Abstract
Background: Next-generation sequencing (NGS) assays are highly complex tests that can vary substantially in both their design and intended application. Despite their innumerous advantages, NGS assays present some unique challenges associated with the preanalytical process, library preparation, data analysis, and reporting. According to a number of professional laboratory organization, control materials should be included both during the analytical validation phase and in routine clinical use to guarantee highly accurate results. The SeraseqTM Solid Tumor Mutation Mix AF10 and AF20 control materials consist of 26 biosynthetic DNA constructs in a genomic DNA background, each containing a specific variant or mutation of interest and an internal quality marker at 2 distinct allelic frequencies of 10% and 20%, respectively. The goal of this interlaboratory study was to evaluate the Seraseq AF10 and AF20 control materials by verifying their performance as control materials and by evaluating their ability to measure quality metrics essential to a clinical test.
Methods: Performance characteristics were assessed within and between 6 CLIA-accredited laboratories and 1 research laboratory.
Results: Most laboratories detected all 26 mutations of interest; however, some discrepancies involving the internal quality markers were observed.
Conclusion: This interlaboratory study showed that the Seraseq AF10 and AF20 control materials have high quality, stability, and genomic complexity in variant types that are well suited for assisting in NGS assay analytical validation and monitoring routine clinical applications.
IMPACT STATEMENT
This manuscript describes a multicenter study evaluating the use of synthetic control material as a quality control (QC) indicator in next-generation sequencing (NGS) assays. It is the first such study to test the performance metrics of various NGS assays and platforms using this QC material. The reproducibility of assays using this QC material in different laboratories with different operators helps assess the quality of NGS results being reported as part of precision medicine applications in oncology. The routine use of QC materials including different mutations will help ensure more standardized performance of NGS testing for guided therapy in patients with cancer.
Precision medicine in oncology routinely uses next-generation sequencing (NGS)9 to screen patient tumor samples for somatic mutations to guide the selection of targeted therapeutics. NGS-based assays have considerably increased the ability to detect clinically actionable variants that are critical for diagnosis, prognosis, and/or making therapeutic decisions.
Clinical NGS assays are highly complex tests that can vary widely in both their design and technology. Factors contributing to the analytical variation of NGS assays include the method of library preparation (amplicon-based or hybrid capture-based); coverage of genomic regions of interest (targeted panel, whole-exome or whole-genome); instrumentation (pH-based or sequencing by synthesis chemistries); and the bioinformatic analysis pipeline (supplied by instrument vendor, open-source tools).
Despite their substantial advantages, NGS assays present some considerable challenges associated with preanalytical processes—nucleic acid extraction (DNA/RNA quantity and quality), library preparation, data analysis (pipeline development and data management), and reporting (variant classification and interpretation). These complexities and challenges require extensive analytical validation to ensure accurate and precise results for NGS assays before clinical use. To date, there are no NGS oncology assays that are approved by the U.S. Food and Drug Administration. The American College of Medical Genetics and Genomics, the Association for Molecular Pathology, and the College of American Pathologists have published consensus-based guidelines for NGS test validation (1–10).
The analytical validation process for NGS assays consists of determining the analytical sensitivity and specificity (accuracy), reproducibility, and limit of detection. In addition, analytical validations should use sample types that are identical to those that will be tested using the assay and of sufficient quantity of variants that represent each of the detectable mutation types [single nucleotide variants (SNVs), copy number variations (CNVs), insertions and deletions (INDELs), fusions, etc.] covered within the assay's targeted regions. Finally, the laboratory should establish a quality management plan that includes proficiency testing (PT), quality control (QC) checkpoints, and QC materials. Both the QC checkpoints and QC materials are crucial for monitoring preanalytical, analytical (library preparation and sequencing), and postanalytical (data analysis) steps in each run (9, 10).
Ideal QC materials should mimic the human genome by containing different mutation types with different allelic frequencies and be stable to monitor the multistep NGS workflow. Because there are no U.S. Food and Drug Administration-approved (or recommended) QC materials for clinical NGS assays, many companies offer commercially available research-use-only control materials, including genomic DNA from blood or fresh/formalin-fixed paraffin-embedded (FFPE) cell lines containing a variety of genomic mutations at different allelic frequencies.
SeraseqTM AF10 and AF20 control materials consist of a mix of 26 biosynthetic DNA constructs, each containing a specific variant or mutation of interest (MOI) and an internal quality marker (IQM) spiked into genomic DNA from a cell line at specific allelic frequencies. The MOIs represent either known pathogenic somatic mutations or rare mutation types in both oncogenes and tumor suppressor genes described in the Catalogue of Somatic Mutations in Cancer (COSMIC) database. The IQMs are inserted adjacent to each MOI to indicate that the mutations detected are from the plasmids and not from the patient sample.
The goal of this multilaboratory study was to evaluate the Seraseq AF10 and AF20 control materials by verifying their performance as control materials and by evaluating their ability to measure quality metrics essential to clinical assays using both intra- and interlaboratory comparisons.
Material and Methods
Preparation of seraseq solid tumor mutation mix AF10/AF20 control materials
Control plasmid DNA constructs were distributed from the Frederick National Laboratory for Cancer Research and all manufacturing procedures were conducted in a good manufacturing practice-compliant facility. The Seraseq AF10 and AF20 control materials contain 26 control-linearized plasmids each containing a specific MOI and an IQM. In total, 26 plasmids containing 4 INDELs, 4 SNVs in homopolymer regions, and 18 SNVs (Table 1) were selected according to the content of commercially available tumor profiling assays and their genomic complexity (such as SNVs in homopolymer regions), which can be a challenge for some sequencing platforms (11). These were pooled at an equal molar ratio and spiked into genomic DNA from a reference lymphoblastoid cell line derived from a healthy individual (GM24385, Coriell Institute for Medical Research) at 10% and 20% allelic frequencies.
SeraseqTM AF10 and AF20 variants list.
Further, 2 distinct steps were performed to prepare the AF10 and AF20 control materials. Quantification at each step was performed using a droplet digital PCR (ddPCR) (Bio-Rad). In the first step, the 26 plasmids solutions were separately quantitated by a specific ddPCR assay for a sequence element common in all 26 plasmids to assign a concentration to each individual linearized plasmid stock. This quantification allowed mixing of the plasmids into a combined pool such that each plasmid was present at the same molar ratio, and the exact number of copies per microliter of each plasmid within the pool was known. In the second step, duplex ddPCR assays were performed to calculate the allelic frequency of 6 individual mutation targets [see Table 1 in the Data Supplement that accompanies the online version of this article at http://www.jalm.org/content/2/2].
Laboratory 2 verified the quality (or sample integrity) of both the AF10 and AF20 control materials using the KAPA hgDNA Quantification and QC Kit. This real-time PCR assay uses 3 reactions that amplify targets of 41 bp, 129 bp, and 305 bp within a conserved single-copy locus in the human genome. Absolute quantification is achieved using the 41-bp target, and quality is assessed by normalizing the concentrations of the longer amplicons to the 41-bp amplicon. Samples with high-quality DNA (or intact/unfragmented DNA) present Q129/Q41 and Q305/Q41 ratios close to 1, and low-quality (or degraded/fragmented DNA) present Q129/Q41 ratios <1 and Q305/Q41 ratios ≪1.
Study design
This interlaboratory study was conducted over an 8-week period. The study was designed so that 4 replicates were measured in weeks 1 and 8 and only 1 measurement was made in the intervening weeks. Therefore, the design had 14 measurements of each Seraseq standard. This would yield a large enough data set to infer whether there was any time-dependent nature to the measurements because of the stability of the Seraseq control materials or other systematic sources of error in the measurement methods.
For each mutation and laboratory, the measured allele frequencies from week 1 (AF0) and from week 8 (AF) were compared as the ratio AF/AF0. If the hypothesis that observations in week 8 are the same as those in week 1 is to be accepted, then the ratio AF/AF0 = 1. JMP 12 (SAS Institute) was used to analyze the stability data (AF/AF0). Nonparametric Wilcoxon rank sum and Kruskal–Wallis ANOVA tests were performed to determine whether there were any differences between laboratories or mutation types.
This interlaboratory study was conducted by a multicenter group of 6 CLIA-licensed clinical laboratories and 1 research laboratory as follows: Participating laboratories included 6 CLIA-licensed laboratories (Bio-Reference Laboratories, Dartmouth-Hitchcock Medical Center, The Jackson Laboratory for Genomic Medicine, Frederick National Laboratory for Cancer Research-Molecular Characterization Laboratory, Virginia Commonwealth University, Weill Cornell Medical Center) and 1 research laboratory (Seracre Lifesciences).
The laboratories each received 4 vials of Seraseq AF10 and AF20 on dry ice; they also received a handbook detailing safety and handling precautions, −20 °C storage instructions with up to 10 freeze–thaw cycles, indications of reagent instability, and instructions on how to use the control material. Most importantly, the handbook contained a comprehensive set of guidelines detailing sample processing, data transfer, and study timeline.
For weeks 1 and 8, the AF10 and AF20 control materials were processed in 4 replicates to characterize any sample degradation and to monitor intrarun variability. To perform library preparation and sequencing, laboratories were asked to use their routine clinical workflow, which can include use of reagents with different lot numbers and multiple operators. After each sequencing run, the laboratories were requested to fill out a data input table and transfer the sequencing files to the iQ NGS QC Management β version 1.0 software analysis for tracking purposes. The data input table contained both analytical data and bioinformatics data.
Next-generation sequencing
All 7 laboratories provided detailed descriptions of their respective assay parameters, such as NGS panel, genomic content, sequencing method and instrument, DNA extraction method, DNA quantification method, and DNA quality detection (see Table 2 in the online Data Supplement). The laboratories also provided a comprehensive overview of their analytical and bioinformatics pipelines, which included DNA input, control material, data analysis pipeline, software versions, and established QC metrics (see Supplemental Table 3 in the online Data Supplement). Six of the 7 laboratories ran a targeted amplicon-based assay (Ion Torrent PGM: 4 and Illumina MiSeq: 2). One laboratory used a hybrid capture-based assay on the Illumina NextSeq 500. With the exception of laboratory 1, all NGS panels used by the laboratories during this study covered all 26 variants present in the Seraseq AF10 and AF20 control materials (see Supplemental Table 3 in the online Data Supplement). The following variants were not covered by the NGS panel used by laboratory 1: C353fs*5 (ATM)10, A466fs*28 (SMAD4), W288fs*12 (NPM1), S249C (FGFR3), W515L (MPL), R1450* (APC), D835Y (FLT3), R201C (GNAS), and V617F (JAK2).
Each CLIA-licensed laboratory processed the Seraseq AF10 and AF20 control materials using their clinical NGS laboratory-developed procedure. Over the course of 8 consecutive weeks, both AF10 and AF20 were processed along with additional control materials for each sequencing run (see Supplemental Table 3 in the online Data Supplement) and clinical sample for each sequencing run.
Data analysis and statistical analysis
For data analysis, the laboratories uploaded the sequencing run files, input tables, and any additional files associated with each run (if applicable) to the iQ NGS QC Management β version 1.0 software. This software allowed the laboratories to not only upload their FASTQ, BAM, and VCF files but also monitor the analytical and postanalytical QC parameters and metrics used by each of the laboratories. In addition, the iQ NGS QC Management β version 1.0 software extracted the allelic frequencies of each MOI present in the Seraseq AF10 and AF20 biosynthetic mixtures, generating a comprehensive database.
Further, 3 different statistical analyses were performed: (a) MOI and IQM detection, (b) comparison of sequencing platform and targeted sequencing methods, and (c) stability and reproducibility. MOI and IQM detection was performed to verify the performance of the control materials in each laboratory. It consisted of pooling each laboratory's data, splitting it into 2 groups (AF10 and AF20), and calculating the mean, SD, and coefficient of variation for each of the 26 variants (and IQMs) detected in all sequencing runs across the 7 laboratories. Comparison of sequencing platform and targeted sequencing methods was designed to evaluate the performance of the control materials using different sequencing platforms and methods. It included the data generated in the first analysis and the data input table provided by each laboratory for each run. Finally, the stability and reproducibility analyses were performed to verify the degradation and reproducibility over the 8-week study period. For each mutation and laboratory, the allelic frequencies from week 1 (AF0) and week 8 (AF) were compared as the ratio AF/AF0. Because some laboratories did not perform this study in 8 weeks, AF was calculated as their last week. The nonparametric Wilcoxon rank sum and Kruskal–Wallis ANOVA tests were performed using the JMP software (SAS Institute). The following 2 additional distributions were performed to parameterize the data: log-normal (best 2-parameter distribution) and Johnson SU distribution (a 4-parameter distribution).
Results
Precise quantification of each plasmid using ddPCR allowed the Seraseq control materials to have the same molar ratio and the exact number of copies per microliter of each plasmid within the pool. The average percentage of allelic frequencies and the SD of the 6 targets was 10.3% (0.58) and 18.0% (0.44), respectively, indicating that accurate titration was achieved for both control materials.
To assess the integrity (or quality) of Seraseq AF10 and AF20, laboratory 2 performed real-time PCR using the KAPA Quantification and QC Kit on 2 aliquots of each control material. All 4 aliquots showed similar concentrations for all 3 amplicon reactions and “Q ratios” of 1.
The goal of the first analysis was to verify the performance of the AF10 and AF20 control materials in each laboratory. The data were organized into 2 groups—allelic frequency of each individual MOI and allelic frequency of the MOIs according to their type. Further, 13 of the 26 MOIs in the control mix were detected by all laboratories (Table 2). Laboratory 1 used an assay that did not cover 9 of the 26 mutations. Laboratories 1, 2, 6, and 7 detected all variants covered by their sequencing panel (Table 2). Laboratories 3, 4, and 5 detected the majority of the INDELs and SNVs present in both AF10 and AF20. Most of the MOIs and their corresponding IQMs were detected by the laboratories. However, some discrepancies were identified in both sequencing platform groups.
Variants detected (
), not detected (
), or detected at least in 1 run (
) in AF10/AF20 by each laboratory.
When comparing allelic frequency to variant type, Laboratories 1, 2, and 6 had the lowest SDs (AF10: 1.59–1.80%, AF20: 2.19–2.99%) (see Fig. 1 and Table 4 in the online Data Supplement). INDELs and SNVs in homopolymer regions showed the highest SDs (Fig. 1 and Table 4 in the online Data Supplement), and it was not sequencing platform- or target sequencing method-dependent. Seraseq AF10 showed the following SDs for deletions, insertions, SNVs and SNVs in homopolymer regions: 0.91%–2.59%, 0.93%–3.18%, 1.58%–2.69%, and 1.48%–4.86%, respectively. For Seraseq AF20, the SDs were 0.85%–3.09%, 1.10%–5.12%, 2.41%–3.47%, and 1.70%–6.01%, respectively.
NGS panel used by laboratory 1 does not cover both insertions present in AF10 and AF20 (SMAD4: p.A466fs*28 and NPM1: p.W288fs*12).
The second analysis was assessed to verify the performance of the AF10 and AF20 control materials by grouping the laboratories according to the sequencing platform and the targeted sequencing method. In each group, 3 laboratories failed to detect at least 1 MOI and had issues detecting some IQMs [sequencing platform: Ion Torrent PGM (laboratory 4) and Illumina (laboratory 3 and laboratory 5); targeted sequencing method: amplicon-based (laboratory 4 and laboratory 5), hybrid-capture (laboratory 3)]. Furthermore, 3 laboratories in each group had high SDs for INDELs [sequencing platform: Ion Torrent PGM (laboratory 7: 2.17%–3.09%) and Illumina (laboratory 3: 2.80%–5.12% and laboratory 5: 2.36%–4.66%); targeted sequencing method: amplicon-based (laboratory 5: 2.36%–4.66% and laboratory 7: 2.17%–3.09%), hybrid-capture (laboratory 3: 2.80%–5.12%)]. For the AF10, 4 laboratories in each group had high SDs for SNVs in homopolymer regions [sequencing platform: Ion Torrent PGM (laboratory 4: 2.41% and laboratory 7: 2.48%) and Illumina (laboratory 3: 2.03% and laboratory 5: 4.86%); targeted sequencing method: amplicon-based (laboratory 4: 2.41%, laboratory 5: 4.86%, and laboratory 7: 2.48%), hybrid-capture (laboratory 3: 2.03%)]. For the AF20, 6 of 78 laboratories showed high SDs for SNVs in homopolymer regions (laboratory 1: 4.16%, laboratory 2: 2.39%, laboratory 3: 3.08%, laboratory 4: 4.39%, laboratory 5: 6.01%, and laboratory 7: 3.24%). Laboratory 6 showed a SD of 1.70%.
Finally, the third analysis was performed to verify stability over the course of the 8-week study using the nonparametric Wilcoxon rank sum and Kruskal–Wallis ANOVA tests. This analysis showed that there were no differences between laboratories or mutation types. In each case, no difference was seen with >99% confidence interval. This enabled all of the data to be pooled and to probe whether AF/AF0 = 1. The distribution of AF/AF0 was positively skewed and very tightly distributed around AF/AF0 = 1 (Fig. 2). The median degradation was AF/AF0 = 0.99. Using the Johnson SU distribution for the AF10 standard, the mode for AF/AF0 = 0.96, the median = 1.01, and the 95% confidence interval = 0.64–1.52. For the AF20 standard, the mode for AF/AF0 = 0.96, the median = 1.08, and the 95% confidence interval = 0.74–1.85. The reproducibility experiments within and between laboratories showed concordance for variant detection and allelic frequency (Fig. 3). However, several laboratories showed small variations in allelic frequency during the study (Fig. 3).
The allelic frequencies from week 1 (AF0) and from the last week (AF) were compared as the ratio AF/AF0. The distribution of AF/AF0 was clearly positively skewed and very tightly distributed around AF/AF0 = 1, and the median degradation was AF/AF0 = 0.99. Log-normal and Johnson–SU distributions were performed and the hypothesis AF/AF0 = 1 could not be rejected with >95% confidence.
Discussion
At present, there is no ideal control material approved or recommended to monitor the performance of NGS-based tumor profiling tests. CAP/CLIA guidelines and professional laboratory organizations recommend the use of control materials to monitor performance parameters throughout the NGS pipeline. Some vendors offer commercially available types of control materials, such as genomic DNA from blood or cell lines, FFPE cell lines, and synthetic DNA. All have advantages and disadvantages (12). Ideal control materials should mimic the different variant types found in the human genome (6, 12). Reference materials should also contain variants located in challenging regions.
The DNA integrity assessment performed by Laboratory 2 showed that both control materials had similar concentrations for all 3 amplicon reactions and Q ratios of 1. Samples with high DNA quality were expected to have these characteristics, suggesting high DNA quality in the GM24385-derived AF10 and AF20 pools. This study showed that both biosynthetic mixes were highly stable, showing no degradation during the 8-week period of the study. Both control materials generated reproducible results when performed by different operators using different sequencing platforms and target sequencing methods.
Recently, Sims et al. (13) published a study on plasmid-based control materials, CPSGs (control plasmid spiked-in genomes), as multiplex quality control materials for clinical NGS assays. They compared the performance of biosynthetic DNA plasmid control materials with cell line FFPE gDNA using different NGS assays designed for different platforms. Their results showed that the plasmid constructs were detected with similar efficiencies on 2 sequencing platforms and target amplicon sequencing methods as the gDNA that was extracted from the FFPE fixed cells bearing the same types of variants over a wide range of allelic frequencies. CPSGs were useful for evaluating assay performance across multiple library preparation methods, sequencing platforms, and data analysis pipelines, and it was reproducible over long periods. Although Sims et al. (2016) (13) showed that CPSGs showed great promise as standardized control materials for clinical NGS assays, our study showed that the AF10 and AF20 control materials could be used as control materials for assay validation and for QC across multiple analytically validated NGS assays in CLIA-certified laboratories.
The Seraseq AF10 and AF20 control materials had many variants at multiple allelic frequencies of 10% and 20%, respectively. This study showed that the majority of the 26 MOIs were detected by laboratories at 7 centers using different protocols. Among the 26 MOIs, INDELs and SNVs in homopolymer regions showed the highest SDs independently of sequencing platforms and targeted sequencing methods. One of the limitations of the Ion Torrent chemistry is the incidence of errors in sequencing homopolymer regions (11), which requires the use of software capable of recalibrating for these errors (14).
Although most of the MOIs and their corresponding IQMs present in the control materials were detected by the laboratories, there were some isolated discrepancies identified in both sequencing platforms. The Ion Torrent PGM failed to detect (or detected sporadically) some IQMs close to the end of the amplicon, and the Illumina platforms failed to detect (or detected sporadically) IQMs located on the primer binding region of the amplicon. This study showed that the discrepancies in which the IQM was not detected were dependent on the platform used. Because the IQM indicates the accuracy and stability of both AF10 and AF20 by showing that the MOIs detected are from the control materials and not from the patient sample, a potential solution for these discrepancies would be to modify them to be sequencing platform-independent. Another potential solution would be to establish individual performance metrics by the laboratories during the analytical validation process. According to Sims et al. (13), MOI and IQM discrepancies could be resolved by either moving the primers further away and downstream from the MOI or by moving the molecular barcode toward the center of the amplicon.
The Seraseq AF10 and AF20 control materials have characteristics of robust, highly multiplexed control materials for assay comparison, proficiency testing and tracking, and trending assay performance over time. The control materials are stable through multiple freeze–thaw cycles; the DNA is accurately quantified by ddPCR; and the materials contain SNVs, SNVs in homopolymer regions, and small and large INDELS. This interlaboratory study showed that the Seraseq AF10 and AF20 control materials are appropriate for use as control materials during analytical validation process and as positive control samples in routine clinical tests.
The AF10 and AF20 control materials can be used to show an assay's ability to consistently detect clinically significant mutations located in regions that are technically challenging to sequence. In addition, the control materials can be used to show an assay's ability to consistently detect mutations that are bioinformatically challenging to identify, such as large INDELs, e.g., the L858R mutation (located in a homopolymeric region) of the EGFR gene and a 15 bp or 18 bp deletion in exon 19 of the EGFR gene. Both mutations confer sensitivity to first-, second-, and third-generation EGFR tyrosine kinase inhibitors in patients diagnosed with non-small cell lung carcinoma (NSCLC). The detection of both clinically and nonclinically relevant variants in the control material showed excellent performance of the validated NGS assay. Finally, Seraseq AF10 and AF20 may be used as routine control materials during the analytical validation process and in clinical NGS sequencing runs for cancer panels. This is the first time that data from several CLIA laboratories have been compared in a fully transparent way.
Additional Content on this Topic
Karen Page, David S. Guttery, Daniel Fernandez-Garcia et al. Clin Chem 2017;63:532–41
Acknowledgments
We thank Guru Ananda from the Jackson Laboratory for Genomic Medicine; David J. Sims, Kneshay N. Harper, and Vivekananda Datta from the Molecular Characterization and Clinical Assay Development Laboratory; and Jessica Dickens and Alice Ku from SeraCare Life Sciences Inc. for their assistance in this project.
Footnotes
↵9 Nonstandard abbreviations:
- NGS
- next-generation sequencing
- FFPE
- formalin-fixed paraffin-embedded
- MOI
- mutation of interest
- IQM
- internal quality marker
- SNVs
- single nucleotide variants
- CNVs
- copy number variations
- INDELs
- insertions and deletions
- PT
- proficiency testing
- MOI
- mutation of interest
- IQM
- internal quality marker
- ddPCR
- droplet digital PCR
- gDNA
- genomic DNA
- CLIA
- Clinical Laboratory Improvement Amendments.
↵10 Human genes:
- ATM
- ATM serine/threonine kinase
- SMAD4
- SMAD family member 4
- NPM1
- nucleophosmin
- EGFR
- epidermal growth factor receptor
- FGFR3
- fibroblast growth factor receptor 3
- GNAQ
- G protein subunit alpha q
- AKT1
- AKT serine/threonine kinase 1
- BRAF
- B-Raf proto-oncogene, serine/threonine kinase
- KRAS
- KRAS proto-oncogene, GTPase
- PIK3CA
- phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha
- NRAS
- NRAS proto-oncogene, GTPase
- TP53
- tumor protein p53
- CTNNB1
- catenin beta 1
- IDH1
- isocitrate dehydrogenase [NADP(+)] 1, cytosolic
- MPL
- MPL proto-oncogene, thrombopoietin receptor
- APC, APC
- WNT signaling pathway regulator
- FLT3
- fms related tyrosine kinase 3
- PDGFRA
- platelet derived growth factor receptor alpha
- RET
- ret proto-oncogene
- GNAS
- GNAS complex locus
- KIT
- KIT proto-oncogene receptor tyrosine kinase
- JAK2
- Janus kinase 2.
Authors' Disclosures or Potential Conflicts of Interest: Upon manuscript submission, all authors completed the author disclosure form.
Employment or Leadership: R. Daber, Bio Reference Laboratories; L. Davis, Frederick National Laboratory of Cancer Research; R.K. Garlick, A. Ferreira-Gonzalez, C.I. Dumur, S. Haralampu, and G.J. Tsongalis, SeraCare Life Sciences Inc.
Consultant or Advisory Role: R. Daber, SeraCare; R. Harrington, Seracare.
Stock Ownership: S. Haralampu, Equity.
Honoraria: F.B. de Abreu, Seracare.
Research Funding: SeraCare provided reagents and materials for this study. R.K. Garlick, SeraCare.
Expert Testimony: None declared.
Patents: None declared.
Other Remuneration: F.B. de Abreu, SeraCare.
Role of Sponsor: The funding organization played no role in the design of study, choice of enrolled patients, review and interpretation of data, or preparation or approval of manuscript.
- Received January 12, 2017.
- Accepted May 5, 2017.
- © 2017 American Association for Clinical Chemistry
References
In this issue
