8875, 100) | 1310 | 30.7971 | 99.62% (35235/35370) [99.55%, 99.68%] | | --- | --- | --- | --- | --- | | Not Difficult to Sequence^{2} | [2.5, 5.3) | 618 | 3.7020 | 85.43% (14255/16686) [84.89%, 85.96%] | | | [5.3, 7.95) | 905 | 6.9862 | 97.92% (23927/24435) [97.73%, 98.09%] | | | [7.95, 11.925) | 1562 | 9.8605 | 99.33% (41893/42174) [99.25%, 99.41%] | | | [11.925, 17.8875) | 1780 | 14.6688 | 99.50% (47820/48060) [99.43%, 99.56%] | | | [17.8875, 100) | 2931 | 30.8226 | 99.73% (78923/79137) [99.69%, 99.76%] | | GC Rich (> 65% GC Content) | [2.5, 5.3) | 234 | 3.7603 | 88.45% (5588/6318) [87.63%, 89.21%] | | | [5.3, 7.95) | 286 | 6.9820 | 99.42% (7677/7722) [99.22%, 99.56%] | | | [7.95, 11.925) | 464 | 9.9140 | 99.48% (12463/12528) [99.34%, 99.59%] | | | [11.925, 17.8875) | 645 | 14.6332 | 99.66% (17355/17415) [99.56%, 99.73%] | | | [17.8875, 100) | 992 | 30.7219 | 99.68% (26698/26784) [99.60%, 99.74%] | | GC Depleted (< 25% GC Content) | [2.5, 5.3) | 11 | 4.0082 | 81.82% (243/297) [77.03%, 85.79%] | | | [5.3, 7.95) | 16 | 6.6877 | 89.81% (388/432) [86.60%, 92.32%] | | | [7.95, 11.925) | 27 | 10.0193 | 98.08% (715/729) [96.80%, 98.85%] | | | [11.925, 17.8875) | 29 | 14.6474 | 99.87% (782/783) [99.28%, 99.98%] | | | [17.8875, 100) | 52 | 32.2239 | 99.15% (1392/1404) [98.51%, 99.51%] | | High Coverage^{3} | [2.5, 5.3) | 260 | 3.8131 | 92.85% (6518/7020) [92.22%, 93.43%] | | | [5.3, 7.95) | 358 | 6.9746 | 99.51% (9619/9666) [99.35%, 99.63%] | | | [7.95, 11.925) | 591 | 9.8107 | 99.71% (15910/15957) [99.61%, 99.78%] | | | [11.925, 17.8875) | 647 | 14.6451 | 99.85% (17443/17469) [99.78%, 99.90%] | | | [17.8875, 100) | 970 | 31.1293 | 99.83% (26146/26190) [99.77%, 99.87%] | | Low Coverage^{3} | [2.5, 5.3) | 235 | 3.7646 | 73.70% (4676/6345) [72.60%, 74.76%] | | | [5.3, 7.95) | 329 | 6.8764 | 93.08% (8268/8883) [92.53%, 93.59%] | | | [7.95, 11.925) | 563 | 9.9280 | 97.99% (14896/15201) [97.76%, 98.20%] | | | [11.925, 17.8875) | 667 | 14.6296 | 98.63% (17763/18009) [98.45%, 98.79%] | PMA P260004: FDA Summary of Safety and Effectiveness Data Page 32 {32} | | [17.8875, 100) | 1094 | 32.1506 | 99.49% (29386/29538) [99.40%, 99.56%] | | --- | --- | --- | --- | --- | | Base Substitution: A>C/T>G | [2.5, 5.3) | 38 | 3.6501 | 77.68% (797/1026) [75.03%, 80.12%] | | | [5.3, 7.95) | 32 | 7.0213 | 95.83% (828/864) [94.29%, 96.98%] | | | [7.95, 11.925) | 67 | 9.4166 | 98.40% (1780/1809) [97.71%, 98.88%] | | | [11.925, 17.8875) | 49 | 14.3914 | 99.77% (1320/1323) [99.34%, 99.92%] | | | [17.8875, 100) | 80 | 32.8918 | 99.81% (2156/2160) [99.52%, 99.93%] | | Base Substitution: A>G/T>C | [2.5, 5.3) | 134 | 3.8292 | 83.53% (3022/3618) [82.28%, 84.70%] | | | [5.3, 7.95) | 264 | 6.8985 | 97.04% (6917/7128) [96.62%, 97.41%] | | | [7.95, 11.925) | 284 | 9.4034 | 98.96% (7588/7668) [98.70%, 99.16%] | | | [11.925, 17.8875) | 151 | 14.6656 | 99.17% (4043/4077) [98.84%, 99.40%] | | | [17.8875, 100) | 340 | 34.1939 | 99.55% (9139/9180) [99.39%, 99.67%] | | Base substitution: A>T/T>A | [2.5, 5.3) | 45 | 3.6209 | 83.37% (1013/1215) [81.18%, 85.36%] | | | [5.3, 7.95) | 44 | 6.7378 | 94.36% (1121/1188) [92.90%, 95.53%] | | | [7.95, 11.925) | 53 | 9.8047 | 99.23% (1420/1431) [98.63%, 99.57%] | | | [11.925, 17.8875) | 69 | 14.6117 | 98.39% (1833/1863) [97.71%, 98.87%] | | | [17.8875, 100) | 110 | 31.0764 | 99.63% (2959/2970) [99.34%, 99.79%] | | Base Substitution: C>A/G>T | [2.5, 5.3) | 128 | 3.7348 | 85.91% (2969/3456) [84.71%, 87.03%] | | | [5.3, 7.95) | 168 | 6.9789 | 97.93% (4442/4536) [97.47%, 98.30%] | | | [7.95, 11.925) | 207 | 9.7307 | 99.07% (5537/5589) [98.78%, 99.29%] | | | [11.925, 17.8875) | 233 | 15.0060 | 98.92% (6223/6291) [98.63%, 99.15%] | | | [17.8875, 100) | 350 | 30.5028 | 99.41% (9394/9450) [99.23%, 99.54%] | | Base Substitution: C>G/G>C | [2.5, 5.3) | 169 | 3.7351 | 85.38% (3896/4563) [84.33%, 86.38%] | | | [5.3, 7.95) | 170 | 7.0366 | 98.08% (4502/4590) [97.64%, 98.44%] | | | [7.95, 11.925) | 545 | 10.2124 | 99.41% (14628/14715) [99.27%, 99.52%] | | | [11.925, 17.8875) | 614 | 14.4183 | 99.67% (16524/16578) [99.58%, 99.75%] | PMA P260004: FDA Summary of Safety and Effectiveness Data Page 33 {33} | | [17.8875, 100) | 840 | 29.3716 | 99.88% (22652/22680) [99.82%, 99.91%] | | --- | --- | --- | --- | --- | | Base Substitution: C>T/G>A | [2.5, 5.3) | 423 | 3.6991 | 86.49% (9878/11421) [85.85%, 87.10%] | | | [5.3, 7.95) | 645 | 6.9598 | 98.21% (17103/17415) [98.00%, 98.40%] | | | [7.95, 11.925) | 1083 | 9.9049 | 99.30% (29036/29241) [99.20%, 99.39%] | | | [11.925, 17.8875) | 1547 | 14.6770 | 99.53% (41571/41769) [99.46%, 99.59%] | | | [17.8875, 100) | 2521 | 30.8058 | 99.69% (67858/68067) [99.65%, 99.73%] | | SNVs in 16-plex | [2.5, 5.3) | 3 | 4.2448 | 80.25% (65/81) [70.30%, 87.46%] | | | [5.3, 7.95) | 2 | 6.4727 | 98.15% (53/54) [90.23%, 99.67%] | | | [7.95, 11.925) | 1 | 11.7453 | 100% (27/27) [87.54%, 100%] | | | [11.925, 17.8875) | 32 | 15.7336 | 99.65% (861/864) [98.98%, 99.88%] | | | [17.8875, 100) | 262 | 37.1043 | 99.80% (7060/7074) [99.67%, 99.88%] | $^{1}$ SNV targets for the IMvigor011 study population most commonly occurred on five genes: TP53, FGFR3, KDM6A, DMD, or HUWE1. SNVs within these genes represented 0.3% of all potential 16-plex targets among the study subjects in the clinical validation study. No eligible SNVs from KDM6A, DMD and HUWE1 were found in the samples for this study. $^{2}$ SNVs identified in the difficult to sequence whole genome sequencing (WGS) panel generated by the consortium Genome in a Bottle (NIST, Genome in a Bottle Project, 2024). $^{3}$ High Coverage: ≥ 75th percentile of tumor total DOR per sample; Low Coverage: ≤ 25th percentile of tumor total DOR per sample. $^{4}$ Analysis includes all SNVs identified at least once across replicates with an average VAF above 2.5%. ### c. Precision of Plasma Workflow The study evaluated within-run and between-run / within-laboratory precision of the Signatera CDx plasma workflow under varied conditions including different reagent lots, instruments, and operators. The precision of ctDNA MRD-positive samples was evaluated using six ctDNA MRD-positive clinical MIBC cfDNA samples, diluted using healthy donor cfDNA to 1.5x LoD. Each sample was tested using three replicates across three reagent lots, three instrument sets, and three operators, for a total of 81 replicates per sample and 486 total data points. The study generated a total of 463 data points after QC exclusions. The precision from ctDNA MRD-negative samples was evaluated using three pools of cfDNA from healthy donors tested using two replicates for each combination of variance components generated across three reagent lots, three instrument sets, and three operators, using three 16-plex primer pools from MIBC clinical samples used in the same precision study, for a total of 54 replicates per sample and 162 total data points. The study generated PMA P260004: FDA Summary of Safety and Effectiveness Data Page 34 {34} a total of 156 data points after QC exclusions. Both sample types were processed at the 10ng minimum cfDNA input. The point estimate for within-run precision was calculated as the percentage of sample-level replicates that agreed with the majority ctDNA MRD call within a sequencing run across all sample-run combinations. The point estimate for between-run / within-laboratory precision was calculated as the percentage of sample-level replicates that agreed with the majority ctDNA MRD call among all replicates of that sample across all runs. The point estimates for sample-level within-run and between-run / within-laboratory precision were both 100% overall (Table 21) and by factor (Table 22). **Table 21. Summary of Plasma Workflow Precision Study (Sample Level)** | Sample Type | Total Sample Replicates (N) | Within-run Precision (%) (n/N) [95% CI] | Between-run / Within-laboratory Precision (%) (n/N) [95% CI] | | --- | --- | --- | --- | | ctDNA MRD Positive | 463 | 100% (463/463) [99.2%, 100%] | 100% (463/463) [99.2%, 100%] | | ctDNA MRD Negative | 156 | 100% (156/156) [97.6%, 100%] | 100% (156/156) [97.6%, 100%] | **Table 22. Summary of Plasma Workflow Precision by Study Factor (Sample Level)** | Factor | Total Sample Replicates (N) | Within-run Precision (%) (n/N) | Between-run / Within-laboratory Precision (%) (n/N) | | --- | --- | --- | --- | | Reagent Lot 1 | 159 | 100% (159/159) [97.6%, 100%] | 100% (159/159) [97.6%, 100%] | | Reagent Lot 2 | 161 | 100% (161/161) [97.7%, 100%] | 100% (161/161) [97.7%, 100%] | | Reagent Lot 3 | 143 | 100% (143/143) [97.4%, 100%] | 100% (143/143) [97.4%, 100%] | | Instrument Set 1 | 161 | 100% (161/161) [97.7%, 100%] | 100% (161/161) [97.7%, 100%] | | Instrument Set 2 | 161 | 100% (161/161) [97.7%, 100%] | 100% (161/161) [97.7%, 100%] | PMA P260004: FDA Summary of Safety and Effectiveness Data Page 35 {35} | Factor | Total Sample Replicates (N) | Within-run Precision (%) (n/N) | Between-run / Within-laboratory Precision (%) (n/N) | | --- | --- | --- | --- | | Instrument Set 3 | 141 | 100% (141/141) [97.3%, 100%] | 100% (141/141) [97.3%, 100%] | | Operator 1 | 162 | 100% (162/162) [97.7%, 100%] | 100% (162/162) [97.7%, 100%] | | Operator 2 | 158 | 100% (158/158) [97.6%, 100%] | 100% (158/158) [97.6%, 100%] | | Operator 3 | 143 | 100% (143/143) [97.4%, 100%] | 100% (143/143) [97.4%, 100%] | Precision for individual samples is summarized in **Table 23**. Precision was 100% across patients' samples, with variant level concordance aggregated within a sample ranging from 62.4% - 71.3% across the positive samples and 98.4% - 99.8% across the negative samples. Across individual variants, the mean within-run precision is 79.2%, ranging from 70.1% to 100%, and the mean between-run / within-laboratory precision is 65.7%, ranging from 49.4% to 100% (**Table 24**). Variant-level precision across a range of genomic contexts demonstrated negative call rates for modal negative variants (as defined by majority call across replicates) ranging from 59.1% (C>T/G>A) to 74.3% (A>C/T>G) across variant classifications and positive call rates for modal positive variants ranging from 55.8% (C>A/G>T) to 73.9% (A>T/T>A) (**Table 25**). **Table 23. Summary of Plasma Workflow Precision Study per Sample** | Specimen | Modal Call Status | # Replicates | Sample Call Concordance (n/N) [95% CI] | # Positive Targets | | Sample Mean VAF | | Variant Call Concordance (n/N) [95% CI] | | --- | --- | --- | --- | --- | --- | --- | --- | --- | | | | | | Mean (SD) | %CV | Mean (SD) | %CV | | | 1 | Positive | 78 | 100% (78/78) [95.3, 100] | 8.0 (2.1) | 26.0 | 0.0653 (0.0223) | 34.1 | 67.4% (841/1248) [64.7%, 69.9%] | | 2 | Positive | 77 | 100% (77/77) [95.2, 100] | 7.4 (2.3) | 31.0 | 0.0603 (0.0201) | 33.3 | 62.4% (769/1232) [59.7%, 65.1%] | | 3 | Positive | 77 | 100% (77/77) [95.2, 100] | 8.3 (2.0) | 23.5 | 0.0614 (0.0183) | 29.9 | 62.4% (769/1232) [59.7%, 65.1%] | PMA P260004: FDA Summary of Safety and Effectiveness Data Page 36 {36} | Specimen | Modal Call Status | # Replicates | Sample Call Concordance (n/N) [95% CI] | # Positive Targets | | Sample Mean VAF | | Variant Call Concordance (n/N) [95% CI] | | --- | --- | --- | --- | --- | --- | --- | --- | --- | | | | | | Mean (SD) | %CV | Mean (SD) | %CV | | | 4 | Positive | 77 | 100% (77/77) [95.2, 100] | 6.4 (2.1) | 32.3 | 0.0534 (0.0196) | 36.7 | 71.3% (878/1232) [68.7%, 73.7%] | | 5 | Positive | 76 | 100% (76/76) [95.2, 100] | 7.5 (1.9) | 25.0 | 0.0463 (0.0188) | 40.5 | 64.1% (779/1216) [61.3%, 66.7%] | | 6 | Positive | 78 | 100% (78/78) [95.3, 100] | 7.6 (2.0) | 25.8 | 0.0574 (0.0181) | 31.5 | 66.7% (833/1248) [64.1%, 69.3%] | | 7 | Negative | 52 | 100% (52/52) [93.1, 100] | 0.0 (0.2) | NE | 0 (0) | NE | 99.2% (825/832) [98.3%, 99.6%] | | 8 | Negative | 52 | 100 (52/52) [93.1, 100] | 0.1 (0.3) | NE | 0 (0) | NE | 98.4% (819/832) [97.3%, 99.1%] | | 9 | Negative | 52 | 100 (52/52) [93.1, 100] | 0.0 (0.1) | NE | 0 (0) | NE | 99.8% (830/832) [99.1%, 99.9%] | **Table 24. Distribution of Plasma Precision across Variants** | | Min | Q1 | Median | Mean | Q3 | Max | | --- | --- | --- | --- | --- | --- | --- | | **Within-run Precision** | 70.1 % | 74.8% | 77.9% | 79.2% | 81.9% | 100% | | **Between-run / Within-laboratory Precision** | 49.4% | 55.8% | 61.5% | 65.7% | 72.7% | 100% | *…