Organisation – QIAGEN
QIAGEN is a leading company in the field of molecular biology. Started in the early 1980’s, they began by making extractions of DNA from biological samples in the laboratory simpler and safer. Since those early days of biotechnology, the company has grown consistently and gone from a small entrepreneurial start-up to become a global market leader, with around 5000 employees in a hierarchical structure spread across more than 25 countries, offering over 500 core products and supporting in excess of 500,000 customers world-wide. From CEO to Technical Assistant, or even placement student, everyone from the top down in the organization embraces core values, like innovation and integrity, to drive quality in their work, in order to ‘make improvements in life possible’. With products for use in all areas of life sciences, forensics, molecular diagnostics and food safety, customers include doctors, vets, scientists, researchers, forensic experts and police departments, food standards agencies and even government departments.
Their recently introduced strategic framework of ‘Sample to Insight’ is based on a desire to provide any of these customers, wherever they may be, with a full portfolio of molecular testing applications, equipment and products, including fully automated workflows, to gain an insight from a sample. This means QIAGEN are market leaders in handling, extraction, purification and stabilization of DNA, RNA and proteins from any biological sample type, no matter how difficult this challenge may be. For example, QIAGEN are closely involved in supporting a project run in conjunction with the U.N. and I.C.M.P. (International Commission on Missing Persons) helping identifying remains from mass graves in war torn areas through proprietary Next Generation Sequencing (NGS) techniques developed specially for highly degraded DNA.
On joining QIAGEN I was trained to work in the qPCR (quantitative polymerase chain reaction) product development department. Providing customers of QIAGEN with better, quicker, more efficient means of gaining valuable molecular insights from their biological samples, launching new products to market as well as improving existing ones. Molecular insights can include anything from companion diagnostics, ensuring cancer patients get the most effective treatment, personalised healthcare and monitoring due to accurately pinpointing specific mutations linked to their type of cancer, to forensic investigators providing irrefutable DNA evidence to assist in prosecuting a criminal. Here at QIAGEN Manchester in the qPCR department a lot of the work is on companion diagnostics for oncology, or other pathological conditions.
I worked on different projects, both directly and in a key support role as a member of the pre-analytical (P.A.) team. Core skills developed include planning experiments, time management, data analysis, presentations and report writing. Laboratory experience and extensive training available has meant I have learned, or improved, many new skills, techniques and the use of new equipment. The P.A. Team carry out extractions of DNA and RNA on a daily basis from a wide variety of biological sample types, using fully automated elements of the QIAGEN workflow like the QIASymphony, as well as manual extractions such as detailed in this report. Other experience gained, not detailed in the following report, includes reverse transcription of RNA to cDNA, Sanger and pyro-sequencing, NGS, quantitation and evaluation of extracted DNA and RNA using spectrophotometry and fluorometry, physical fragmentation of DNA via sonication, preparation of master mixes of re-agents, dilutions and other RT-qPCR based work than that described in this report. Approximately 20% of my time during my placement year has been spent on training and development, 30% on my student project, the remaining 50% working directly for QIAGEN.
Diagnostic errors in the pre-analytical phase
According to the 2015 report by the Institute of Medicine, ‘Improving Diagnosis in Healthcare’, errors or delays in diagnosis such as; failure to utilise correct and up to date testing methods, failure to act correctly on results or failure to interpret results correctly, for example due to poor or degraded samples giving false or inconclusive results, cause up to 10% of patient deaths and around 17% of all adverse events patients experience during treatment. These figures vary geographically and by method of data collection, they are also not deemed adequate to extrapolate the incidence of diagnostic errors in clinical practice today, but should be troubling particularly as many of these deaths and poor treatment decisions leading to patient harm have the potential to be avoided. Human error accounts for some, but it is argued that although such isolated errors occur, these often interact with or are instigated by other, more basic, systemic errors. Schiff et al. (2009) define diagnostic error as “any mistake or failure in the diagnostic process leading to a misdiagnosis, a missed diagnosis, or a delayed diagnosis” and divides the diagnostic process into seven stages, of which at least four can be classed as part of the ‘pre-analytical’ phase.
Errors or failures in diagnostic decisions should not be compounded by providing physicians with inaccurate or misleading information for them to base their therapeutic recommendations on. In modern healthcare perhaps 70% or more of clinical decisions are now based on results of laboratory testing, testing that can be complex and carries with it an inherent risk of mistakes. While it is true that advances in technology and improving standards of quality control have significantly reduced error rates, some studies have shown most laboratories focus on the analytical aspect only when assessing quality. Plebani (2006) summarises a growing body of evidence to suggest that anywhere from 46 – 68.2% of total laboratory errors occur during the pre-analytical phase, this means pre-analytical errors make up the majority of total test errors, although post-analytical errors were also high at an estimated 18.5 – 47%.
The pre-analytical phase typically starts outside of the laboratory setting, with testing and sample collection from the patient, followed by storage and transport of samples prior to analysis, often this is carried out off-site from the original test location. Many factors can impact the sample quality depending on the nature of sample, this report will discuss predominantly the collection, storage and testing of blood.
Kaushik and Green (2014) recognise the importance of standards, both internal such as Standard Operating Procedures (SOP’s) and full compliance with external Government guidelines and best practice, in improving patients experience by significantly reducing the incidence of pre-analytical errors. These errors reduce confidence in healthcare, can damage reputations and have huge financial implications, ultimately, they can also endanger patient health, so what more needs to be done?
In 2009 the European Union, under its 7th Framework programme, recognized the scale of the problem and the costs to healthcare and damage to patients and their families that were being caused by avoidable diagnostic errors. In the changing healthcare landscape, as genetic medicine and molecular biological analysis becomes more and more prevalent, the European Commission funded a project aimed at improving the potential and utility of in-vitro diagnostics, through the creation of new standards and quality assurance schemes for the collection, handling and processing of blood, tissue, tumour and other sample types. Set up to standardise practices across hospitals, doctors’ surgeries, laboratories and other testing sites for the collection, handling, storage and processing of patient samples, all of which are critical steps in the pre-analytical phase, the SPIDIA project (Standardisation and improvement of generic Pre-analytical tools and procedures for In-vitro DIAgnostics) brought together 16 companies and research institutions from 11 countries across Europe, headed and coordinated by QIAGEN GmbH Ltd.
Accurate analysis and diagnosis of samples, affected by factors such as temperature, time lapse between taking and processing samples, stabilization, fixation and preservation methods, is vitally dependent on the sample integrity. The molecular profile of samples can mean target biomarkers can change or disappear due to incorrect handling. The commission realized molecular diagnostics using DNA and RNA would be a vital and integral part of future European healthcare, as part of a new era of personalised healthcare, so over four and a half years of collaborative research and knowledge sharing several new pre-analytical technologies were developed. Within the CEN/Technical Committee 140 for “In vitro medical devices”, SPIDIA’s results enabled the development and introduction of the first 9 CEN Technical Specifications (CEN/TS) for pre-analytical workflows in Europe (www.spidia.eu, 2018) These achievements will not only help to begin standardising and improving the pre-analytical workflow and reducing diagnostic errors, but have lead on to further projects such as SPIDIA4P, standardising pre-analytical workflows in personalized medicine as well as the STRATFix project here in the UK.
Further to the SPIDIA European project, a three year, £1m UK project was launched early in 2015, funded by the Technology Strategy Board – Innovate UK. The STRATFix consortium was set up to develop integrated formaldehyde free systems for biological sample collection, stabilisation and nucleic acid purification (CSP Systems) and to attempt to integrate these into routine clinical practice within the NHS. The aim was to provide ‘across the board’ solutions for fixing and preserving solid tissues, fine needle aspirate and core needle biopsies and for stabilising ccfDNA in blood and liquid biopsies to enable more reliable and accurate downstream genetic and molecular biology based analysis. Project coordinator QIAGEN is a market leader in CSP systems to cover the entire “sample-to-insight” workflow. STRATFix was a UK-wide collaboration, including as members 8 opinion-leading UK NHS hospitals, thus project outcomes have the potential for uptake by the whole UK pathology community. Here the approach was first to provide the systems to support improving pre-analytical workflows, future work is being planned to incorporate these into new, applicable ISO standards (www.spidia.eu/stratfix-project, 2018).
Modern oncology is increasingly moving towards cancer treatments tailored specifically to individual patient’s needs, this is usually referred to as ‘personalised healthcare’ or ‘stratified medicine’. STRATFix comes from the full project title: ‘Enabling STRATified medicine with novel FIXatives for improved pre-analytical pathology workflows’. Research has found that current sample preparation methods allow morphological assessment of cells and tissues, yet do not always preserve nucleic acids well enough for the genetic and molecular analysis required for stratified medicine. The STRATFix study aims to improve these methods, especially preservation of nucleic acids in samples, so that a full range of test results can be obtained to help select and tailor the right treatment for individual patients (www.gtr.ukri.org., 2015).
The STRATFix project was divided into three major areas based on the main biological sample types. The business-led consortium developed prototypes of commercially relevant products, whilst the participation of a diverse geographical network of NHS partners assisted in the adoption of these novel tools and products into routine clinical practice. As can be seen in Fig. 1 below the scientific approach was to address the problem working from most to least invasive biopsy type. Work Package 1 involved implementation of the PAXgene Tissue system into routine clinical use. Work Package 2 required the development and validation of a new CSP-system for fine needle aspiration (FNA) cytology samples. Finally, in Work Package 3 the subject of investigation was liquid biopsies, in particular blood, and the development and validation of a circulating cell free DNA (ccfDNA) stabilising CSP system.
Figure 1: Scientific approach for STRATFix project, key driver was developing formaldehyde free preservation and stabilization methods for all the main biopsy/sample types, optimizing the pre-analytical workflows for each sample type, then validating each for use in clinical settings. Work was split into 3 key areas, starting with most invasive procedures.
Speicher et al, (2015) firmly believe in the tremendous potential for liquid biopsies in future developments in oncology, stratified medicine and the application of true targeted therapies, otherwise known as companion diagnostics. In oncology, “liquid biopsy” refers to the analysis of ccfDNA to establish, non-invasively from the peripheral blood, the characteristics of a tumour genome in cancer patients. It is the ease and non-invasive nature of taking blood samples that makes them ideal for cancer patients, who may not be strong enough to undergo surgery to obtain sections of tumours, or the tumours themselves may not be readily accessible via surgery. Liquid biopsies can be taken regularly and often, so allow doctors to monitor the progress of disease, metastasis and/or mutations and responses to treatment, giving a clear, ongoing and up to date picture of individual patient’s health. Another issue is that of tumour heterogeneity, along with selection and evolution, which means tumours can and often do develop treatment resistance. Standard biopsies only provide ‘snap-shots’, to overcome this methods are needed for rapid, cost-effective, and non-invasive identification of biomarkers at various time points during the course of the disease. Because circulating tumour DNA (ctDNA) is a potential surrogate for the entire tumour genome, liquid biopsies can be used to obtain the genetic data needed (Heitzer et al, 2015).
ccfDNA in oncology
It was in 1948 that Mandel and Métais first identified cell free nucleic acids in human blood, yet this was not linked to disease states for nearly 30 years until Leon et al (1977) reported on the correlation between high levels of ccfDNA in cancer patients and how these levels reduced post therapy, however, they found ‘normal’ levels in ~50% of patients with cancer and at this time felt ccfDNA concentration would have a low diagnostic value. During the mid to late nineteen nineties the importance of cell free nucleic acids as oncological biomarkers started to become clearer, when Sorenson et al (1994) found mutated RAS genes and Nawroz et al (1996) discovered microsatellite alterations on ccfDNA in the blood of cancer patients. Over the past two decades much research has been conducted in this area, as liquid biopsies with reliable biomarkers for cancer could prove a very useful diagnostic tool, however, levels of ccfDNA can also be affected by non-cancer related pathologies (Schwarzenbach et al, 2011). For the purpose of this report the terms ‘ccfDNA’ and ‘ctDNA’ have been used interchangeably, as many types of ccfDNA exist.
It has been demonstrated that ccfDNA yields are significantly higher in patients with malignant tumours than in healthy individuals, yet the ccfDNA burden has also been found to increase in patients with benign tumours and other inflammatory diseases, as yet the physiological events leading to increased ccfDNA levels in liquid biopsies have not been fully characterised. However, the release of nucleic acids into the blood stream is believed to be linked to apoptosis and necrosis of cancer cells, as well as secretion and metastasis. Some estimates indicate that for a ‘typical’ patient with a 100g tumour, equating to ~3 x 1010 tumour cells, up to 3.3% tumour DNA can enter the circulation every day. Even with fairly large patient to patient variability healthy subjects have concentrations between 0 and ~100ng/ml ccfDNA in their blood, with an average of 30ng/ml ccfDNA, as compared to a range of between 0 and >1,000ng/ml of whole blood, with an average of 180ng/ml ccfDNA measured in cancer patients (Schwarzenbach et al, 2011).
As mentioned above, ccfDNA yields can be indicative of disease states, however, the critical clinical applications only begin here. Prognosis and response to treatment can be quickly and easily measured as proportion of ccfDNA correlates to size and stage of tumour progression, generally more advanced or aggressive tumours can contribute between 10-100 times more ccfDNA than those in early stages. Higher ccfDNA yields also indicate worse prognosis and lower survival rates, while significant drops in ccfDNA yields are indicative of favourable responses to treatments and can be detected earlier than with any other current test method (Stewart et al, 2018). It is then molecular identification of mutational causes underlying the patients’ cancer type that can guide tailored treatment, as well as ongoing monitoring of patients to detect any acquisition of resistance mutations as timely identification allows early intervention.
Elshimali et al (2013) discuss the clinical utilisation of ccfDNA as an oncological biomarker in a wide range of carcinomas and the significant progress made in detection and quantitative or qualitative analysis of tumour mutations and alterations such as DNA integrity, genetic abnormalities and/or methylation and microsatellite instabilities as markers for diagnosis, prognosis and monitoring. Yet they highlight the problems with lack of standard practices, particularly in the pre-analytical phase, as well as a need for further clinical studies to truly optimize and implement this type of testing into clinical practice.
These findings are echoed by Merker et al (2018) in a joint review by the American Society of Clinical Oncology and College of American Pathologists, they found little evidence to support the clinical validity and widespread utility of ccfDNA assays, predominantly due to a paucity of clinical trials to date. This was with the clear exception of assays already validated by regulatory approval (CE, FDA, etc.). The review also highlighted the pre-analytical parameters as critical to the accuracy of any such assays and in particular discussed the impact of tube type and specimen handling on ccfDNA detection. They found no head to head performance comparison studies had been reported at the time of publication, which is an area this report aims to address. They found current evidence strongly suggests plasma as the optimal sample type for extracting and characterising ccfDNA as opposed to serum, primarily due to the fact that genomic DNA released from lysis of leukocytes, which dilutes and masks ccfDNA, is found at much lower levels in plasma. This is even though actual levels of ccfDNA are typically higher in serum than plasma.
EGFR and cancer
It is well documented that the family of epidermal growth factor receptors (EGFR) are typically over expressed, over active, or otherwise mutated in a wide range of cancers or tumour types, see table 1 below for summary. The ranges reported may be wide due to a variety of methods used to establish expression levels and clinical studies to date have failed to find a direct correlation between expression levels and prognosis for cancer patients, however, numerous clinical and preclinical studies have established the involvement of EGFR signalling pathways in tumour pathogenesis. It has been hypothesized therefore that establishing global expression levels of these receptors could have diagnostic, prognostic and therapeutic applications and benefits to patients (Normanno et al, 2008).
Table 1: Expression of EGFR family of receptors in human carcinomas (Normanno, N. et al, 2008)
EGFR has been well established as an important oncogene, despite the many differences in study methodology, mutant EGFR genes have been widely reported to correlate with a poor patient prognosis and aggressive tumour phenotype. Particularly as EGFR signalling has been linked not only to increased tumour survival and proliferation, but also metastasis, migration and invasion due to interactions with cell adhesion and motility molecules. EGFR antagonists may stem metastasis and can offer improved therapeutic responses from cancer patients with certain specific EGFR mutations (Box, C. et al, 2008).
Phase II clinical trials of the drugs erlotinib and gefitinib, both EGFR Tyrosine Kinase Inhibitors (TKI’s), on Non-Small Cell Lung Cancers (NSCLC) showed certain groups of patients exhibited remarkable therapeutic responses and clinical improvements with treatment, profound enough to be classified as a ‘Lazarus effect’. These were found to be associated with somatic activation mutations in the tyrosine kinase domain (TKD). ‘Classical’ TKD mutations such as Deletions and L858R are very sensitive to these drugs. Conversely T790M was subsequently discovered as an acquired secondary resistance mutation that can convert sensitized mutant receptors to drug resistant, thus identifying which mutations are present in patients has a huge impact on the effectiveness of drug therapies (Eberhard, D.A., 2008).
According to Eberhard (2008) NSCLC’s typically exhibit considerable tumour heterogeneity. This can be both within a single tumour with differentiated sub-populations of cells as well as between different tumours in the same patient, but also in the same tumour mass over a period of time, as tumours can evolve resistance during treatment (e.g. acquisition of T790M mutation). Therefore, reliable detection of mutant EGFR oncogenes via liquid biopsies offers a superior process to monitor ongoing disease states.
Molecular stratification and detection of specific EGFE mutations allows targeted therapy, QIAGEN’s therascreen EGFR Plasma RGQ PCR Kit is one of only two in vitro diagnostic tests approved by the FDA and CE marked for the detection of exon 19 deletions and exons 20 and 21 substitutions (T790M and L858R respectively) in the EGFR gene (Stewart et al, 2018). This kit provides a qualitative assessment of the mutation status using ccfDNA from liquid biopsies and the results can aid doctors in identifying patients with non-small cell lung cancer (NSCLC) who may benefit from treatment with IRESSA (gefitinib).
Study aims and objectives
This study covers an area of Work Package 3 (WP3.3) within the larger STRATFix project, this will be discussed towards the end of the report in conjunction with other WP3 activities conducted by other members of the consortium and other scientists at QIAGEN. This study aims to investigate ccfDNA stability in blood stored in different tubes. There are three key objectives to be addressed by this study:
1. Comparison of the effectiveness of ccfDNA stabilisation between PAXgene ccfDNA blood tubes against K2-EDTA blood tubes across multiple time points – 0, 1, 3 and 7 days (when kept in storage at ambient room temperature, 15-25°C), to be assessed by RT-qPCR using QIAGEN therascreen EGFR PCR kits.
2. Comparison of the NEW QIAamp Minelute ccfDNA Mini/Midi Kit and OLD QIAamp Circulating Nucleic Acid Kit for extraction of ccfDNA from plasma to ensure equivalency between extraction methods in terms of DNA yields.
3. Validation of the NEW QIAamp Minelute ccfDNA Mini/Midi Kit in conjunction with PAXgene ccfDNA stabilising blood collection tubes for STRATfix project WP3 as a complete CSP system workflow for liquid biopsies.
It is hypothesised that EDTA blood tubes will not stabilise ccfDNA when stored at room temperature for anything over 1 day, while it is hoped the PAXgene ccfDNA stabilising blood tubes will for a period of at least 7 days. This would represent a significant potential improvement to the pre-analytical workflow for CSP systems for liquid biopsies intended for molecular analysis for patient companion diagnostics, or stratified medicine. We also hypothesised that the newly introduced QIAGEN QIAamp ccfDNA minelute extraction kit will be quicker and easier to use then the current QIAGEN QIAamp circulating nucleic acid extraction kit due to the fact it utilises magnetic bead technology, it is hoped the results show equivalent or higher yields of extracted ccfDNA using the new kit compared to the old kit, for like for like samples.
Materials & Methods
All work was done according to relevant QIAGEN S.O.P.’s or handbooks (including H;S, PPE, etc.) and recorded on approved experimental worksheet supplements and in the associated laboratory notebook (no. 1048). Lab book, supplements folder and raw data files may be available to review upon request. Wet work and planning was carried out with Jessica Jefferies, a placement student from Manchester Metropolitan University. For the purpose of scientific rigour all elements of laboratory wet work were split 50/50, with each student carrying out alternating repeats, to reduce any potential operator bias in results.
In order to mimic circulating tumour DNA, ctDNA or ccfDNA, and enable downstream detection and analysis of the stabilization of ccfDNA in the alternative blood tubes, via the QIAGEN Therascreen EGFR PCR kit, it was first necessary to obtain the relevant mutant DNA and dilute and digest to similar sized fragments and an approximate concentration to give a Mutant Allelic Fraction (% M.A.F.) in the spiked blood samples, similar to what may be found in clinical samples. The reference stock of genomic EGFR mutation DNA from human cell lines was bought in from Horizon Discovery; EGFR ?E746-A750 Reference Standard for deletions in Exon 19 (HD251), EGFR T790M Reference Standard (HD258) and EGFR L858R Reference Standard (HD254) for specific mutations found in Exons 20 and 21 respectively. The original Horizon DNA concentrations of 50ng/µL, heterozygous for EGFR mutations and 50% M.A.F. was calculated using an online calculating tool from the URI Genomics ; Sequencing Centre to give ~37,000 copies/µL (based on normal human genome size and average base pair weight of 650 Daltons). This was then diluted by a factor of 12.3? using QIAGEN AVE Buffer (ion free, highly purified RNase free water + 0.04% sodium azide) to obtain sufficient working stock of each mutant DNA at ~3,000 copies/µL.
The mutant EGFR genomic DNA is from human cell lines, in order to represent circulating tumour DNA, or ccfDNA, it was necessary to fragment this long genomic DNA into smaller pieces, the aim was to have numerous fragments of smaller lengths, between ~50-800bp, with less than 2% of the total number of fragments ;1000bp long. It had been found in previous QIAGEN SRATFix experiments that random digestion of the genomic DNA, using either physical fragmentation methods such as sonication or non-specific endonucleases, e.g.; recombinant DNAse1, led to an inability to detect the target mutations in downstream applications, due to cutting within the target sites and an inability for primers to bind in PCR, or to detect target sequences using NGS. Therefore, considerable time and effort by scientists Maryam Zahedi and Natasha Cant (QIAGEN) went into selecting highly specific, targeted, restriction digestion enzymes, using the online tool NEBcutter V2.0. Enzymes were carefully selected on the following basis; the restriction site must not present near or inside the interest site, it should not present in the Internal Control sequence or binding site for forward or reverse primers, the enzyme should be inactivated by heat and not be methylation sensitive. See Fig. 2 and 3 for enzyme selection details and estimated fragments, in order to simplify workflow and reduce costs the same enzymes were used, where possible, and the minimum number of enzymes were used to achieve the desired results. It was still necessary to have a modified procedure for the T790M and L858R mutations from the Deletions. The EGFR Deletion mutant reference DNA was digested using three restriction endonucleases; AvaII (R0135S), HincII (R0103S) and ScrFI (R0110S). The EGFR T790M and L858R mutant reference DNAs was digested using three restriction endonucleases; ApoI (R0566S), EcoRI (R0101S) and Hpy1881 (R0617S). All in order to obtain DNA fragments varying in length from 80-800 bp (all enzymes supplied by New England Biolabs).
Fragments of appropriate lengths were created, specifically containing EGFR Mutations target sequences = DNA Base Change – Deletions: AAGGAATTAAGAGAAGCA ? AA in Exon 19 (COSMIC ID 12678), DNA Base Change – T790M: ACG?ATG in Exon 20 (COSMIC ID 6240), DNA Base Change – L858R: CTG?CGG in Exon 21 (COSMIC ID 6224). COSMIC = Catalogue Of Somatic Mutations In Cancer.
RESTRICTION ENZYMES chosen for Deletions (Exon 19)
Enzyme Name Restriction Sequence Restriction Sites
G GWC C
CC N GG
CCAGG, CCTGG, CCGGG, CCCGG
GTCAAC, GTTAAC, GTCGAC, GTTGAC
RESTRICTION ENYZMES CHOSEN FOR: T790M (Exon 20) L858R (Exon 21)
Enzyme Name Restriction Sequence Restriction Sites
ApoI R AATT Y AAATTT, AAATTC, GAATTT, GAATTC
EcoRI G AATT C GAATTC
Hpy1881 TC N GA TCAGA, TCTGA, TCCGA, TCGGA
Figure 2: Comparison and justification for selection of the 3 restriction endonuclease enzymes for genomic DNA digestion containing EGFR Deletions mutation in Exon 19, addition of AvaII did not significantly increase % desired ccfDNA fragments.
Figure 3: Comparison and justification for selection of the 3 restriction endonuclease enzymes for genomic DNA digestion with EGFR T790M/L858R mutations in Exons 20 ; 21, addition of HincII did not significantly increase % desired ccfDNA fragments.
Using working stock of reference DNA (at 3000 copies/µL) and 10x CutSmart Buffer (New England Biolabs) correct quantities of all reagents were calculated to result in final diluted master mixes, including enzymes, at concentration of ~50copies/µL. Components were mixed (before adding enzymes) by pipetting. For Deletions AvaII, ScrFI and HincII were added to the reaction mix and gently mixed by pipetting up and down. Mix was incubated at 37°C for 15 minutes, then at 80°C for 20 minutes to stop the reaction. The T790M and L858R were combined, EcoRI and Hpy1881 were added to the reaction and mixed gently by pipetting up and down. Mix was incubated at 37°C for 15 minutes, then ApoI added to the reaction and mixed gently by pipetting up and down. Finally this was incubated at 50°C for 15 minutes and then at 80°C for 20 minutes to stop the reaction. Digested DNA stored in -20oC freezing conditions.
DNA fragment analysis
The Agilent 2200 Tape Station was used for analysis of ccfDNA fragments created by restriction digestion. This platform provides simple, faster, more reliable gel electrophoresis. Using high sensitivity (HS) D1000 Tape, the gel images and DNA size peak analysis charts were used to assess the quality of restriction digested DNA (see results section). All operating procedures as outlined in the instructions/ user manual were followed. Samples were removed from freezer and thawed at room temperature for 30 minutes. Tape station reagents were removed from the fridge to equilibrate to room temperature for 30 minutes then tubes containing Tape Station reagents and samples were mixed by vortexing. All reagents and DNA samples to be analysed, including controls, were briefly centrifuged. 2µl of High Sensitivity D1000 DNA Ladder was added to the first well, 2µl of digested DNA samples and controls to the sample wells, with 2µl of HSD1000 Sample Buffer added to all (all Tape Station reagents supplied by Agilent).
Wells were covered with plate foil and mixed on a vortex at 2000 rpm for 1 min, the plate was briefly centrifuged to collect samples (important: check here there are no bubbles in any wells, remove with a pipette tip if present). The plate, the loading tips and the HS D1000 DNA screen tape were loaded into the Tape Station and the Tape Station operated as described in the MAN-SOP-44-01-002 (or manual).
10 healthy volunteer blood donors consented to give 80mL each of whole blood, collected into 8no. 10mL blood tubes, 4no. each of standard commercially available BD potassium-salt K2EDTA haematology blood storage tubes (in common use clinically) with a nominal concentration of 1.8mg EDTA per mL blood as an anti-coagulant, and 4no. each into PAXgene ccfDNA blood tubes from Pre-AnalytiX with a 1.5mL volume of proprietary preservation and stabilization solution. Blood tubes once filled were inverted 10 times to thoroughly mix fixative/preservative and immediately processed and logged, given unique MAN ID’s for traceability and transferred to a level 3 laboratory for storage and spiking with mutant EGFR DNA.
Blood spiking & storage
All work on potentially infectious material (blood/plasma) was carried out in a Class II microbiological safety cabinet within the pathogen suite at QIAGEN, including spiking, preparation and extractions. To mimic clinical samples all freshly drawn whole blood from healthy donors, collected into either PAXgene ccfDNA (cat. # 768115) or K2EDTA blood tubes (cat. # 367525), were spiked with EGFR mutant ccfDNA. Stocks of ccfDNA were made previously by enzymatic restriction digestion of human genomic DNA containing 3 x EGFR mutations (Deletions, T790M, L858R) and dilution to an estimated 50 copies/µl. This spike-in must be completed as soon as possible after blood drawing, ideally the same day, in this experiment all blood samples from all donors in all tube types were spiked within 2 hours of collection.
The Deletions mutant EGFR ccfDNA working stock was prepared separately to the T790M and L858R mutations, due to different enzymes required for digestion. The 2 working stocks were combined and diluted with nuclease free water by a dilution factor of 10, to reach concentration of ~5 copies of each mutation/µl. Each blood tube was spiked with 200µl of this solution, containing ~1000 copies of each EGFR mutation, then inverted to mix thoroughly, giving an approximate concentration of only 100 copies of each mutation per mL in the whole blood. This was in order to be close to the Limit of Detection (L.O.D) for these mutations, to thoroughly test the extraction methods being compared. LOD’s for this assay vary by mutation type, this will be discussed further in the Results section but range from /=1500bp, E5 = digested T790M ; L858R DNA with large band/peak of fragments ~200bp + spread of fragments 500-1000bp.
To compare the effectiveness of ccfDNA stabilisation between the two different blood tubes, as well as evaluating and validating the performance of the new QIAamp Minelute ccfDNA extraction kit against the current QIAamp Circulating Nucleic Acid extraction kit, eluates from each extraction at each time point were quantitatively amplified using real time quantitative PCR with the QIAGEN EGFR PCR kit. A control reaction determined if sufficient amplifiable ccfDNA was present in each sample and will be used in the analytical calculations to determine if mutations are present and detectable in the 3 separate assays with mutation-specific amplification refractory mutation system (ARMS) primers for the T790M, L858R and Deletions EGFR mutations. See Fig. 8 below for an overview of the mutation analysis process.
Figure 8: Flowchart for mutation analysis of Ct values for EGFR mutations generated by RT-qPCR amplification of ccfDNA samples
Ct values or cycle threshold values, are the number of cycles of PCR required for the fluorescent signal for each assay to cross a pre-defined threshold. Allele-specific amplification was achieved by ARMS, utilising thermostable Taq DNA polymerase (from Thermus Aquaticus bacterium) to distinguish between matched and mismatched bases at the 3′ end of a PCR primer. When the primer matches, amplification will proceed with full efficiency. When the 3′ base is mismatched, only low-level amplification will occur. Therefore, any mutated sequence was selectively amplified, even in samples where the majority of the ccfDNA did not carry the mutation(s). 2 replicates for each sample were analysed and averages taken.
Scorpions are bifunctional molecules containing a PCR primer covalently linked to a probe. The probe is composed of a fluorophore; Carboxyfluorescein (FAM) or Hexachlorofluorescein (HEX), and a quencher. The latter quenches the fluorescence of the fluorophore. When the probe binds to the ARMS amplicon during PCR, the fluorophore and quencher separate, leading to a detectable increase in fluorescence.
The FAM (excitation 492nM and emission of 517nm) signal was read by the green channel and HEX (excitation 535nM and emission of 556nm) signal was read by the yellow channel on the RGQ MDx. The FAM signal will be used to determine mutation detection, while the HEX signal was used to assess sample validity or amplification failure as an Internal Control, employing a non-EGFR related oligonucleotide target sequence, an unlabelled primer, and a HEX labelled Scorpions primer.
Table 3 shows FAM ?Ct value cut-offs for reliable detection of each specific EGFR mutation, as shown in Fig.7 the ?Ct for each PCR product was calculated by deducting the Control Ct from each Mutation Ct.
Table 3: EFGR RT-qPCR Mutation ?Ct Cut-off Values (?CT = mutation CT – control CT) for RGQ FAM/Green Channel
Ct values for all samples, for each assay, were automatically generated following input of threshold parameters into QIAGEN RGQ software Version 126.96.36.199, see table 4 below for analysis parameters.
Table 4: EFGR RT-qPCR software analysis parameters
Raw Ct values were generated by running qPCR on 2 repeats per sample, for each mutation assay, then these were averaged and averages added for all donors. The means of each were used for comparisons between tube types, time points and extraction methods, meaning each mutation Ct or ?Ct, for each tube type, at each time point, for each extraction kit is the average of ~20 data points or replicates, excluding some minor omissions due to invalid PCR run or invalid sample data, as well as some operator errors.
Figure 9 shows the average Ct values for the Control reaction (A), by tube type (orange = K2-EDTA tubes, blue = PAXgene ccfDNA tubes), by extraction kit (solid line = OLD, dashed line = NEW), over the 4 time points from 0-7 days. Average Ct values at T0 for EDTA tubes were 30.1 (old kit) and 30.58 (new kit) a nominal difference of only -0.42, for PAXgene tubes Ct values were 30.45 (old) and 31.32 (new) so a bigger variance was observed of 0.87. One issue with these initial Control Ct values was the PAXgene T0 and T1 average CT values were 31.32 and 31.42 respectively, where the kit threshold for these values is a maximum of 31.1. This will be discussed in more detail in the Discussion section but in consultation with the lead scientists on the project it was decided these values were acceptable, as ?Ct values for all the mutation assays fell within range. The higher control Ct indicated not enough amplifiable DNA was present in the samples, however, after amplification it was sufficient to successfully detect all mutations.
The Control Ct for PAXgene (old kit) dropped from 30.45 (Standard Deviation 0.525) at T0 to 30.35 at T7 (StDev 0.263), a variance of only -0.1 across all 4 time points showing a very stable and consistent yield of ccfDNA extracted. Control Ct for PAXgene (new kit) as mentioned ranged from 31.32 (StDev 0.567) at T0 to 30.78 (StDev 0.474) at T7, fairly stable with a total variance of -0.64 across all 4 time points. In the EDTA tubes an obvious difference became evident after just 3 days ambient storage. Both old and new extraction kits follow a very similar pattern, T0 Control Ct for each were 30.1 (old) and 30.58 (new), T1 29.94 and 30.2 respectively, from T3 Control Ct’s drop significantly indicating a much greater abundance of amplifiable DNA. The possible causes will be discussed later in this report, T3 Ct values were 28.32 (old) and 29.95 (new) but by T7 these had fallen to 26.72 and 27.22 respectively, a decrease in Ct of -3.38 (old) and 3.36 (new) which could mean up to an 8-10 X increase in background or WT DNA. The data was more variable for EDTA tubes, standard deviations were typically >0.5 (old) and >1 (new).
Figure 9 shows the three mutation assays in charts (B), (C) and (D) and the average ?Ct values for each tube type at each time point, by extraction kit. The black dotted lines indicate the ?Ct threshold cut-off for reliable mutation positive identification, as per Table 3. Any ?Ct values falling above the threshold lines are deemed ‘false negative’ as all samples contained the same original concentration of mutant DNA. It is worth noting the similarity in trend for both old and new kits for each tube type for (A)-(D), the orange and blue lines (tube type) whether solid (old kit) or dashed (new kit) exhibit little discrepancy from each other. Overall there is also a clear correlation between the observed drop in Control Ct value (A) for EDTA tubes and the increase in ?Ct values for each mutation assay (B, C, D), the relationship and possible root causes for these observations are reviewed in the following discussion section.
Figure 9: RT-qPCR average Ct values for Control Reaction (A) and average ?Ct values for 3 EGFR mutations, T790M (B), Deletions (C), L858R (D) for all time points (0, 1, 3 & 7 days) by tube type (PAXgene/blue or EDTA/orange) versus extraction kit (Old/solid vs New/dashed) showing significant drop in Control Ct after 3 days for EDTA tubes (A) and increased ?Ct for EDTA tubes after 3 days for each mutation; false negative mutation calls after 3 days for T790M (B) and L858R (D) and 7 days for Deletions (C)
For the T790M mutation (B) actual Ct values remained consistent, regardless of tube type, extraction method or time point. From the raw data, average Ct values for PAXgene tubes (old kit) varied by 1.23 cycles from a lowest Ct of 35.75 at T0 to 36.98 at T3 and for the PAXgene tubes (new kit) only by 0.46 cycles over the 7 days. This consistency matches the consistent Control Ct values for the PAXgene tubes and when ?Ct values were calculated a negligible increase is apparent from T0 to T7 but all fall well below the threshold value and were deemed as positively identifying the T790M mutant ccfDNA. By comparison the average Ct values for EDTA tubes had variance of 1.21 and 1.23 cycles for the old and new kits, ranges from 35.73-36.94 for the old kit and 36.21-37.44 for the new kit. The observed large shift in ?Ct values shown in chart (B) after 3 and 7 days is likely primarily due to the decrease in Ct for the Control reaction at these time points. ?Ct values for PAXgene tubes only varied by ~1 cycle, the ?Ct values for EDTA tubes exhibited much greater variation with ?Ct >10 at T7 for both old and new kits. The significant increase in ?Ct for EDTA tube samples at T3 and T7 led to false negative mutation calls.
Chart (C) for the Deletions EGFR mutation displays a similar pattern to (B), however, raw Ct data differs slightly, the Ct values overall were lower than those for both L858R and T790M by approximately 1 cycle across all tube types and time points for both extraction kits. Actual average Ct values for PAXgene tubes for the old kit ranged from 34.39 at T0 to 35.7 at T7, variance of 1.31 cycles, while for the new kit the variance was only 0.28, from 35.68 at T0 to just 35.95 at T7. EDTA tubes again had a greater variance in average Ct values for both old and new kits, 1.72 cycles difference from 34.75 at T0 to 36.47 at T7 and 1.87 cycles difference from 34.63 to 36.5 respectively. ?Ct values for PAXgene tubes ranged from 3.94 to 5.36 for the old kit and only varied by 0.88 cycles for the new kit from 4.36 up to 5.24 so all fell well below the mutation positive threshold and resulted in 100% accurate mutation calls for all time points. These lower average Ct values and slightly higher ?Ct threshold for this mutation assay combined resulted in accurate mutation positive calls for EDTA tubes, for both old and new kits, for T0 through to T3, but at T7 ?Ct values were 9.75 (old) and 9.64 (new) which gave false mutation negative data. Again a significant difference in ?Ct compared to PAXgene tubes and from T0 to T7 was evident, ?Ct for EDTA tubes increased from 4.65 to 9.75 for the old kit and rose from 4.37 to 9.64 for the new kit, variance >5 cycles.
Close similarities to (B) and to a lesser extent (C) continue for chart (D) in Fig. 9 for the L858R mutation. As shown ?Ct increases steadily for EDTA tubes for both old and new extraction kits from T0 to T7 while PAXgene ?Ct values increase only nominally from T0 to T3 and then actually drop slightly to T7 but overall remain fairly stable. The actual average Ct values closely match the data for chart (B), T790M, ranging from 36.21 up to 37.58 for PAXgene old kit, with a variance of 1.38 cycles and from 37.41 to 38.14 for the new kit, a difference of 0.73. ?Ct values for PAXgene start at 5.75 at T0 for the old kit and rise by 1.52 to 7.28 at T3 before dropping slightly to 6.57 by T7, for the new kit they are more consistent with a variance of only 1.03 cycles, from 6.14 at T0 up to 7.17 at T3 before decreasing at T7 to 6.48. All PAXgene tube samples at all time points for both old and new extraction kits provided 100% accurate mutation positive identification. This was not the case for the EDTA tubes, while average Ct values showed similar differences over the 4 time points to those for the L858R mutation, starting at 36.3 for the old kit and 36.17 for the new these increased by T7 to 37.73 and 37.7 respectively. The variance of around 1.5 cycles was greater than that of either kit for PAXgene tubes but less than the variance in ?Ct values, 4.84 and 4.85 for old and new. ?Ct values ranged from 6.2 to 11.03 for the old kit and from 5.59 to 10.44 for the new, which were the highest of any mutation and false negative calls were made from T3.
Fig. 10 below shows ?Ct values for each mutation assay, split between tube type (PAXgene on the left and EDTA on the right of each chart), with side by side comparison of the ?Ct values for the samples extracted using the old kit (hatched columns) as opposed to the new kit (solid columns) for each time point. Although the ?Ct values clearly increase for EDTA tubes over time, as shown in the previous Fig. 9, the comparison here was for the extracted yields of detectable ccfDNA using each kit.
The patterns were almost identical for all 3 mutation assays, (A) for Deletions, (B) for L858R and (C) for T790M. There are no significant differences between average ?Ct values for each mutation assay, for each tube type, at each time point. While ?Ct values increase gradually from T0 to T7 for EDTA tubes, the levels are comparable for each time point for each of the extraction methods. ?Ct values do increase slightly for the PAXgene tubes after T0, these level off at T3 or in the case of L858R (B) and T790M (C) appear to decrease slightly by T7, however, as mentioned above the variance is very low for these data.
One trend within the data was interesting and will be discussed further in the next section, the majority of average ?Ct values for PAXgene tubes show higher ?Ct values for the new kit when compared to the old kit, however, this trend is reversed for the EDTA tubes where ?Ct values are higher for old than the new. On average across the 3 mutation assays ?Ct values for PAXgene tubes were approximately 0.24 cycles greater for the new kit than for the old, whereas for EDTA tubes ?Ct values were approximately 0.27 cycles lesser for the new kit compared to the old.
Figure 10: RT-qPCR average ?Ct values for 3 EGFR mutations, T790M (A), Deletions (B), L858R (C) for all time points (0, 1, 3 & 7 days) by tube type (PAXgene or EDTA) versus extraction kit (Old vs New) showing equivalence between extraction methods
This study showed PAXgene ccfDNA stabilising blood storage tubes can both stabilise ccfDNA and help prevent lysis of other cells, thus releasing gDNA, for at least 7 days when stored at room temperature (15-25oC) and result in 100% accurate and reliable mutation identification even after 7 days storage at ambient temperature with the QIAGEN EGFR PCR Kit. Comparatively K2-EDTA blood tubes did not prevent cell lysis and gDNA release and do not appear to stabilise ccfDNA, after just 3 days storage at ambient temperature it appears the release of genomic WT DNA masked EGFR mutant ccfDNA signals leading to ‘false negative’ inaccurate EGFR mutation identification with the QIAGEN EGFR PCR Kit.
This indicates that PAXgene ccfDNA stabilising blood tubes are superior to K2-EDTA blood tubes for storing and stabilising liquid biopsies (blood), for periods of time longer than 1 day, at ambient storage temperatures, leading to improved reliability and accuracy for molecular biology and NGS analysis.
The NEW QIAGEN QIAamp ccfDNA MinElute Extraction Kit is directly comparable in terms of extracted ccfDNA yields with its predecessor the OLD QIAGEN QIAamp Circulating Nucleic Acid Extraction Kit, yet overall is quicker and easier to use, leading to labour savings of around 1 hour per 24 sample extraction.
The key study objective was met and the QIAamp ccfDNA Minelute Extraction Kit has been fully validated for use with the PAXgene ccfDNA stabilising blood collection tubes as an integral liquid biopsy CSP system and improved pre-analytical workflow, as part of WP3 of the STRATFix project. The work done by QIAGEN and other members of the SPIDIA and STRATFix consortiums will be invaluable for companion diagnostics in oncology going forward. Technologies developed such as the PAXgene ccfDNA stabilisation blood tubes, PAXgene Tissue and PAXgene FNA Fix will improve the pre-analytical phase of CSP systems for the major biological sample types used for cancer diagnosis, prognosis and monitoring. PAXgene blood tubes offer more reliable isolation and identification of ccfDNA even after storage, transport and sample handling across different sites, so will help ‘make improvements in life possible’.
Suggestions for further studies to build on this report would be utilisation of the exact specified equipment from the handbook for the QIAamp ccfDNA Minelute Extraction Kit, although it is considered unlikely this would have any impact on the results gathered during this study. Ensuring the 1.5ml additive in the PAXgene tubes is accounted for when calculating concentrations for spiking in future is important, although this is likely to improve on the data gathered in this study as this calculation error resulted in more diluted ccfDNA. A larger sample set would help in increasing the generalisability of these results, as the number of data points was not sufficient for rigorous statistical analysis. Also, to assist in wider generalisability repeating the study using actual clinical samples, as opposed to contrived samples as used, would be necessary to accurately determine the clinical validity and utility in a full clinical trial.