Collaborative study to evaluate the proposed WHO 1st international reference panel for genomic KRAS codons 12 and 13 mutations

WHO/BS/2017.2317 ENGLISH ONLY

EXPERT COMMITTEE ON BIOLOGICAL STANDARDIZATION Geneva, 17 to 20 October 2017 Collaborative study to evaluate the proposed WHO 1st International Reference Panel for genomic KRAS codons 12 and 13 mutations Pia Sanzone1, Ross Hawkins1, Eleanor Atkinson2, Peter Rigsby2, and Jennifer Boyle1,3 Divisions of Advanced Therapies1 and Biostatistics2, National Institute for Biological Standards and Control (NIBSC), Blanche Lane, South Mimms, Hertfordshire, EN6 3QG, United Kingdom. 3 Principal Investigator (email: jennifer.boyle@nibsc.org; telephone: +44 (0)1707 641000) NOTE: This document has been prepared for the purpose of inviting comments and suggestions on the proposals contained therein, which will then be considered by the Expert Committee on Biological Standardization (ECBS). Comments MUST be received by 18 September 2017 and should be addressed to the World Health Organization, 1211 Geneva 27, Switzerland, attention: Technologies, Standards and Norms (TSN). Comments may also be submitted electronically to the Responsible Officer: Dr M. Nübling at email: nueblingc@who.int © World Health Organization 2017

All rights reserved. Publications of the World Health Organization are available on the WHO web site (www.who.int) or can be purchased from WHO Press, World Health Organization, 20 Avenue Appia, 1211 Geneva 27, Switzerland (tel.: +41 22 791 3264; fax: +41 22 791 4857; e-mail: bookorders@who.int). Requests for permission to reproduce or translate WHO publications – whether for sale or for noncommercial distribution – should be addressed to WHO Press through the WHO web site: (http://www.who.int/about/licensing/copyright_form/en/index.html). The designations employed and the presentation of the material in this publication do not imply the expression of any opinion whatsoever on the part of the World Health Organization concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. Dotted lines on maps represent approximate border lines for which there may not yet be full agreement. The mention of specific companies or of certain manufacturers’ products does not imply that they are endorsed or recommended by the World Health Organization in preference to others of a similar nature that are not mentioned. Errors and omissions excepted, the names of proprietary products are distinguished by initial capital letters. All reasonable precautions have been taken by the World Health Organization to verify the information contained in this publication. However, the published material is being distributed without warranty of any kind, either expressed or implied. The responsibility for the interpretation and use of the material lies with the reader. In no event shall the World Health Organization be liable for damages arising from its use. The named authors alone are responsible for the views expressed in this publication.

WHO/BS/2017.2317 Page 2

Summary An international collaborative study assessed the suitability of a panel of genomic DNA (gDNA) materials as the proposed World Health Organization 1st International Reference Panel for genomic KRAS codons 12 and 13 mutations, NIBSC code 16/250, for use in the standardization of KRAS oncogene codons 12 and 13 mutation-based diagnostics. The panel comprised eight freeze-dried gDNA materials of the seven most-common colorectal cancer (CRC)-associated KRAS codons 12 and 13 mutations; NM_033360.3 (KRAS) c.35G>C (p.Gly12Ala, hereafter referred to as p.G12A; NIBSC material code 16/252), c.34G>T (p.Gly12Cys, p.G12C; 16/258), c.35G>A (p.Gly12Asp, p.G12D; 16/260), c.34G>C (p.Gly12Arg, p.G12R; 16/254), c.34G>A (p.Gly12Ser, p.G12S; 16/256), c.35G>T (p.Gly12Val, p.G12V; 16/264), c.38G>A (p.Gly13Asp, p.G13D; 16/262), plus a wild-type KRAS codons 12 and 13 material (16/266). Participants evaluated the materials using their routine diagnostic methods, and against in-house controls (previously characterized patient samples and cell line-derived gDNA) or commercial materials. Where possible, results were reported quantitatively in order to assign consensus values to each of the materials. Fifty six laboratories in thirty four countries performed sixty eight testing methods on the panel, of which thirty six reported quantitative data. Conclusions from this study indicated that all eight materials were suitable for use as reference materials in the genomic diagnosis of KRAS codons 12 and 13 mutations, with verified performance in next-generation sequencing (NGS), Sanger sequencing, real-time PCR, pyrosequencing, digital PCR (dPCR), Matrix Assisted Laser Desorption/Ionization-Time of Flight (MALDI-TOF) mass spectrometric analysis (MassARRAY®), KRAS StripAssay®, high resolution melt analysis (HRM), Amplification Refractory Mutation System-PCR (ARMS-PCR), PCR-Reverse Sequence Specific Oligonucleotide probe technique (PCR-rSSO), minisequencing, and restriction fragment length polymorphism analysis (RFLP). The proposed consensus mutation percentage for each material is derived from the median value of NGS and dPCR methods as: 65.7% KRAS p.G12A (16/252), 99.98% p.G12C (16/258), 71.5% p.G12D (16/260), 85.6% p.G12R (16/254), 99.7% p.G12S (16/256), 49.7% p.G12V (16/264), 66.9% p.G13D (16/262), and wild-type KRAS codons 12 and 13 (16/266). The collaborative study also analysed the response of these materials to dilution (with wild-type KRAS codons 12 and 13 material 16/266). These dilution data were used to calculate the consensus KRAS mutant and total copy number for each material. These consensus copy number data can be applied to a mathematical formula, with which the end-user may calculate how to prepare further standards at lower KRAS codons 12 and 13 mutation percentages from each of the seven mutant KRAS materials (by dilution with wild-type KRAS codons 12 and 13 material 16/266, or another wild-type gDNA aligned to 16/266). These materials and their dilutions will enable the calibration of assays, kits, and secondary standards for the seven most-common CRC-associated KRAS codons 12 and 13 mutations. All collaborative study participants agreed with the proposed genotype and consensus mutation percentage for each material, along with its consensus KRAS copy number data (and associated dilution formula), and approved the panel as the proposed WHO 1st International Reference Panel for genomic KRAS codons 12 and 13 mutations (NIBSC panel code 16/250).

WHO/BS/2017.2317 Page 3

Introduction Single nucleotide variants (SNVs) in the KRAS (Kirsten rat sarcoma 2 viral oncogene homolog) gene are present in approximately 30% of human cancers, and are particularly common in adenocarcinomas of lung, pancreas, and colon (COSMIC; Karnoub & Weinberg, 2008). Approximately 40% of CRCs are associated with KRAS mutations, with approximately 90% of these mutations occurring in codons 12 and 13 of KRAS exon 2 (COSMIC; Neumann et al., 2009). KRAS is a downstream component of the epidermal growth factor receptor (EGFR)-led RAS/MAPK (mitogen-activated protein kinase) signalling pathway, with EGFR regulating cell proliferation, apoptosis, and tumour-induced neoangiogenesis. KRAS mutations can lead to constitutive activation of KRAS (as GTP-bound KRAS), resulting in uncontrolled activation of downstream pathways. Anti-EGFR monoclonal antibodies (cetuximab and panitumumab) are available for the treatment of metastatic CRC, however, their treatment efficacy is limited to a subset of patients since EGFR-independent, constitutive activation of the RAS pathway impairs response to anti-EGFR treatment (Chan, 2015 and references within). Thus KRAS-activating mutations can predict resistance to anti-EGFR monoclonal antibody treatment, with KRAS mutation screening necessary prior to treatment (Amado et al., 2008; Douillard et al., 2013; Peeters et al., 2015). Whilst there are currently no therapies which directly target mutant KRAS, it is an active area of development (Ostrem et al., 2013; Zimmermann et al., 2013). Of all KRAS mutations reported in human tumours, by far the most frequent are KRAS c.35G>A (p.G12D), c.35G>T (p.G12V), c.38G>A (p.G13D), c.34G>T (p.G12C), c.35G>C (p.G12A), c.34G>A (p.G12S), and c.34G>C (p.G12R; COSMIC; Table 1). Each of these seven mutations is represented in the proposed WHO 1st International Reference Panel for genomic KRAS codons 12 and 13 mutations. Less frequent, CRC-associated mutations are not represented, including others in KRAS codons 12, 13, 59, 61, 117, and 146, and those in BRAF, PIK3CA, AKT1, SMAD4, PTEN, NRAS, and TGFBR2.

WHO/BS/2017.2317 Page 4

KRAS mutation p.G12D p.G12V p.G13D p.G12C p.G12A p.G12S p.G12R Other

Percentage total reported incidence (%) 33.6 22.7 12.5 11.2 5.4 4.5 3.1 7.0

Table 1. KRAS variant incidence. The percentage incidence of the most common KRAS variants were calculated from variant counts on COSMIC; the seven most common mutations are represented in the proposed WHO 1st International Reference Panel for genomic KRAS codons 12 and 13 mutations. The availability of KRAS codons 12 and 13 mutation primary standards should improve the quality of CRC genomic diagnostics by enabling the calibration of assays and kits, and the derivation of secondary standards for routine diagnostic use in determining testing accuracy and sensitivity, thus providing inter-laboratory comparison towards the harmonisation of KRAS codons 12 and 13 testing. The proposed WHO 1st International Reference Panel for genomic KRAS codons 12 and 13 mutations is intended as a panel of primary standards or calibrants in DNA-based genotyping of KRAS codons 12 and 13 mutations, and has been validated in this collaborative study encompassing NGS, Sanger sequencing, real-time PCR, pyrosequencing, dPCR, MassARRAY (Agena Bioscience, Hamburg, Germany), KRAS StripAssay (ViennaLab Diagnostics, Vienna, Austria; based on mutant-enriched PCR and reverse hybridization), HRM, ARMS-PCR, PCRrSSO, minisequencing, and RFLP methods. The panel comprises eight freeze-dried human gDNA materials produced from mutant and wild-type KRAS codons 12 and 13 cell lines, providing standards for the seven most-common KRAS codons 12 and 13 mutations, plus a wild-type KRAS control or diluent. The proposed consensus mutation percentages established from the international collaborative study involving 56 laboratories are the median values of all NGS and dPCR methods, the most condordant of all quantitative methods: 65.7% KRAS p.G12A (16/252), 99.98% p.G12C (16/258), 71.5% p.G12D (16/260), 85.6% p.G12R (16/254), 99.7% p.G12S (16/256), 49.7% p.G12V

WHO/BS/2017.2317 Page 5 (16/264), 66.9% p.G13D (16/262), and wild-type KRAS codons 12 and 13 (16/266). In expressing levels of somatic mutations in cancer, currently the overwhelmingly usual unit of clinical reporting is as percentage, and thus the consensus values for these reference materials are also reported as such (i.e. KRAS codon 12 or 13 mutant alleles as a percentage of total KRAS alleles). However, recently published data indicates the occurrence of KRAS allelic heterogeneity and mutant allele copy number gains in tumours (Birkeland et al., 2012; Kerr et al., 2016; Mekenkamp et al., 2012; Sasaki et al., 2011). It is therefore probable that the use of ‘percentage’ to define mutational load is simplistic, with ploidy and gene copy number variation potentially meaning that samples with very different KRAS mutation content could be characterized as having similar mutation percentages. It is considered that the use of tumour-derived cell lines closely mimics this in vitro tumour genomic heterogeneity, thus achieving some commutability. The collaborative study was able to derive mutant and total consensus KRAS copy numbers in the mutant KRAS materials, based upon their response to dilution with the wild-type KRAS codons 12 and 13 material 16/266; this information will also be provided to end-users. Moreover, a proposed dilution formula (based upon the calculated KRAS copy numbers) determines how the end-user may prepare standards at lower calculated KRAS codons 12 and 13 mutation percentages for each mutant material (by dilution with wild-type KRAS codons 12 and 13 material 16/266, or another wild-type gDNA aligned to 16/266). The use of multiple standards at a range of mutation percentages, and for each of the seven variants, will enable assay calibration across a wide mutation percentage range. A total of 2,057 panels of the proposed WHO 1st International Reference Panel for genomic KRAS codons 12 and 13 mutations (panel code 16/250) are available from the National Institute for Biological Standards and Control (NIBSC, UK). These standards are intended for use in in vitro diagnostics and relate to BS EN ISO 17511:2003 Section 5.5.

Aims of the Collaborative Study The study evaluated the panel of eight freeze-dried gDNA materials each of a different KRAS codons 12 and 13 genotype in an international collaborative study involving laboratories using a variety of diagnostic genotyping techniques, thereby assessing the panel’s suitability as the WHO 1st International Reference Panel for genomic KRAS codons 12 and 13 mutations, for use as a primary reference material in the calibration of secondary standards, kits, and assays. All data were used to establish the genotypes; quantitative data were used to establish a consensus KRAS codons 12 or 13 mutation percentage for each of the materials. The materials were also evaluated at several dilutions (each material diluted in the nominal wild-type KRAS codons 12 and 13 material 16/266). These data were used to derive consensus mutant and total KRAS copy numbers for each of the mutant materials, and to establish a formula which determines how a dilution should be performed (with the wild-type KRAS codons 12 and 13 material 16/266, or another wild-type gDNA aligned to 16/266) to generate standards at any specified lower KRAS codons 12 and 13 mutation percentage.

WHO/BS/2017.2317 Page 6

Candidate Materials Eight materials were evaluated as candidates for the proposed WHO 1st International Reference Panel for genomic KRAS codons 12 and 13 mutations. All materials were of freeze-dried, purified gDNA extracted from eight cell lines of either mutant or wild-type KRAS codons 12 and 13 genotypes. Materials were freeze-dried in glass ampoules as an established format for ensuring long-term stability of gDNAs. Ideally, the formulation for reference materials should be as close as possible to the usual patient analyte, cover the entire analytical process, and be applicable to methods in use throughout the world. However, it is essential that the formulation be stable for many years, and that it is practically possible to produce batches of sufficient size to satisfy demand over a similar period of time. Additionally, it should ideally be possible to generate replacement standards from the same source material to ensure consistency in formulation and to minimize value drift. It would be impossible to obtain sufficient primary patient material to produce mutant KRAS codons 12 and 13 reference materials at both high quality and sufficient quantity. Also, since the process of DNA extraction from cultured cells is different from that of solid tissue (typically formalin-fixed paraffin-embedded (FFPE) sections) or blood (if the analyte is circulating tumour DNA (ctDNA)), providing the materials as cultured cells would not provide standardization for this step of the process in the most optimal way. Materials were provided as high molecular weight gDNA rather than the fragmented DNA often obtained from FFPE sections, or present as ctDNA, due to stability concerns and the intent to provide materials applicable to potentially any substrate used for KRAS mutation detection. The nominal wild-type KRAS ATDB102 lymphoblastoid cell line was established at NIBSC following Epstein-Barr virus (EBV) transformation of isolated monocytes from a whole blood sample provided by a consenting healthy donor, and was confirmed as having a diploid genomic content by karyotyping (data not shown). The nominal mutant KRAS codons 12 and 13 human cancer cell lines were derived from patient tumour tissue and obtained from the European Collection of Cell Cultures (ECACC; Public Health England, Salisbury, UK; Table 2). All cell lines were tested and found negative for HIV1, HTLV1, Hepatitis B, and Hepatitis C by PCR; master and working cell banks were produced in-house to ensure a continual future cell supply. Large-scale cell culture was carried out (in-house and at ECACC), and frozen cell pellets at 1 x 108 cells prepared. Genomic DNA was extracted from the cell pellets using Gentra Puregene chemistry with a Gentra Autopure LS robot (Qiagen, Manchester, UK). The DNA extraction process involved RNAse treatment, protein denaturation, protein removal, and 70% ethanol washing. The use of 70% ethanol is an established method for viral inactivation (Roberts & Lloyd, 2007). Additionally, gDNA extracted in-house using the same purification procedure from other EBV-transformed cell lines did not show EBV infectivity (Hawkins et al., 2010). However, these materials should be handled with care, and according to local laboratory safety precautions for biological materials. Many tumours exhibit high levels of genomic instability, including mutation mosaicism and variability in gene copy number, zygosity, and overall ploidy. Cell lines derived from tumours are believed to provide a snapshot of the tumour at the time of biopsy (Lansford et al., 1999),

WHO/BS/2017.2317 Page 7 with evidence to support this including data from histopathology, molecular genetics, receptor expression, gene expression, and drug sensitivity (Masters, 2000). However, it is unclear as to what extent variability continues to occur within the cell line over time. Overall it is expected that the materials used in this study are a useful mimic to the in vivo genomic complexity and variability of a tumour sample, and thus some commutability is achieved. Furthermore, since these materials are each prepared as a large batch, they are a long-term source of an unchanging genomic content. KRAS codon 12 and 13 genotypes for each of the eight materials were indicated by COSMIC and/or in-house droplet dPCR (ddPCR, BioRad, Hercules, CA, USA). Droplet dPCR was also used to confirm the presence of two copies of the wild-type KRAS allele in the ATDB102 cell line, which when used in the dilution of the mutant KRAS materials (as material 16/266) enabled the calculation of the KRAS allelic ratio (mutant: wild-type) and total KRAS copy number (mutant plus wildtype) in each of the mutant materials (see KRAS Copy Numbers: Establishment of a Dilution Formula, below). Each of the gDNA materials was prepared at approximately 10 µg/ml DNA concentration in 2.0 mM Tris, 0.2 mM EDTA, with 5 mg/ml D-(+)-trehalose dehydrate (Sigma-Aldrich, St. Louis, MO, USA; Table 3). Aliquots of 0.5 ml (1.0 ml for nominal wild-type KRAS material 16/266) were dispensed into 3 ml autoclaved DIN glass ampoules (Schott, Pont-sur-Yonne, France) using an automated AFV5090 ampoule filling line (Bausch & Strobel, Ilfshofen, Germany) with the bulk continually stirred at a slow rate using a magnetic stirrer whilst at ambient temperature. The homogeneity of the fill was determined by on-line check-weighing of the wet weight of triplicate ampoules for every 90 ampoules filled, with ampoules outside the defined specification (0.5000 g to 0.5300 g; 1.0000 g to 1.0150 g for material 16/266) discarded. The ampoules were partially stoppered with 13 mm Igloo stoppers (West, St Austell, UK) before the materials were freezedried in a CS15 (Serail, Argenteuil, France) to ensure long-term stability: the ampoules were frozen to -50°C, with primary drying at -35°C, 50 µbar, for 30 hours (-40°C, 30 µbar, for 30 hours for material 16/266), followed by secondary drying at +30°C, 30 µbar, for 40 hours. The vacuum was then released and the ampoules back-filled using boil-off gas from high purity liquid nitrogen (99.99%), before stoppering in situ in the dryer and flame sealing of the ampoules. Measurement of the mean oxygen head space after sealing served as a measure of ampoule integrity. This was measured non-invasively by frequency modulated spectroscopy (FMS 760, Lighthouse Instruments, Charlottesville, VA, USA), based upon the Near Infra-Red absorbance by oxygen at 760 nm when excited using a laser. Controls of 0% and 20% oxygen were tested before samples were analysed to verify the method. Twelve ampoules were tested at random from each material; oxygen should be less than 1.14%. Residual moisture content was measured for the same 12 ampoules per material using the coulorimetric Karl Fischer method in a dry box environment (Mitsubishi CA100, A1 Envirosciences, Cramlington, UK) with total moisture expressed as a percentage of the mean dry weight of the ampoule contents. Individual ampoules were opened in the dry box and reconstituted with approximately 1-3 ml Karl Fischer analyte reagent which was then injected back into the Karl Fischer reaction cell and the water present in the sample determined colourmetrically. Dry weight was determined for six ampoules per material weighed before and after drying, with the measured water expressed as a percentage of

WHO/BS/2017.2317 Page 8 the dry weight. Residual moisture levels of less than 1% are typically obtained, but where the dry weight is low (as here) the moisture level can be higher, with the materials still expected to demonstrate long-term stability (as seen for the similarly prepared WHO 1st International Genetic Reference Panel for Prader Willi & Angelman Syndromes, NIBSC panel code 09/140, which continues to demonstrate high stability eight years post-manufacture). Ongoing stability will be confirmed by accelerated degradation studies (see Degradation Studies, below). Upon reconstitution with 100 µl nuclease-free water, the DNA concentration was approximately 50 µg/ml in 10 mM Tris, 1 mM EDTA (1x TE buffer) with 25 mg/ml D-(+)-trehalose dehydrate, excepting nominal wild-type KRAS codons 12 and 13 material 16/266 which was reconstituted with 200 µl nuclease-free water to give a DNA concentration of approximately 125 µg/ml in 1x TE buffer with 25 mg/ml D-(+)-trehalose dehydrate (for further dilution with 1x TE buffer to achieve 50 µg/ml DNA concentration). Homogeneity of each fill was determined by analysis of ampoules from the beginning, middle, and end of the filling process with quality and quantity of the freeze-dried gDNAs confirmed by 260/280 nm absorbance (Nanodrop, Thermo Fisher Scientific, Wilmington, DE, USA), Qubit fluorometric DNA quantification (Thermo Fisher Scientific), TapeStation electrophoresis (Agilent, Santa Clara, CA, USA), and ddPCR, which also acted as a pilot study to determine the performance of the materials in this increasingly-used diagnostic technique (Table 3). Lower DNA integrity number (DIN) was noted for materials 16/252, 16/258, and 16/266, as compared with the other materials, although the DNA quality was still within the acceptable range. Microbiological results were negative for all eight materials. The ampoules are stored at -20°C at NIBSC under continuous temperature monitoring for the lifetime of the product. Shipping will typically be at ambient temperature, as studies have indicated the retained stability of the materials at elevated temperatures (5 months at +56°C, see Degradation Studies, below).

NIBSC code

Nominal KRAS codon 12 or 13 mutation p.G12A p.G12C p.G12D p.G12R p.G12S p.G12V p.G13D

Originating cell line

Human tissue source myeloma (blood) pancreatic carcinoma lung adenocarcinoma pancreatic adenocarcinoma lung carcinoma ovarian metastasis of primary colon adenocarcinoma supraclavicular lymph node metastasis of colon adenocarcinoma EBV-transformed lymphocytes (blood)

16/252 16/258 16/260 16/254 16/256 16/264 16/262 16/266

RPMI 8226 MIA-Pa-Ca-2 SK LU 1 PSN1 A549 SW 626 LoVo ATDB102

Wild-type

WHO/BS/2017.2317 Page 9 Table 2. Source cell lines of the eight materials of the proposed WHO 1st International Reference Panel for genomic KRAS codons 12 and 13 mutations. KRAS codon 12 and 13 genotypes were indicated by COSMIC and/or in-house ddPCR. NIBSC code Nominal KRAS codons 12 and 13 mutation Date filled Mean DNA concentration upon filling (µg/ml; n=15 to 21) Mean fill mass (g; n=78 to 133) Mean pH (n= 10-14) Coefficient of variation of fill mass (%; n= 78 to 133) Mean dry weight (g; n= 6) Coefficient of variation of dry mass (%; n= 6) Mean residual moisture after lyophilisation (%; n= 12) Coefficient of variation of residual moisture (%; n= 12) Mean residual oxygen (%; n= 12) Coefficient of variation of residual oxygen (%; n= 12) Mean DNA concentration upon reconstitution (µg/ml; n= 3) Mean OD ratio (260/280 nm; n= 3) Mean 16/252 p.G12A 30/09/16 11.04 (~5µg total; 21) 0.5246 (99) 7.0 (14) 0.33 (99) 0.002 11.00 16/258 p.G12C 24/11/16 11.39 (~5µg total; 21) 0.5175 (107) 6.5 (14) 0.99 (107) 0.002 5.10 16/260 p.G12D 28/10/16 10.74 (~5µg total; 21) 0.5254 (109) 7.0 (14) 0.48 (109) 0.002 6.43 16/254 p.G12R 30/9/16 12.93 (~5µg total; 21) 0.5251 (81) 7.0 (14) 0.56 (81) 0.002 4.47 16/256 p.G12S 14/10/16 10.98 (~5µg total; 15) 0.5238 (113) 6.5 (10) 0.87 (113) 0.002 4.01 16/264 p.G12V 14/10/16 10.45 (~5µg total; 21) 0.5261 (102) 6.5 (14) 0.22 (102) 0.002 4.15 16/262 p.G13D 28/10/16 11.87 (~5µg total; 21) 0.5233 (133) 6.5 (14) 0.93 (133) 0.002 3.57 16/266 Wildtype 10/11/16 28.11 (~25µg total; 18) 1.0084 (78) 7.0 (12) 0.15 (78) 0.005 1.89

4.10346

2.13715

2.62897

2.96266

2.55461

2.56359

2.48148

1.90700

27.64

30.10

24.44

24.14

12.88

38.37

25.14

29.48

0.44

0.49

0.35

0.54

0.40

0.28

0.40

0.64

24.41

24.29

28.54

21.52

31.23

45.65

50.96

22.89

51.80

60.30

57.40

60.00

57.80

57.20

68.07

118.67

1.91 6.9

1.92 7.0

1.93 8.7

1.73 9.1

1.91 9.4

1.90 9.1

1.93 9.1

1.89 7.9

WHO/BS/2017.2317 Page 10 TapeStation DIN (n= 3) Mean KRAS mutation % (ddPCR; n= 9) Coefficient of variation of KRAS mutation % (%; n=9) Number of ampoules available Presentation Excipient Address of facility where material was processed Present custodian Storage temperature

67.00

99.97

71.53

85.80

99.95

51.47

67.90

0.00

0.39

0.02

0.43

0.55

0.05

1.43

0.77

N/C

2547

2526

2574

1997

2100

2572

2328

2057

Sealed, glass DIN ampoules, 3 ml Trehalose, 5 mg/ml in 2.0 mM Tris, 0.2 mM EDTA buffer

NIBSC, South Mimms, Hertfordshire, UK

NIBSC, South Mimms, Hertfordshire, UK -20oC

Table 3. Production and testing summary of the eight materials of the proposed WHO 1st International Reference Panel for genomic KRAS codons 12 and 13 mutations. N/C, not calculated as all values were zero.

Participants Sixty laboratories were recruited to the collaborative study, predominantly through membership of the European Molecular Genetics Quality Network (EMQN; Manchester, UK) and UK National External Quality Assessment Service (UK NEQAS) for Molecular Genetics (Edinburgh, UK), and also publications on KRAS genomic diagnostics, and personal contacts, ensuring maximal coverage of the principal KRAS codons 12 and 13 mutation diagnostic techniques. Four laboratories were unable to proceed with the study, either due to import constraints (n=1) or limited laboratory resources (n=3); all remaining fifty six laboratories participated in the study and returned data (Appendix I). Thirty four countries were represented by the participants returning results, encompassing Europe, Asia, North America, South America, and Australia. Each laboratory was assigned a code number (1 to 60) which does not reflect the order of listing in Appendix I. Where laboratories submitted data from more than one method, each method is referred to by an alphabetical suffix, for example 2a and 2b for laboratory 2 methods a and b. Data from a total of 68 methods were returned, including 36 quantitative data sets.

WHO/BS/2017.2317 Page 11

Study Design Triplicate coded samples of the panel of eight gDNA materials (n= 24) were sent to each laboratory with instructions for reconstitution and storage. Overall the materials were each to be tested at four different dilutions (crude, 1:2, 1:2.5, and 1:5), by dilution with the nominal wildtype KRAS codons 12 and 13 material 16/266 (total n=96). However, since it was not reasonable to request each laboratory analyse such a high number of samples, the materials and their dilutions were distributed amongst the participants based upon their method and reported assay sensitivity; details of the collaborative study design are provided in Appendices II and III. Participants were asked to perform their routine testing method(s) for the investigation of KRAS codons 12 and 13 mutations by testing the 24 coded samples in groups of 8 at a single dilution (n=8), plus 1 of those samples at an additional dilution (n=1), on 3 separate days (in total, n=27). Participants were requested to use different batches of reagents and/or different operators if possible, alongside in-house patient samples (or other control materials) if typically used. Laboratories were asked to report quantitative results where possible, together with the clinical interpretation. Overall findings for each sample and raw data, for example Sanger sequencing traces or dPCR counts, were to be returned, together with full details of the techniques used, any reference samples used, and reasons for failure of any of the samples tested.

Collaborative Study Methods Twelve principal methods were used by the fifty six participants of the collaborative study (Table 4). Ten laboratories used two methods or method modifications (laboratories 2, 15, 20, 24, 27, 30, 40, 52, 55, and 58), whilst laboratory 54 used three method modifications, thereby giving a total of sixty eight methods. Quantitative data were reported for 36 methods; 32 methods reported qualitative data only. Following the distribution of the collaborative study report, laboratory 13 withdrew their data from the study (see Comments from the Participants, below); subsequent analyses are of data from 55 participants, and 67 methods, of which 35 were quantitative. Details of each method are provided in Appendix IV.

WHO/BS/2017.2317 Page 12 Principal Testing Method Next-generation sequencing 1 7 4 25 29 Participating Laboratory/Method Number 30b 33 34 38 43 48 50 54a 54b 54c 58a 58b Total 17

Sanger sequencing

8

16

18

20b

27b

28

30a

31

32

40b

42

44

53

13

Real-time PCR

2b

3

9

17

19

24b

26

27a

40a

46

51

11

Pyrosequencing

5

6

10

23

37

49

59

7

digital PCR MALDI-TOF mass spectrometry (MassARRAY®) KRAS StripAssay®

60

21

52b

55b

4

2a

11

13

22

4

24a

41

47

3

High Resolution Melt analysis

15a

20a

2

ARMS-PCR

35

55a

2

PCR-rSSO

39

52a

2

Minisequencing

36

56

2

RFLP

15b

1

Overall total

68

Table 4. Methods used by collaborative study participants. Principal testing methods used in the detection of KRAS codons 12 and 13 mutations by participants in the collaborative study. Methods reporting quantitative data are in bold (n=36, including three MassARRAY methods which were noted to be semi-quantitative; green); all other methods reported qualitative data (n=32). Method 30a was HRM followed by Sanger sequencing; method 31 comprised wild-type allele-clamped PCR followed by Sanger sequencing of the amplified mutant allele; method 32 comprised peptide nucleic acid (PNA) wild-type allele-clamped PCR followed by Sanger sequencing; method 53 utilized selective cleavage of wild-type amplicons with restriction endonucleases followed by Sanger sequencing (orange). Real time PCR methods 17, 40a, and 51 utilized the IdyllaTM KRAS mutation test (Biocartis, Mechelen, Belgium) which includes an integrated sample preparation method (yellow). Following the distribution of the collaborative study report, laboratory 13 withdrew their data from the study (red).

WHO/BS/2017.2317 Page 13

Results Expected Results The 24 blinded materials comprised triplicate samples of each of the 8 materials; NIBSC material code 16/252 (nominal mutation KRAS p.G12A), 16/258 (p.G12C), 16/260 (p.G12D), 16/254 (p.G12R), 16/256 (p.G12S), 16/264 (p.G12V), 16/262 (p.G13D), and 16/266 (wild-type KRAS codons 12 and 13; Table 5). The nominal KRAS codons 12 and 13 genotypes were indicated by COSMIC and/or in-house ddPCR. Each material was tested at a single dilution (crude, 1:2, 1:2.5, or 1:5; n=24), plus one material was tested at an additional dilution (crude, 1:2, 1:2.5, or 1:5; n=3), distributed across three separate days (in total, n=27). The expected mutation percentages for each material/dilution are not shown as these were to be determined by the collaborative study.

WHO/BS/2017.2317 Page 14

NIBSC code Nominal KRAS codons 12 and 13 mutation Sample 1 Sample 2 Sample 3 Sample 4 Sample 5 Sample 6 Sample 7 Sample 8 Sample 9 Sample 10 Sample 11 Sample 12 Sample 13 Sample 14 Sample 15 Sample 16 Sample 17 Sample 18 Sample 19 Sample 20 Sample 21 Sample 22 Sample 23 Sample 24

16/252

16/258

16/260

16/254

16/256

16/264

16/262

16/266

p.G12A 

p.G12C

p.G12D

p.G12R

p.G12S

p.G12V

p.G13D

Wildtype

                      

Table 5. Collaborative study expected results. Participants tested 24 blinded samples which comprised triplicates of the 8 materials of the different nominal KRAS codons 12 and 13

WHO/BS/2017.2317 Page 15 genotypes. Each of these materials was tested as crude, 1:2, 1:2.5, or 1:5 diluted with the nominal wild-type KRAS codons 12 and 13 material (16/266); expected mutation percentages are not shown as they were to be established by the collaborative study.

Results returned by Participants Quantitative Data Quantitative data were reported for 35 methods including NGS, Sanger sequencing, real-time PCR, pyrosequencing, dPCR, and MassARRAY (a semi-quantitative method), and the means of the triplicate sample values for each method and material calculated as mutation percentage (for full data see Appendix V). Particular observations were: a. method 1 reported KRAS p.G12A in sample 2 of material 16/254 (at 1:2.5 dilution); b. method 7 reported 45% KRAS p.G12S for sample 23 of material 16/256 (at 1:5 dilution), which was considered an outlier and excluded from further analysis; c. method 29 additionally reported STK11 p. P281L and p.F354L in all three samples of material 16/254 (at 1:5 dilution), STK11 p.Q37* in all three samples of material 16/256 (at 1:5 dilution), and EGFR p.R836R in all three samples of material 16/264 (at crude and 1:5 dilution); d. method 33 reported 15% KRAS p.G12S for sample 14 of material 16/256 (at 1:2 dilution), which was considered an outlier and excluded from further analysis, and additionally reported IDH2 c.435delG (p.T146fs) in all three samples of material 16/262 (crude); e. method 43 was unable to test samples 11 and 18 of material 16/262 at 1:2.5 dilution, due to resource availability; f. method 58a reported KRAS p.G13D in sample 17 of material 16/260 (at 1:5 dilution); g. method 46 reported >75% KRAS p.G12A for sample 1 of material 16/252 (at 1:2.5 dilution), >50% KRAS p.G12C for sample 20 of material 16/258 (at 1:2.5 dilution), and unknown percentage KRAS p.G12S for sample 3 of material 16/256 (at 1:2.5 dilution), all likely due to the absence of high percentage controls for quantification in these assays; h. method 6 was unable to quantify KRAS c.G34 mutations, and thus the three samples of material 16/258 were reported as KRAS p.G12C of unknown percentage (at 1:5 dilution), the three samples of material 16/254 were reported as KRAS p.G12R of unknown percentage (at 1:5 dilution), and the three samples of material 16/256 were reported as KRAS p.G12S of unknown percentage (at 1:5 dilution); i. method 55b did not perform dPCR analysis (and therefore quantification) of materials 16/258 (crude), 16/254 (at 1:5 dilution), 16/256 (at 1:5 dilution), or 16/262 (at 1:5 dilution), although genotypes were identified in the laboratory’s other method, 55a. Quantitative data from real-time PCR analysis were clearly distinct from those of other quantitative methods, were reported by only two methods (methods 3 and 46), and thus were excluded from further analysis to avoid the influence of these outlying data. It was furthermore apparent that data from all remaining quantitative methods (n=33) could not be considered as one

WHO/BS/2017.2317 Page 16 dataset as there were clear differences between some of the principal methods’ results, meaning that the derivation of overall average values for each material would likely not be in agreement with any one method. Therefore, quantitative data from NGS, Sanger sequencing, pyrosequencing, dPCR, and MassARRAY were considered separately (Table 6; Figure 1). Summary statistics for each principal method and material were calculated as mean and median (Table 7). As there were insufficient data to confirm the assumption of a normal distribution of results within all methods for each material (see Figure 1), and to avoid the influence of any outliers, the median value was used as an appropriate summary statistic for each material (and dilution).

WHO/BS/2017.2317 Page 17 Material Dilution crude 1:2 1:2.5 1:5 crude 1:2 1:2.5 1:5 crude 1:2 1:2.5 1:5 crude 1:2 1:2.5 1:5 crude 1:2 1:2.5 1:5 crude 1:2 1:2.5 1:5 crude 1:2 1:2.5 1:5 Principal method, Laboratory/method code Sanger Pyroseq dPCR MassARRAY 1 7 4 25 29 30b 33 34 43 48 50 54a 54b 54c 58a 58b 16 28 31 5 6 10 23 37 49 60 21 52b 55b 2a 11 22 65.0 65.0 65.7 66.0 66.9 65.9 63.9 63.3 64.3 67.0 67.0 53.4 57.0 31.3 30.3 31.7 43.3 33.6 31.3 31.9 23.7 26.6 29.1 28.3 23.7 26.3 28.1 25.5 27.4 10.7 13.0 12.1 14.0 12.7 12.7 10.7 12.3 14.7 15.0 12.4 16.0 17.7 100.0 100.0 99.7 99.9 99.4 100.0 95.0 97.1 95.0 100.0 100.0 59.4 58.7 53.7 55.9 80.0 59.9 64.0 59.1 57.3 46.3 37.5 53.2 48.7 47.0 55.5 49.2 42.9 61.3 28.7 29.2 24.7 42.0 29.0 20.7 22.0 42.3 29.4 29.0 48.3 44.7 NGS 38 71.0 40.3 26.7 85.0 72.0 62.0 99.7 42.7 21.5 46.3 20.7 15.0 66.7 30.3 20.7 33.2 20.3 37.5 17.3 46.1 21.4 51.0 23.1 24.9 11.0 12.0 14.3 67.7 66.9 39.9 31.8 26.7 41.4 21.0 49.7 49.9 26.0 26.7 12.0 18.0 67.7 36.7 20.3 17.0 11.7 10.7 66.1 66.2 67.7 80.0 50.0 37.0 27.0 30.3 43.4 39.2 15.0 49.3 47.1 47.4 73.5 58.2 100.0 99.8 46.5 71.6 57.6 52.6 43.3 22.0 19.3 40.0 40.0 27.5 30.0 29.0 8.8 66.5 17.7 50.4 25.0 54.3 36.2 26.2 45.2 24.3 86.7 85.6 85.0 78.1 69.7 59.3 51.7 99.9 99.6 100.0 99.3 98.0 100.0 43.3 36.0 17.7 54.2 43.8 36.3 80.9 86.5 86.0 72.3 71.5 46.6 71.5 71.7 70.8 49.7 42.0 26.7 19.3 85.0 78.3 62.0 58.0 100.0 99.0 75.5 80.0 61.7 46.0 25.7 54.4 31.7 89.0 77.9 70.3 50.7 40.7 32.9 71.0 74.8 74.3 71.5 51.6 41.8 25.7 83.0 85.8 75.6 71.9 58.4 93.7 99.9 50.0 41.7 22.1 43.7 51.5 29.9 23.8 13.7 63.7 67.9 43.3 35.7 20.0 69.8 46.9 53.8 72.2 75.1 49.2 40.3 43.0 99.5 52.2 40.2 28.0 23.7 24.0 50.3 27.6 25.4 37.5 12.5 24.7 26.5 53.6 40.7 40.4 33.3 33.0 24.2 45.3 41.7 63.3

16/252

16/258

16/260

16/254

16/256

16/264

16/262

Table 6. Summary data for quantitative methods in the collaborative study. Mean mutation percentages for triplicate tested samples in all quantitative methods were derived (n=33 methods; excluding real-time PCR methods 3 and 46). Method 7 reported 45% KRAS p.G12S for one sample of material 16/256 at 1:5 dilution, which was considered to be an outlying value (compared with the other two samples of 20% and 23%), and so was excluded from the mean calculation (orange); method 33 reported 15% KRAS p.G12S for one sample of material 16/256 at 1:2 dilution, which was considered to be an outlying value (compared with the other two samples of 47% and 46%), and so was excluded from the mean calculation (orange); method 43 tested only one sample of material 16/262 (at 1:2.5 dilution; orange). Results are reported to one decimal place.

WHO/BS/2017.2317 Page 18

Figure 1. Mean KRAS codons 12 and 13 mutations percentage values for triplicate tested materials in all quantitative methods in the collaborative study (n=33 methods; excluding quantitative real-time PCR methods 3 and 46). Data are shown for each material at crude, 1:2, 1:2.5, and 1:5 dilutions for each quantitative method.

WHO/BS/2017.2317 Page 19

Material Dilution crude 1:2 1:2.5 1:5 crude 1:2 1:2.5 1:5 crude 1:2 1:2.5 1:5 crude 1:2 1:2.5 1:5 crude 1:2 1:2.5 1:5 crude 1:2 1:2.5 1:5 crude 1:2 1:2.5 1:5

16/252

16/258

NGS 65.5 30.8 26.9 12.3 99.8 56.9 46.4 28.0 71.5 48.1 40.9 26.6 85.1 78.1 71.7 57.2 99.5 49.5 43.4 20.0 48.7 26.0 23.8 13.1 67.0 39.9 33.9 20.4

Sanger 63.3 37.5 23.7

Principal method mean Pyro 65.7 33.6 27.2 13.5 95.0 96.1 80.0 62.0 47.0 55.5 80.0 61.7 46.0 25.7 85.0 78.3 62.0 42.3 73.4 54.4 50.7 35.1 86.0 76.7 70.3 58.0 97.6 54.2 43.8 17.7 47.0 27.5 25.0 13.3 65.1 43.4 39.2 28.7

dPCR 60.2 31.6 25.5 13.7 100.0 58.2 46.1 29.2 70.7 49.2 41.8 24.9 79.0 75.4 71.9 58.4 99.7 51.1 41.7 22.1 50.9 27.6 23.8 13.1 60.8 42.0 35.7 20.0

MassARRAY 57.0 27.4 16.8

61.3 46.5

NGS 65.7 30.8 27.5 12.7 100.0 57.3 47.5 28.7 71.5 48.1 41.2 26.4 85.6 78.1 71.8 57.9 99.7 49.5 43.0 21.2 49.3 26.0 24.0 12.0 66.9 39.9 33.2 20.3

Principal method median Sanger Pyro dPCR 63.3 65.7 60.2 37.5 33.6 31.6 23.7 27.2 25.5 13.5 13.7 95.0 96.1 100.0 80.0 62.0 58.2 47.0 55.5 46.1 80.0 61.7 46.0 25.7 85.0 78.3 62.0 100.0 43.3 36.0 40.0 40.0 29.5 80.0 50.0 37.0 42.3 74.3 54.4 50.7 32.9 86.0 76.7 70.3 58.0 99.0 54.2 43.8 17.7 47.0 27.5 25.0 13.3 65.1 43.4 39.2 28.7 29.2 70.7 49.2 41.8 24.9 79.0 75.4 71.9 58.4 99.7 51.1 41.7 22.1 50.9 27.6 23.8 13.1 60.8 42.0 35.7 20.0

MassARRAY 57.0 27.4 16.8

61.3 46.5

NGS -0.2 0.0 -0.5 -0.4 -0.1 -0.4 -1.1 -0.6 0.0 0.0 -0.2 0.2 -0.5 0.0 -0.1 -0.7 -0.2 0.0 0.4 -1.1 -0.6 0.0 -0.2 1.1 0.1 0.0 0.7 0.1

Principal method mean-median Sanger Pyro dPCR 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 -1.0 0.0 0.0 2.2 0.0 0.0 0.0 0.0 -1.4 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.1 0.0 0.0 0.0 0.0 0.0 0.0

MassARRAY 0.0 0.0 0.0

0.0 0.0

16/260

53.8 43.5 63.3 49.2 41.7

53.8 43.5 63.3 49.2 41.7

0.0 0.0 0.0 0.0 0.0

16/254

16/256

100.0 43.3 36.0 40.0 40.0 29.5 80.0 50.0 37.0

40.2 25.2

40.2 24.0

0.0 1.2

16/264

37.5 25.6

37.5 25.6

0.0 0.0

16/262

40.4 33.2

40.4 33.2

0.0 0.0

Table 7. Comparison of mean and median KRAS mutation percentages for quantitative methods in the collaborative study. Subtraction of the overall median percentage from the overall mean percentage for each principal method and material/dilution showed minimal impact on the use of either median or mean as consensus values.

WHO/BS/2017.2317 Page 20 Next, agreement between quantitative principal methods was considered and Lin’s concordance correlation coefficient (ρC) was calculated for each pair of principal methods using the median values shown in Table 7 (Figure 2). MassARRAY showed poor concordance with the four quantitative methods (ρC < 0.80 in all cases), which may be attributed to the semi-quantitative reporting; NGS, Sanger sequencing, pyrosequencing, and dPCR showed good concordance with each other (ρC > 0.90 in all cases). Specifically, Sanger sequencing showed good concordance with NGS, pyrosequencing, and dPCR (0.90 < ρC < 0.95), whereas pyrosequencing showed stronger concordance with NGS and dPCR (ρC > 0.95), and dPCR showed the strongest concordance with NGS (ρC = 0.995). Finally, to determine appropriate consensus mutation percentages for each material, only data from the three methods with the strongest concordance were considered (NGS, dPCR, and pyrosequencing; Table 8). If NGS data alone were used to assign the consensus mutation percentages, this would reflect the frequency at which this method appears to be used (n=17 of 67 methods in the collaborative study), and acknowledges the likely increasing use of NGS as more laboratories adopt this method, but it may be unwise to consider data from only one principal method. NGS and dPCR have very different underlying methodologies, yet have very strong concordance in this study. Thus the inclusion of dPCR data, with this technique gaining consideration as the new ‘gold standard’ high-sensitivity method in absolute quantification of molecular markers, strengthens the dataset (total n=21). Addition of the quantitative pyrosequencing data (total n=27) would result in consensus mutation percentages derived from three technically-different principal methods, but since pyrosequencing has less concordance with NGS than does dPCR (Figure 2), its impact on the overall data must be considered. This is noted especially with the increase in inter-quartile ranges (IQRs) determined when pyrosequencing is also included (for some materials/dilutions; Table 8). Therefore, consensus mutation percentages are presented as the overall median value for each material according to NGS and dPCR (Table 8).

WHO/BS/2017.2317 Page 21

0.927

0.980

0.937

0.995

0.920 7

0.977

0.736

0.684

0.812

0.742

Figure 2. Agreement of median KRAS mutation percentages for each pair of principal methods. The median values for the seven mutant KRAS materials and their dilutions are shown for each quantitative principal method. Both axes are percentage mutant KRAS; solid lines represent perfect agreement; dashed lines indicate fitted Deming regression models. Lin’s concordance correlation coefficient (ρC) was calculated for the median KRAS mutation percentages for each pair of principal methods. Good concordance is shown in light green (0.90 ≤ ρC < 0.95); excellent concordance (ρC ≥ 0.95) is shown in dark green.

WHO/BS/2017.2317 Page 22

Material

Dilution crude 1:2 1:2.5 1:5 crude 1:2 1:2.5 1:5 crude 1:2 1:2.5 1:5 crude 1:2 1:2.5 1:5 crude 1:2 1:2.5 1:5 crude 1:2 1:2.5 1:5 crude 1:2 1:2.5 1:5

16/252

16/258

NGS only Median IQR 65.7 1.0 30.8 27.5 2.6 12.7 1.5 100.0 0.3 57.3 3.5 47.5 5.7 28.7 71.5 48.1 41.2 26.4 85.6 78.1 71.8 57.9 99.7 49.5 43.0 21.2 49.3 26.0 24.0 12.0 66.9 39.9 33.2 20.3 5.8 0.5 3.5 1.9 1.3 1.3 3.9 0.4 1.6 1.8 2.6 2.9 3.0 1.3 4.9 2.6

Principal methods NGS & dPCR NGS, dPCR, & Pyro Median IQR Median IQR 65.7 1.0 65.7 1.8 31.3 0.4 31.3 0.6 26.6 2.8 26.6 2.3 12.7 0.9 12.7 1.3 99.98 0.1 100.0 0.5 58.0 2.8 58.9 2.6 47.5 5.3 48.7 6.6 29.0 71.5 48.3 41.8 25.9 85.6 75.6 71.9 58.2 99.7 51.1 42.7 21.4 49.7 26.8 23.8 12.2 66.9 40.7 34.5 20.3 4.5 0.6 3.4 1.7 2.4 1.0 0.4 2.9 0.4 3.2 1.6 1.6 2.9 2.3 1.8 2.4 1.5 4.3 1.9 29.0 71.5 49.7 41.9 26.7 85.6 75.6 71.7 58.1 99.7 52.2 43.0 21.2 49.7 27.5 24.4 12.2 66.7 42.0 35.7 20.3 3.7 1.0 4.7 3.7 7.3 2.3 2.4 1.3 1.9 0.6 2.6 1.7 2.7 3.1 1.6 1.7 3.0 1.6 2.8 4.6 6.7

16/260

16/254

16/256

16/264

16/262

Table 8. Overall median KRAS mutation percentages and inter-quartile ranges for quantitative principal methods in excellent concordance; NGS, dPCR, and pyrosequencing. The IQR is not shown where less than four mean values for triplicate tested samples are available for the principal method. Data are reported to 1 decimal place, except for crude material 16/258 which is reported to 2 decimal places to capture the apparent (low-level) presence of wild-type KRAS allele in this material (green).

WHO/BS/2017.2317 Page 23

KRAS Copy Numbers: Establishment of a Dilution Formula In this study, each crude material of the proposed WHO 1st International Reference Panel for genomic KRAS codons 12 and 13 mutations was subjected to three dilutions (1:2; 1:2.5, and 1:5, by combination with the nominal wild-type KRAS codons 12 and 13 material 16/266). The presence of two copies of the wild-type KRAS allele in material 16/266 was determined by ddPCR by reference firstly to MRC-5, a primary diploid cell line derived from normal lung tissue of a 14 week-old male foetus (Jacobs et al., 1970) and commonly used in vaccine development, and for in vitro cytotoxicity testing, and secondly to a commercial human gDNA derived from multiple anonymous donors (catalogue number G3041, Promega, Madison, WI, USA). Use of the wild-type KRAS codons 12 and 13 material 16/266 in the dilution of the mutant KRAS materials enabled the calculation of the KRAS allelic ratio (mutant: wild-type) and total KRAS copy number (mutant plus wild-type) in the latter, based upon the dilution response for each of the materials, which varied from quasi-linear to hyperbolic, and reflected the differing genomic complexity of these materials (Figure 3). A model-fitting algorithm was performed using Python 2.7 SciPy, with the best fitting model given by: y= x/(ax+b) where a and b are the coefficients obtained for each of the mutant materials after the convergence of the fitting algorithm to the dilution response (Table 9). To validate the fit of the dilution model for each of the mutant materials, particularly to lower mutation percentages, additional dilutions for each of the mutant materials were evaluated inhouse with ddPCR. The resulting mutation percentages were consistent with those predicated by the model (Appendix VI), thereby demonstrating the accuracy of the model fitting. Due to the large number of samples already tested by the collaborative study laboratories, it was impractical to request they assess further dilutions. However, the strong agreement of the in-house ddPCR data with the collaborative study consensus mutation percentages and the model-derived data provide confirmation of the suitability of the in-house ddPCR in verifying the calculations (Appendix VI).

WHO/BS/2017.2317 Page 24

Figure 3. Dilution responses of the seven mutant KRAS materials of the proposed WHO 1st International Reference Panel for genomic KRAS codons 12 and 13 mutations. Data shown are the consensus mutation percentages for each of the materials and their dilutions (crude, 1:2, 1:2.5, and 1:5). The blue lines represent the best fit dilution response.

WHO/BS/2017.2317 Page 25 coefficient a Material value 16/252 16/258 16/260 16/254 16/256 16/264 16/262 -0.00059 0.00307 0.00758 0.01035 0.00081 0.00450 0.00577 lower 95% -0.00190 0.00205 0.00702 0.01012 0.00034 0.00239 0.00492 upper 95% 0.00072 0.00409 0.00814 0.01059 0.00127 0.00661 0.00662 SE 0.00067 0.00052 0.00029 0.00012 0.00024 0.00108 0.00043 value 0.01584 0.00697 0.00646 0.00139 0.00923 0.01569 0.00923 coefficient b lower 95% 0.01469 0.00618 0.00608 0.00129 0.00884 0.01399 0.00859 upper 95% 0.01698 0.00776 0.00683 0.00150 0.00963 0.01739 0.00986 SE 0.00059 0.00040 0.00019 0.00005 0.00020 0.00087 0.00032

Table 9. Derivation of the a and b coefficients for the seven mutant KRAS material dilution curves. Coefficients, their lower and upper 95% confidence intervals, and standard error were derived from a model-fitting algorithm using Python 2.7 SciPy. SE, standard error.

The above model was used to derive KRAS exon 2 zygosity and consensus KRAS copy number for each mutant material. Furthermore, it was used to determine how the end-user may dilute each mutant material (with wild-type KRAS codons 12 and 13 material 16/266, or an in-house wild-type gDNA aligned to 16/266) to achieve standards at any desired lower mutation percentage. This was done by derivation of the coefficients for the KRAS allelic ratio (mutant: wild-type) and the total KRAS copy number for each material as shown in table 10. The variable zygosity and KRAS copy number for each mutant material is illustrated with the following examples: 1. For material 16/256 (KRAS p.G12S), the proposed consensus mutation percentage for the crude material is 99.7% mutant, which approximately halves to 51.1% mutant following a 1:2 dilution (one part crude material: one part wild-type diluent). The near-halving in value (by dilution with two wild-type copies) indicates that the KRAS exon 2 copy number is close to two. From the data modelling, the overall population-average allelic content is derived as 2.16661 mutant copies and 0.00802 wild-type copies, with a total of 2.17463 alleles. It is clear that the material is not quite (dizygous) homozygous mutant from the fact the consensus percentage dilution relationships are not quite linear. The extremely low number of wild-type copies may be correct and possibly due to reversion mutation or contamination or may be in truth be zero, but called as 0.00802 due to error margins in the mathematical modelling; 2. For material 16/252 (KRAS p.G12A), the consensus mutation percentage for the crude material is 65.7% which approximately halves to 31.3% mutant following a 1:2 dilution. The approximate halving in value again indicates that the KRAS exon 2 copy number in this material is approximately two. In this case however, the zygosity in the population of

WHO/BS/2017.2317 Page 26 cells used to generate the material was not simply heterozygous or homozygous, but 1.26302 mutant alleles plus 0.66246 wild-type alleles, a total of 1.92548 alleles; 3. The situation in the example of material 16/254 (KRAS p.G12R) is further complicated; the crude material consensus mutation percentage of 85.6% does not halve following a 1:2 dilution but rather reduces to only 75.6%, indicating a KRAS exon 2 copy number of greater than 2. The figure of 85.6% mutant in the crude material indicates an allelic ratio of approximately 6:1, i.e. approximately 6 mutant alleles for each wild-type allele. A sample with this allelic ratio and a total copy number of 6 would be expected to give a mutant percentage of approximately 66.7% following a 1:2 dilution. But as the mutant consensus percentage reduces to only 75.6%, it is apparent that the KRAS exon 2 copy number is greater than seven. The mathematical modelling derives values of overall population allelic content as 14.37815 mutant alleles to 2.50899 wild-type alleles with a total allelic content of 16.88713. In the proposed use of these materials as calibrants and in determining assay limit of detection, the end-user may produce a standard at any desired lower consensus percentage by dilution of the crude mutant material with the wild-type KRAS codons 12 and 13 material 16/266 (or a wildtype DNA calibrated to 16/266) by using the formula: ((((mutant KRAS copy number / wanted %)*100)- total KRAS copy number)/2)+1. For example, to prepare a standard of consensus mutation percentage 25% for material 16/254 (KRAS p.G12R), the allelic content figures are thus: ((((14.37815/25)*100)- 16.88713)/2)+1 = 21.31. Meaning that a 1 in 21.31 dilution of material 16/254 (1 part material 16/254 plus 20.31 parts wild-type KRAS codons 12 and 13 material 16/266) will yield a further standard of consensus mutant percentage 25% KRAS p.G12R, for example 1µl 16/254, plus 20.3µl 16/266. It will be possible for NIBSC to supply online interactive plots if end-users do not wish to perform their own calculations. The Instructions for Use will also contain dilution examples for the generation of a range of typical consensus mutant percentages (Appendix VII). Thus, the proposed consensus mutation percentages for the crude materials and the dilution formula can be used to prepare a range of standards at multiple mutation percentages from which assay calibration can be achieved. It should be noted however, that the calibration can only be achieved at and below the crude material consensus mutation percentage. Furthermore, it should be emphasised that these data derived from the consensus values are not necessarily empirical, but the materials and their dilutions achieve standardization since all laboratories will be deriving the same values.

WHO/BS/2017.2317 Page 27

Material

Consensus mutant KRAS copy number 1.26302 2.86821 3.09789 14.37815 2.16661 1.27470 2.16802

Consensus wild-type KRAS copy number 0.66246 0.01205 1.25031 2.50899 0.00802 1.29879 1.08336

Consensus total KRAS copy number 1.92548 2.88025 4.34820 16.88713 2.17463 2.57349 3.25138

16/252 16/258 16/260 16/254 16/256 16/264 16/262

Table 10. Calculated consensus mutant and wild-type KRAS copy number from mathematical modelling in each of the crude mutant materials. Formulae used to derive mutant KRAS copy number, wild-type KRAS copy number, and total KRAS copy number are: mutant KRAS copy number= (2/(100*b)); wild-type KRAS copy number= (((2*a)+(2*b))/b)(2/(100*b)); total KRAS copy number= (((2*a)+(2*b))/b).

Consensus Value Assignment Using the median KRAS mutation percentage for each of the seven materials, derived from the mean quantitative value of triplicate samples tested by NGS and dPCR methods, the genotype and consensus mutation percentage for each of the eight materials in the WHO 1st International Reference Panel for genomic KRAS codons 12 and 13 mutations is shown (Table 11). These data will be reported in the Instructions for Use, along with the IQR for each consensus mutation percentage (Appendix VII). End-users will be able to further dilute the mutant materials (with wild-type KRAS codons 12 and 13 material 16/266, or another wild-type gDNA calibrated to material 16/266) using a dilution formula, to achieve further standards at a range of lower consensus mutation percentages from which assay calibration is achieved. The proposed dilution formula to be used is as follows: ((((mutant KRAS copy number / wanted %)*100)- total KRAS copy number)/2)+1 where the consensus mutant KRAS copy number and total KRAS copy number are specific to each material (Table 11). End-users will be referred to the WHO report (via the Instructions for Use) for further details of this complex data analysis.

WHO/BS/2017.2317 Page 28

KRAS codon 12 Material or 13 mutation 16/252 p.G12A 16/258 p.G12C 16/260 p.G12D 16/254 p.G12R 16/256 p.G12S 16/264 p.G12V 16/262 p.G13D 16/266

Consensus mutation percentage (%) 65.7 99.98 71.5 85.6 99.7 49.7 66.9

Consensus IQR mutant KRAS copy number 1.0 0.1 0.6 1.0 0.4 2.9 1.5 1.26302 2.86821 3.09789 14.37815 2.16661 1.27470 2.16802 Wild-type

Consensus total KRAS copy number 1.92548 2.88025 4.34820 16.88713 2.17463 2.57349 3.25138

Table 11. Consensus values for the eight materials of the proposed WHO 1st International Reference Panel for genomic KRAS codons 12 and 13 mutations. KRAS genotype, consensus mutation percentage, and consensus KRAS copy numbers for use in calculating how each mutant material may be diluted to prepare further standards at lower mutation levels, are shown.

Qualitative Data Thirty two methods reported qualitative data (for full data see Appendix VIII). These data could not be used in the derivation of the consensus values. However, they valuably demonstrate the performance of the materials with Sanger sequencing, real-time PCR, KRAS StripAssay, HRM, ARMS-PCR, PCR-rSSO, minisequencing, and RFLP. Qualitative data were broadly reported as either ‘detected’ (where the actual genotype was reported, for example KRAS p.G12A), or ‘mutated’ where the specific genotype was not reported, but established to not be wild-type (for KRAS codons 12 and 13). Method 26 was able to further distinguish between mutations in KRAS codons 12 or 13, such that results were reported as KRAS p.G12X or p.G13X. All methods were able to report the correct mutations at all dilutions performed, except method 35 which reported ‘wild-type’ for the three samples of material 16/252 (at 1:2 dilution), KRAS p.G12C or p.G12V for sample 9 of material 16/258 (at 1:2 dilution), and ‘wild-type’ for the three samples of material 16/254 (at 1:2 dilution). Method 26 additionally reported KRAS p.A59X in samples 15 and 22 of material 16/254 (at 1:5 dilution), and in sample 13 of material 16/264 (crude and 1:5 dilution). Method 56 reported the failure of the forward primer in SNaPshot analysis to amplify KRAS c.G35, c.G37, and c.G38 for the three samples of material 16/258 (crude only; although c.G34T was amplified, and the reverse primer did amplify all locations, with p.G12C concluded).

Material 16/266 (nominal wild-type KRAS codons 12 and 13) All 67 methods reported results for material 16/266, the nominal wild-type KRAS codons 12 and 13 material (for full data see Appendix IX). All methods used similar terminology in reporting, broadly ‘wild-type’ or ‘no mutations detected’, with the following additional observations:

WHO/BS/2017.2317 Page 29 1. Method 38 reported a 3’UTR variant (KRAS c.*2505T>G) at 49.71% (sample 7), 51.26% (sample 10), and 49.86% (sample 19); 2. Method 17 reported a mutation in KRAS p.A59 (KRAS c.175G>A, or c.176C>A, or c.176C>G resulting in KRAS p.A59E, or p.A59G, or p.A59T) in sample 7 only. This sample has since been re-tested and reported as wild-type for KRAS codons 12, 13, 59, 61, 117, and 146; 3. Method 6 reported the wild-type genotype percentage as 99% (sample 7), 98% (sample 10), and 100% (sample 19); 4. Method 10 reported the KRAS codons 12 and 13 mutations percentage as 0% for samples 7, 10, and 19; 5. Method 60 reported the KRAS codons 12 and 13 mutations percentage as 0.00% for samples 7, 10, and 19; 6. Method 21 reported the wild-type genotype percentage as 100% for samples 7, 10, and 19.

Testing carried out by Participants More than one batch of consumables (for example, reagents, sequencing cartridges, or flow cells), or different instruments (for example, sequencing instruments) were reported as used in twenty nine methods. Testing was noted as carried out by more than one operator for thirty six methods.

Reference Samples used in the Study Positive and negative controls used by the participants were as follows: twelve methods used gDNA extracted from wild-type or mutant KRAS cell lines, including two methods which diluted the gDNA to be at or near the assay’s limit of detection; eleven methods used control materials provided with their commercial kit; nine methods used previously characterized clinical samples from patients and healthy controls, including two methods which specified the use of FFPE specimens, and two methods which diluted these samples to be at or near the assay’s limit of detection; six methods were reported to use commerically-available reference materials; two methods used plasmid controls, with one method using the plasmids at various KRAS mutation percentages; nine further methods also used control DNA materials but did not specify the source or type, with one laboratory noting that their control KRAS mutation percentage was independently verified (by NGS); one method utilized mutant KRAS double stranded DNA oligonucleotides combined with wild-type gDNA (and restriction endonuclease-treated) or (restriction endonuclease-treated) wild-type gDNA alone. The use of no template control samples was also noted for 18 methods. Eleven methods were reported to be used without control samples, although one method’s results were noted to be aligned to the human genome reference (as a quality control check).

WHO/BS/2017.2317 Page 30

Comments from the Participants Few laboratories provided additional comments concerning the collaborative study and materials. One laboratory commented that this was a great initiative [in the standardization of KRAS codons 12 and 13 mutation diagnostic testing]. Four laboratories noted that the collaborative study instructions were clear and the study was well organized. One laboratory would prefer to report the data using an Excel spreadsheet, which is noted. One laboratory withdrew their data from the study upon reading the collaborative study report, as they detected some errors in their reported data which they considered were not reflective of the actual results obtained. Furthermore, this laboratory was not in agreement with the grouping of their data with others’ as one principal method, as they considered the panels and equipment used to not be directly comparable. One laboratory commented that glass ampoules were less-preferable than plastic tubes. This issue is recognized, and NIBSC is accruing long-term stability data for gDNA stored in plastic tubes as a possible alternative format. However, freeze-drying in glass ampoules which can be completely sealed is the currently preferred method for ensuring the long-term stability of WHO International Standards. One laboratory commented that the materials appeared homogeneous upon quantification, with DNA concentration close to the expected 50 µg/ml. Another laboratory noted that their assay input volume could have been much reduced [due to the higher than usual DNA concentration]. Four laboratories commented on the comparability of these materials with patient samples; one laboratory noted that the high quality and concentration of DNA (which is easily amplified and analysed) was not reflective of the poor DNA quality and yield typically seen in FFPE tissue samples; one laboratory would prefer standardization of their laboratory process to include DNA extraction from the FFPE samples; two laboratories noted that the high level mutation percentages apparent in the materials were not reflective of that usually observed in patient samples; one laboratory sought clarification as to how high quality gDNA could standardize assays which usually measure fragmented DNA (from FFPE or as ctDNA). It is considered that these gDNA materials can achieve standardization of KRAS diagnostics by acting as primary standards. However, this is just a first step of what will undoubtedly be a multi-step process towards the diagnostic standardization of an extremely complex area of biology. The materials are not intended as ‘run controls’ i.e. reference materials which are representative of ‘usual’ patient samples, and serve as ‘stop/go’ controls to determine if the assay is working as it should on a particular day. As highest order materials, they are intended as calibrants for secondary standards, kits, or assays, such that all materials and processes align to a common reference. Our reasons for using gDNA are practical, and the reduced commutability with FFPE-extracted DNA or indeed ctDNA is recognised, but as a standardization effort, the batch size, stability, homogeneity, replaceability, and usability in as many diagnostic approaches as possible are importantly addressed. Once standardization begins, by all laboratories aligning to the same reference materials, any differences between methods are revealed, adjustments can be made, and harmonization is achieved.

WHO/BS/2017.2317 Page 31

Degradation Studies Accelerated Degradation Studies Multiple samples were reserved for in-house long-term accelerated degradation studies by storage at elevated temperatures (+37°C, +45°C, and +56°C), as well as for real-time stability monitoring at -20°C. Preliminary samples were assessed after five months’ storage and demonstrated no apparent degradation at +56°C, -20°C, or the baseline temperature of -150°C, as measured by electrophoresis (TapeStation; Figure 4), 260/280 nm absorbance (Nanodrop), DNA quantification (Qubit), and ddPCR (Table 12), with data comparable to that seen at the time of manufacture (Table 3), including a retained lower DIN for materials 16/252, 16/258, and 16/266, as compared with the other materials. The absence of degradation at elevated temperature resulted in the inability to predict loss of real-time stability. However, assurance that the materials are suitable for shipping at ambient temperatures was provided. Samples will continue to be assessed on a regular basis (typically annually) to ensure ongoing long-term stability for the lifetime of the panel. Previous experience with similar gDNA reference panels has demonstrated ongoing real-time stability at least twelve years post-manufacture.

Figure 4. TapeStation analysis of the eight materials of the proposed WHO 1st International Reference Panel for genomic KRAS codons 12 and 13 mutations following 5 months’ storage at various temperatures. High quality gDNA as indicated by a high molecular weight band and the absence of lower molecular weight fragmented gDNA, and as quantified by a high DIN, was apparent for all materials, including at elevated temperature, indicating the absence of degradation. Lane 1, DNA ladder (catalogue number 5190-6292, Agilent); lane 2, positive control (catalogue number G3041, Promega); lane 3, negative control (in-house degraded gDNA); lanes 4-27, ampoules stored for five months at -150°C, -20°C, or +56°C for each material as follows: 16/252 (KRAS p.G12A),16/258 (p.G12C), 16/260 (p.G12D), 16/254 (p.G12R), 16/256 (p.G12S), 16/264 (p.G12V), 16/262 (p.G13D), and 16/266 (wild-type).

WHO/BS/2017.2317 Page 32 16/252 (KRAS p.G12A) -150 -20 +56 Mean OD ratio 260/280 nm (n= 3) Mean DNA concentration (µg/ml; n= 3) Mean KRAS mutation % (ddPCR; n= 3) Coefficient of variation of KRAS mutation % (%; n=3) 1.90 42.33 1.92 46.00 1.88 46.33 16/258 (KRAS p.G12C) -150 -20 +56 1.93 44.83 1.93 45.07 1.90 48.63 16/260 (KRAS p.G12D) -150 -20 +56 1.88 42.17 1.91 47.67 1.87 45.73 Material code (genotype), storage temperature (°C) 16/254 (KRAS p.G12R) 16/256 (KRAS p.G12S) -150 -20 +56 -150 -20 +56 1.87 52.43 1.88 49.13 1.83 44.50 1.92 45.43 1.92 46.17 1.90 40.33 16/264 (KRAS p.G12V) -150 -20 +56 1.89 42.50 1.88 42.37 1.88 43.13 16/262 (KRAS p.G13D) -150 -20 +56 1.88 49.37 1.90 53.13 1.87 47.33 16/266 (wild-type) -150 -20 +56 1.87 122.67 1.88 121.00 1.88 124.67

65.80

65.87

67.30

99.99

99.98

99.97

71.60

72.03

71.60

85.17

85.40

85.40

99.94

99.99

99.96

50.30

50.17

50.63

67.90

67.57

68.17

0.00

0.00

0.00

1.50

2.05

1.71

0.01

0.03

0.01

0.14

1.40

0.78

1.39

0.47

0.82

0.07

0.02

0.04

1.43

2.48

1.60

1.45

0.82

0.98

N/C

N/C

N/C

Table 12. Nanodrop, Qubit, and ddPCR analyses of the eight materials of the proposed WHO 1st International Reference Panel for genomic KRAS codons 12 and 13 mutations following five months’ storage at various temperatures. High quality gDNA as indicated by expected 260/280 nm absorbances, approximately consistent DNA concentrations, and reproducible KRAS mutation percentages, as measured by ddPCR, including at elevated temperature, indicated the absence of degradation in all materials. N/C, not calculated as all values were zero.

WHO/BS/2017.2317 Page 33

Post-Reconstitution Stability Studies End-users are recommended to use the materials on the day of reconstitution. However, in-house analysis determined reconstituted freeze-dried gDNA to be stable for at least four days at +4°C, or two months at -20°C (Figure 5 and Table 13), with data comparable to that seen at the time of manufacture (Table 3), including a retained lower DIN for materials 16/252, 16/258, and 16/266, as compared with the other materials.

Figure 5. Post-reconstitution analysis of the eight materials of the proposed WHO 1st International Reference Panel for Genomic KRAS codons 12 and 13 mutations by TapeStation. High quality gDNA as indicated by a high molecular weight band and the absence of lower molecular weight fragmented gDNA, and as quantified by a high DIN, was apparent for all materials, indicating post-reconstitution stability following 4 days at +4°C or 2 months at 20°C. Lanes 1 and 12, DNA ladder (Agilent); lanes 2 and 13, positive control (Promega); lanes 3 and 14, negative control (in-house degraded gDNA); lanes 4 to 11, reconstituted materials stored for four days at +4°C, and lanes 15 to 22 for two months at -20°C as follows: 16/252 (KRAS p.G12A),16/258 (p.G12C), 16/260 (p.G12D), 16/254 (p.G12R), 16/256 (p.G12S), 16/264 (p.G12V), 16/262 (p.G13D), and 16/266 (wild-type).

WHO/BS/2017.2317 Page 34 16/252 (KRAS p.G12A) +4/4d -20/2mon Mean OD ratio 260/280 nm (n= 3) Mean DNA concentration (µg/ml; n= 3) Mean KRAS mutation % (ddPCR; n= 3) Coefficient of variation of KRAS mutation % (%; n=3) 1.89 47.90 1.93 48.40 16/258 (KRAS p.G12C) +4/4d -20/2mon 1.90 48.60 1.91 46.80 16/260 (KRAS p.G12D) +4/4d -20/2mon 1.93 53.80 1.98 48.47 Material code (genotype), storage temperature (°C)/time 16/254 (KRAS p.G12R) 16/256 (KRAS p.G12S) 16/264 (KRAS p.G12V) +4/4d -20/2mon +4/4d -20/2mon +4/4d -20/2mon 1.90 56.50 1.88 47.37 1.93 52.53 1.97 47.50 1.88 51.40 1.88 45.33 16/262 (KRAS p.G13D) +4/4d -20/2mon 1.93 58.67 1.92 42.67 16/266 (wild-type) +4/4d -20/2mon 1.90 121.00 1.91 122.00

66.60

68.23

99.99

99.95

74.13

71.77

84.90

84.63

99.97

99.77

50.27

50.77

67.33

68.07

0.00

0.00

1.56

1.91

0.01

1.35

4.07

0.56

0.54

1.88

0.03

0.17

0.75

2.66

0.60

1.70

N/C

N/C

Table 13. Post-reconstitution analyses of the eight materials of the proposed WHO 1st International Reference Panel for genomic KRAS codons 12 and 13 mutations by Nanodrop, Qubit, and ddPCR. High quality gDNA as indicated by expected 260/280 nm absorbances, approximately consistent DNA concentrations, and reproducible KRAS mutation percentages, as measured by ddPCR, indicating post-reconstitution stability following four days at +4°C or two months at -20°C. N/C, not calculated as all values were zero.

WHO/BS/2017.2317 Page 35

Discussion The first International Genomic Reference Material was approved by the WHO Expert Committee on Biological Standardization (ECBS) in November 2004 and comprised a panel of three materials for the genomic diagnosis of Factor V Leiden. The same approach has been used at NIBSC for the subsequent preparation a range of other International Genomic Reference Materials including for Prothrombin Mutation G20210A, Factor VIII intron 22 inversion, Fragile X, Prader Willi & Angelman Syndromes, and JAK2 V617F. In the current study, a similar approach was adopted: cell lines were established at NIBSC in order to assure a continual future supply of the same gDNA materials, and following large scale cell culture and gDNA extraction, a panel of eight gDNAs was freeze-dried in ampoules to represent a range of clinically-relevant KRAS codons 12 and 13 mutations. Fifty five laboratories participated in an international collaborative study to evaluate the suitability of this panel of gDNAs as the proposed WHO 1st International Reference Panel for genomic KRAS codons 12 and 13 mutations, and reported data from sixty seven methods. The study was designed to determine the performance of the panel in a large number of laboratories using a variety of methods. In order to assess the consistency of the panel’s performance, participants were requested to carry out the study on three separate days and with different operators and batches of reagents where possible. All materials and their replicates performed well in the study, across a range of commonly-used diagnostic methods, with data from NGS, Sanger sequencing, pyrosequencing, and dPCR methods showing good concordance with one other. The reason for the poor concordance of MassARRAY data with that of the quantitative methods is most likely attributable to the semiquantitative reporting of this method, with this discrepancy considered to have no impact on the usual qualitative diagnostic reporting from this technology. The data from the two quantitative real-time PCR methods in the collaborative study were further discordant from all other quantitative methods for reasons unknown. Such data emphasise the need for improved standardization of KRAS codons 12 and 13 mutation testing, which can be aided by the international availability of highest order reference materials. Consensus KRAS mutation percentage values were assigned to each of the materials as the median values of all NGS and dPCR methods, since these data showed the strongest concordance, with NGS recognized as an increasingly used diagnostic method, and dPCR considered by some as a ‘gold standard’ method for sensitive, absolute quantification. However, it is proposed that reporting only KRAS mutation percentage does not account for the genomic complexity of these materials, including actual mutant and wild-type KRAS copy numbers. Such information was derived from the response of the materials to dilution with the (diploid, two wild-type KRAS exon 2 copies) material 16/266, and can be applied to a dilution formula specific to each mutant KRAS material in the preparation of additional standards at any further KRAS codons 12 and 13 mutation percentage. These standards will enable assay calibration across a broad mutation percentage range considered to be within the usual reporting scope of patient mutation levels. It is noted that if insufficient wild-type KRAS codons 12 and 13 material 16/266 were available for the preparation for such dilutions, an in-house wild-type gDNA could be aligned to material 16/266 and then used as the diluent.

WHO/BS/2017.2317 Page 36 Other SNVs in KRAS, and in other CRC-associated genes were reported in the collaborative study assessment of the eight materials (including in the wild-type KRAS codons 12 and 13 material 16/266), also indicative of the value of these materials in mimicking the in vivo genomic complexity and variability of a tumour sample. The provision of the materials as high quality gDNAs is recognized as being dissimilar to the fragmented DNA typically analysed in patient samples, but practical considerations, together with an intent for this reference panel to be potentially applicable to any substrate used in KRAS mutation detection assays, it is considered to be an important first step towards the standardization of KRAS diagnostics. The challenges in characterizing and assigning consensus values to these materials are considered unlikely to be unique to KRAS and mutant KRAS-containing cancer cell lines, or indeed mutant KRAScontaining tumour samples, and so establish critical future considerations for the quantitative measurement of cancer mutations and its standardization.

Conclusions and Proposal The results of this international multi-centre study demonstrated that the following eight preparations are suitable for use as reference materials in laboratories carrying out genotyping of KRAS codons 12 and 13 with the proposed KRAS codons 12 and 13 consensus mutation percentages derived from the median values of NGS and dPCR methods as 16/252 (65.7% KRAS p.G12A), 16/258 (99.98% p.G12C), 16/260 (71.5% p.G12D), 16/254 (85.6% p.G12R), 16/256 (99.7% p.G12S), 16/264 (49.7% p.G12V), 16/262 (66.9% p.G13D), and 16/266 (wild-type KRAS codons 12 and 13). These materials may be diluted (with wild-type KRAS codons 12 and 13 material 16/266 or another wild-type gDNA aligned to 16/266) by application of a calculation specific to each material (based on its consensus mutant and total KRAS copy number), to produce standards at a range of KRAS consensus mutation percentages which enable calibration of quantitative assays. These values are not necessarily empirical; however, if end-users calibrate their assays using these standards, harmonization is achieved. NIBSC would like to propose that the above eight materials be established as the World Health Organization 1st International Reference Panel for genomic KRAS codons 12 and 13 mutations (NIBSC code 16/250), for use in diagnostic techniques including NGS, Sanger sequencing, realtime PCR, pyrosequencing, dPCR, MassARRAY, KRAS StripAssay, high resolution melt analysis, ARMS-PCR, PCR-rSSO, minisequencing, and RFLP, and intended for use by manufacturers for the calibration of diagnostic kits and by clinical laboratories for the calibration of assays and secondary standards used in routine diagnostic assays for KRAS codons 12 and 13 mutation detection.

Acknowledgements We gratefully acknowledge the significant contributions of all collaborative study participants. Particular thanks go to Simon Patton of EMQN (Manchester, UK) and Sandi Deans of UK NEQAS for Molecular Genetics (Edinburgh, UK) for connecting us with many of the

WHO/BS/2017.2317 Page 37 participants. We would also like to extend our gratitude to Paul Matejtschuk and the Standardisation Science group at NIBSC, along with the Standards Processing Division for their development, and processing of the materials.

References Amado RG, Wolf M, Peeters M, Van Cutsem E, Siena S, Freeman DJ, Juan T, Sikorski R, Suggs S, Radinsky R, Patterson SD, Chang DD. Wild-type KRAS is required for panitumumab efficacy in patients with metastatic colorectal cancer. J Clin Oncol. 2008 Apr 1;26(10):1626-34 Birkeland E, Wik E, Mjøs S, Hoivik EA, Trovik J, Werner HM, Kusonmano K, Petersen K, Raeder MB, Holst F, Øyan AM, Kalland KH, Akslen LA, Simon R, Krakstad C, Salvesen HB. KRAS gene amplification and overexpression but not mutation associates with aggressive and metastatic endometrial cancer. Br J Cancer. 2012 Dec 4;107(12):1997-2004 COSMIC v81, released 09-MAY-17, accessed 25th June 2017 (http://cancer.sanger.ac.uk/cosmic/) Chan, E. 2015. KRAS in Colorectal Cancer. My Cancer Genome. https://www.mycancergenome.org/content/disease/colorectal-cancer/kras/ (Updated June 18), accessed 29th June 2017 Douillard JY, Oliner KS, Siena S, Tabernero J, Burkes R, Barugel M, Humblet Y, Bodoky G, Cunningham D, Jassem J, Rivera F, Kocákova I, Ruff P, Błasińska-Morawiec M, Šmakal M, Canon JL, Rother M, Williams R, Rong A, Wiezorek J, Sidhu R, Patterson SD. PanitumumabFOLFOX4 treatment and RAS mutations in colorectal cancer. N Engl J Med. 2013 Sep 12;369(11):1023-34 Eijkelenboom A, Kamping EJ, Kastner-van Raaij AW, Hendriks-Cornelissen SJ, Neveling K, Kuiper RP, Hoischen A, Nelen MR, Ligtenberg MJ, Tops BB. Reliable Next-Generation Sequencing of Formalin-Fixed, Paraffin-Embedded Tissue Using Single Molecule Tags. J Mol Diagn. 2016 Nov;18(6):851-863 Fumagalli D, Gavin PG, Taniyama Y, Kim SI, Choi HJ, Paik S, Pogue-Geile KL. A rapid, sensitive, reproducible and cost-effective method for mutation profiling of colon cancer and metastatic lymph nodes. BMC Cancer. 2010 Mar 16;10:101 Garcia EP, Minkovsky A, Jia Y, Ducar MD, Shivdasani P, Gong X, Ligon AH, Sholl LM, Kuo FC, MacConaill LE, Lindeman NI, Dong F. Validation of OncoPanel: A Targeted NextGeneration Sequencing Assay for the Detection of Somatic Variants in Cancer. Arch Pathol Lab Med. 2017 Jun;141(6):751-758 Hawkins JR, Hawkins M, Boyle J, Gray E, Matejtschuk P, Metcalfe P. Genetic reference materials and their application to haematology. Biologicals. 2010 Jul;38(4):467-73

WHO/BS/2017.2317 Page 38

Jacobs JP, Jones CM, Baille JP. Characteristics of a human diploid cell designated MRC-5. Nature. 1970 Jul 11;227(5254):168-70 Kerr EM, Gaude E, Turrell FK, Frezza C, Martins CP. Mutant Kras copy number defines metabolic reprogramming and therapeutic susceptibilities. Nature. 2016 Mar 3;531(7592):110-3 Karnoub AE, Weinberg RA. Ras oncogenes: split personalities. Nat Rev Mol Cell Biol. 2008 Jul;9(7):517-31 Kluk MJ, Lindsley RC, Aster JC, Lindeman NI, Szeto D, Hall D, Kuo FC. Validation and Implementation of a Custom Next-Generation Sequencing Clinical Assay for Hematologic Malignancies. J Mol Diagn. 2016 Jul;18(4):507-15 Krypuy M, Newnham GM, Thomas DM, Conron M, Dobrovic A. High resolution melting analysis for the rapid and sensitive detection of mutations in clinical samples: KRAS codon 12 and 13 mutations in non-small cell lung cancer. BMC Cancer. 2006 Dec 21;6:295 Lansford CD, Grenman R, Bier H, Somers KD, Kim SY, Whiteside TL, Clayman GL, Welkoborsky H-J, Carey TE. In Human Cell Culture Volume II, Cancer Cell Lines Part 2 (eds Masters JRW, Palsson, B) 185–255 (Kluwer Academic, Dordrecht, 1999) Levi M, Prayogi G, Sastranagara F, Sudianto E, Widjajahakim G, Gani W, Mahanadi A, Agnes J, Khairunisa BH, Utomo AR. Clinicopathological Associations of K-RAS and N-RAS Mutations in Indonesian Colorectal Cancer Cohort. J Gastrointest Cancer. 2017 Jan 3 [Epub ahead of print] Luthra R, Patel KP, Reddy NG, Haghshenas V, Routbort MJ, Harmon MA, Barkoh BA, Kanagal-Shamanna R, Ravandi F, Cortes JE, Kantarjian HM, Medeiros LJ, Singh RR. Nextgeneration sequencing-based multigene mutational screening for acute myeloid leukemia using MiSeq: applicability for diagnostics and disease monitoring. Haematologica. 2014 Mar;99(3):465-73 Masters JR. Human cancer cell lines: fact and fantasy. Nat Rev Mol Cell Biol. 2000 Dec;1(3):233-6 Mekenkamp LJ, Tol J, Dijkstra JR, de Krijger I, Vink-Börger ME, van Vliet S, Teerenstra S, Kamping E, Verwiel E, Koopman M, Meijer GA, van Krieken JH, Kuiper R, Punt CJ, Nagtegaal ID. Beyond KRAS mutation status: influence of KRAS copy number status and microRNAs on clinical outcome to cetuximab in metastatic colorectal cancer patients. BMC Cancer. 2012 Jul 17;12:292 Na II, Rho JK, Choi YJ, Kim CH, Koh JS, Ryoo BY, Yang SH, Lee JC. Clinical features reflect exon sites of EGFR mutations in patients with resected non-small-cell lung cancer. J Korean Med Sci. 2007 Jun;22(3):393-9

WHO/BS/2017.2317 Page 39 Neumann J, Zeindl-Eberhart E, Kirchner T, Jung A. Frequency and type of KRAS mutations in routine diagnostic analysis of metastatic colorectal cancer. Pathol Res Pract. 2009;205(12):85862 Nollau P, Moser C, Weinland G, Wagener C. Detection of K-ras mutations in stools of patients with colorectal cancer by mutant-enriched PCR. Int J Cancer. 1996 May 3;66(3):332-6 Ostrem JM, Peters U, Sos ML, Wells JA, Shokat KM. K-Ras(G12C) inhibitors allosterically control GTP affinity and effector interactions. Nature. 2013 Nov 28;503(7477):548-51 Peeters M, Karthaus M, Rivera F, Terwey JH, Douillard JY. Panitumumab in Metastatic Colorectal Cancer: The Importance of Tumour RAS Status. Drugs. 2015 May;75(7):731-48 Python 2.7 SciPy: Open Source Scientific Tools for Python, 2017-, accessed 30th June 2017 (http://www.scipy.org/) Roberts PL, Lloyd D. Virus inactivation by protein denaturants used in affinity chromatography. Biologicals. 2007 Oct;35(4):343-7 Sasaki H, Hikosaka Y, Kawano O, Moriyama S, Yano M, Fujii Y. Evaluation of Kras gene mutation and copy number gain in non-small cell lung cancer. J Thorac Oncol. 2011 Jan;6(1):1520 Shigaki H, Baba Y, Watanabe M, Miyake K, Murata A, Iwagami S, Ishimoto T, Iwatsuki M, Yoshida N, Baba H. KRAS and BRAF mutations in 203 esophageal squamous cell carcinomas: pyrosequencing technology and literature review. Ann Surg Oncol. 2013 Dec;20 Suppl 3:S48591 Sholl LM, Do K, Shivdasani P, Cerami E, Dubuc AM, Kuo FC, Garcia EP, Jia Y, Davineni P, Abo RP, Pugh TJ, van Hummelen P, Thorner AR, Ducar M, Berger AH, Nishino M, Janeway KA, Church A, Harris M, Ritterhouse LL, Campbell JD, Rojas-Rudilla V, Ligon AH, Ramkissoon S, Cleary JM, Matulonis U, Oxnard GR, Chao R, Tassell V, Christensen J, Hahn WC, Kantoff PW, Kwiatkowski DJ, Johnson BE, Meyerson M, Garraway LA, Shapiro GI, Rollins BJ, Lindeman NI, MacConaill LE. Institutional implementation of clinical tumor profiling on an unselected cancer population. JCI Insight. 2016 Nov 17;1(19):e87062 Singh RR, Patel KP, Routbort MJ, Reddy NG, Barkoh BA, Handal B, Kanagal-Shamanna R, Greaves WO, Medeiros LJ, Aldape KD, Luthra R. Clinical validation of a next-generation sequencing screen for mutational hotspots in 46 cancer-related genes. J Mol Diagn. 2013 Sep;15(5):607-22 Solassol J, Vendrell J, Märkl B, Haas C, Bellosillo B, Montagut C, Smith M, O'Sullivan B, D'Haene N, Le Mercier M, Grauslund M, Melchior LC, Burt E, Cotter F, Stieber D, Schmitt FL, Motta V, Lauricella C, Colling R, Soilleux E, Fassan M, Mescoli C, Collin C, Pagès JC, Sillekens P. Multi-Center Evaluation of the Fully Automated PCR-Based Idylla™ KRAS

WHO/BS/2017.2317 Page 40 Mutation Assay for Rapid KRAS Mutation Status Determination on Formalin-Fixed ParaffinEmbedded Tissue of Human Colorectal Cancer. PLoS One. 2016 Sep 29;11(9):e0163444 Trung NT, Huyen TTT, Hoan PQ, Song LH. Peptide clamped PCR assay for identifying tissue and serum circulating mutant KRAS DNA from colorectal patients. (In Vietnamese) Weyn C, Van Raemdonck S, Dendooven R, Maes V, Zwaenepoel K, Lambin S, Pauwels P. Clinical performance evaluation of a sensitive, rapid low-throughput test for KRAS mutation analysis using formalin-fixed, paraffin-embedded tissue samples. BMC Cancer. 2017 Feb 16;17(1):139 van Dongen JJ, Langerak AW, Brüggemann M, Evans PA, Hummel M, Lavender FL, Delabesse E, Davi F, Schuuring E, García-Sanz R, van Krieken JH, Droese J, González D, Bastard C, White HE, Spaargaren M, González M, Parreira A, Smith JL, Morgan GJ, Kneba M, Macintyre EA. Design and standardization of PCR primers and protocols for detection of clonal immunoglobulin and T-cell receptor gene recombinations in suspect lymphoproliferations: report of the BIOMED-2 Concerted Action BMH4-CT98-3936. Leukemia. 2003 Dec;17(12):2257-317 Yamane LS, Scapulatempo-Neto C, Alvarenga L, Oliveira CZ, Berardinelli GN, Almodova E, Cunha TR, Fava G, Colaiacovo W, Melani A, Fregnani JH, Reis RM, Guimarães DP. KRAS and BRAF mutations and MSI status in precursor lesions of colorectal cancer detected by colonoscopy. Oncol Rep. 2014 Oct;32(4):1419-26 Zimmermann G, Papke B, Ismail S, Vartak N, Chandra A, Hoffmann M, Hahn SA, Triola G, Wittinghofer A, Bastiaens PI, Waldmann H. Small molecule inhibition of the KRAS-PDEδ interaction impairs oncogenic KRAS signalling. Nature. 2013 May 30;497(7451):638-42

WHO/BS/2017.2317 Page 41

Appendix I. Collaborative Study Participants Country Argentina Participant(s) Javier Sfalcin Analía Seravalle Elizabeth Algar Timmy Chan Chelsee Hewitt Kym Pham Michael Christie Anne Bernt Christian Oberkanins Geneviève Vandercruyssen Peter Sillekens Flávia Escremim de Paula Escremim Rui Reis Gabriel Macedo Patricia Ashton-Prolla Sandra Leistner Segal Alicja Parker Wang Dong Yanling Wu Cai Lijun Rastislav Slavkovský Jiri Drabek Jarmila Simova Niels Pallisgaard Saskia Hussung Marie Follo Albrecht Stenzinger Institution CIBIC – Centro de Diagnostico Medico de Alta Complejidad, Rosario Genetics and Molecular Pathology Laboratory, Monash Health, Melbourne Molecular Pathology, Peter MacCallum Cancer Centre, Melbourne University of Melbourne Centre for Cancer Research, Melbourne ViennaLab Diagnostics, Vienna Biocartis NV, Mechelen Molecular Oncology Research Center, Barretos Cancer Hospital, São Paulo Centro de Pesquisa Experimental. Hospital de Clínicas de Porto Alegre Contextual Genomics Inc., Vancouver Gene Tech (Shanghai) Company Ltd., Shanghai Shanghai Topgen Biopharm Technology Co. Ltd., Shanghai Institute of Molecular and Translational Medicine, Palacky University, Olomouc CGB laboratoř a.s., Ostrava Department of Pathology, Zealand University Hospital, Næstved Medical Center- University of Freiburg, Department of Medicine I Institute of Pathology, University Hospital Heidelberg Molecular Diagnostic Unit, Histopathology Department, Hellenic Red Cross General Hospital of Athens

Australia

Australia Australia Austria Belgium Brazil

Brazil Canada China China Czech Republic Czech Republic Denmark Germany Germany

Greece

Loukia Psaridi

WHO/BS/2017.2317 Page 42 Greece Hungary Hungary Hong Kong India Indonesia Ireland Italy Italy Italy Stylianos Ventis Béla Kajtár Attila Mokánszki Gábor Méhes Stella Tsang Chris Wong Rashmi Khadapkar Cynthia Christina Michael Levi Ahmad R. Utomo Reiltin Werner Michael Bennett Matteo Fassan Gianmarco Musciano Silvio Veronese BioAnalytica GenoTypos S.A., Athens Department of Pathology, University of Pécs Department of Pathology, Medical Center, University of Debrecen Hong Kong Molecular Pathology Diagnostic Centre SRL Limited, Mumbai Kalbe Genomics Laboratory (KalGen), Jakarta Department of Pathology, Cork University Hospital Surgical Pathology Unit, Department of Medicine, University of Padua Diatech Pharmacogenetics, Jesi Niguarda Cancer Center, Grande Ospedale Metropolitano Niguarda, Milan Pathological Anatomy, Polytechnic University of the Marche Region, School of Medicine, United Hospitals, Ancona Medical and Biological Laboratories Co. Ltd., Nagoya SRL Inc. Hachioji Laboratory, Tokyo Cancer Research Centre, Institute for Medical Research, Kuala Lumpur Sunway Institute for Healthcare Development, Sunway University, Selangor Radboud University Medical Center, Nijmegen Department of Pathology and Laboratory Diagnostics, Maria Sklodowska-Curie Memorial Cancer Center and Institute of Oncology, Warsaw

Italy

Antonio Zizzi Marina Scarpelli Makoto Kaneda Yukiko Abe Masahiro Kanou Lu Ping Tan Geok Wee Tan Hamidah Akmal Hisham Suat-Cheng Peh Sin-Yeang Teow (Ronald) Riki Willems

Japan Japan Malaysia

Malaysia Netherlands

Poland

Pawel Leszczynski Łukasz Szafron

WHO/BS/2017.2317 Page 43 Bartosz Wasąg Magdalena Chmara Department of Biology and Genetics, Medical University of Gdansk GenoMed - Diagnósticos de Medicina Molecular, S.A., Instituto de Medicina Molecular, Lisbon University Institute of Anatomical and Molecular Pathology, Faculty of Medicine of the University of Coimbra Resident Laboratory Ltd., Oradea National Research Center for Hematology, Moscow Lucence Diagnostics Pte Ltd., Singapore Department of Clinical Genetics, St. Elisabeth Cancer Institute, Bratislava Eone-Diagnomics Inc., Incheon Hospital del Mar, Barcelona Ampligen Diagnósticos, Ponferrada Chula GenePRO Center, Bangkok Intergen Genetics Diagnosis and Research Centre, Ankara Acibadem Labgen Genetic Diagnosis Center, Istanbul Manchester Centre for Genomic Medicine, Saint Mary’s Hospital, Manchester Source BioScience, Nottingham NIBSC, South Mimms All Wales Medical Genetics Service, University Hospital of Wales, Cardiff The Centre for Molecular Pathology, The Royal Marsden NHS Foundation Trust, Sutton

Poland

Portugal

Ana Carla Sousa Lina Carvalho Ana Alarcão Vitor Sousa Ana Filipa Ladeirinha Ferenc Fazakas Andrey B. Sudarikov Min-Han Tan Katarína Závodná Lukáš Šebest Jin-Sik Bae Beatriz Bellosillo Hada Navas Fernández Pilar Arca Chinachote Teerapakpinyo Serdar Ceylaner Haldun Dogan Sait Tümer Cumhur Gökhan Ekmekci Michael Bulman Kevin Chittock Pia Sanzone Rhianedd Ellwood-Thompson Dörte Wren Lisa Thompson

Portugal

Romania Russia Singapore Slovakia South Korea Spain Spain Thailand Turkey Turkey UK UK UK UK

UK

WHO/BS/2017.2317 Page 44 Rajyalakshmi Luthra Keyur P. Patel Bedia A. Barkoh Kristen Floyd Jawad Manekia Lynette Sholl Ngo Tat Trung

USA

The University of Texas MD Anderson Cancer Center, Houston Department of Pathology, Brigham and Women's Hospital, Boston Department of Molecular Biology, Military Central Hospital, Hanoi

USA Vietnam

WHO/BS/2017.2317 Page 45

Appendix II. Collaborative Study Design Material Coded samples Protocol Laboratory number A 16 4 crude* 1:2** 16/252 1, 16, 21 B 5 40 C 1 8 D 7 2 crude 1:2 1:2.5 1:2.5 1:5 Y 49 54 crude 1:2 16/258 8, 9, 20 Z 51 52 AA 53 50 AB 55 56 crude 1:2 1:2.5 1:2.5 1:5 Q 17 34 crude 1:2 16/260 5, 12, 17 R 35 36 59 1:2 1:2.5 S 37 31 T 39 6 crude 1:2.5 1:5 E 9 10 crude 1:2 16/254 2, 15, 22 F 15 3 G 19 14 H 11 38 crude 1:2 1:2.5 1:2.5 1:5 I 33 18 crude 1:2 16/256 3, 14, 23 J 13 20 K 12 22 L 23 24 58 crude M 21 32 crude 1:2 16/264 4, 13, 24 N 27 28 O 25 30 P 29 26 crude 1:2 1:2.5 1:2.5 1:5 U 45 42 crude 1:2 16/262 6, 11, 18 V 43 44 57 1:2 1:2.5 W 41 46 X 47 48 crude 1:2.5 1:5

Dilutions

1:2 1:2.5

1:5

1:5

1:5

1:5

1:2.5 1:5

1:5

1:5

1:5

Each material was provided as triplicate coded ampoules. The materials were tested at four different dilutions overall (crude, 1:2, 1:2.5, and 1:5). Each participant tested all materials at one dilution (*in bold; n=24); one material was additionally tested (in triplicate) at a second dilution (**n=3). Following the distribution of the collaborative study report, laboratory 13 withdrew their data from the study (see Comments from the Participants, above; red).

WHO/BS/2017.2317 Page 46

Appendix III. Example Collaborative Study Protocol

WHO/BS/2017.2317 Page 47

WHO/BS/2017.2317 Page 48

WHO/BS/2017.2317 Page 49

WHO/BS/2017.2317 Page 50

WHO/BS/2017.2317 Page 51

WHO/BS/2017.2317 Page 52

WHO/BS/2017.2317 Page 53

WHO/BS/2017.2317 Page 54

Appendix IV. Details of the Methods used in the Collaborative Study Twelve principal testing methods for the detection of KRAS codons 12 and 13 mutations were used by participants in the collaborative study. Methods reporting quantitative data are in bold (n=35, including MassARRAY which is considered semi-quantitative); all other methods reported qualitative data (n=32). Further details provided for each of the laboratories/ methods were as follows: Method / Lab 1 7 Method outline Next-generation sequencing AmpliSeq Cancer Hotspot Panel v2 massively parallel sequencing, with Ion Torrent S5 Qiagen GeneRead DNAseq Targeted Panels v2 – Human Clinically Relevant Tumour Panel, Qiagen GeneRead DNAseq Panel PCR kit v2, Illumina TruSeq DNA PCR-Free HT Sample Preparation kit fastRAS (KRAS, NRAS, and BRAF genotyping) on Illumina MiSeq using Illumina MiSeq Reagent Nano kit v2, Illumina MiSeq Reporter software using PCR Amplicon workflow TruSight custom-made CRC panel (Illumina), low-input library prep (Illumina) with the Illumina MiSeq with either v2 or v3 cartridge, Variant Studio/in-house bioinformatics pipeline Two-step amplification using Find-ITTM Somatic Hotspot Cancer Panel (Contextual Genomics), bead clean-up (Ampure XP bead), Illumina Nextera XT kit with Illumina MiSeq and Illumina MiSeq Reagent kit v2 (300 cycle), data analysis with Contextual Genomics software, with variant allele frequencies generated with MutationSeq software Amplicon-based using in-house real-time PCR with primer sequences from van Dongan et al. (2003), Fluidigm access array system for enrichment of interest for targeted Massively Parallel Sequencing, custom somatic mutation panel, Agencourt AMPure XP purification, Illumina MiSeq and Illumina MiSeq v2 x300 cycle kit, multiple annotation data sources, and in-house variant viewer (PathOS), with Integrative Genomics Viewer (IGV) Single molecule Molecular Inversion Probe (smMIP)-based library preparation from Eijkelenboom et al. (2016), followed by Illumina NextSeq500 and analysis with SeqNext software Amplification with SOMATIC 1 MASTR Dx kit (Multiplicom), magnetic bead clean-up, Illumina MiSeq and analysis with Sophia DDM v4.3.5 platform (Sophia Genetics) Agencourt AMPure XP beads for DNA purification, QIAseq Human Actionable Solid Tumor Panel, followed by Illumina MiSeq, QIAGEN online data analysis software Customised protocol for KRAS exons 2 and 3 mutations with Ion Torrent PGM, analysis with Ion Torrent Suite and Ion Reporter 5.0

4

25

29

30b

33

34

38 43

WHO/BS/2017.2317 Page 55 Targeted amplification of BRAF exon 15, PIK3CA exons 9 and 20, KRAS and NRAS exons 2, 3, and 4, EGFR exons 18-21, with Illumina MiSeq and MiSeq reporter software Amplification of region of interest by in-house designed PCR primers, sequencing of amplicons by NGS amplicon resequencing, Illumina NexteraXT library prep kit and Illumina MiSeq with 300v2 cartridges, data analysed by IGV software (Broad Institute) Agilent Haloplex Custom Enrichment Panel (CM28) for 28 genes (full exon/coding region), Illumina MiSeq with 600 cycle v3 cartridges, using MiSeq Control software 2.4, MiSeq Reporter software 2.5.1, alignment with SureCall v3.0, with in-house software for SNV (and indel) viewing Custom Illumina TruSeq Cancer Panel (CM53) for 53 genes. Illumina MiSeq with 300 cycle v2 cartridges, using MiSeq Control software 2.4, MiSeq Reporter software 2.5.1, in-house software for SNV (and indel) viewing, reference Luthra et al. (2014) Ampliseq Hotspot V2 panel using IonTorrent, reference Singh et al. (2013) Rapid Heme Panel Amplicon with Illumina MiSeq, reference Kluk et al. (2016) OncoPanel Hybrid Capture, references Garcia et al. (2017), Sholl et al. (2016) Sanger sequencing ABI Prism 3130 Genetic Analyser with Foundation Data Collection v3.0 software, Sequencing Analysis 5.3.1 (Applied Biosystems) QIAquick PCR purification kit (QIAGEN), BigDye v3.1 sequencing reagents (Life Technologies), DyeEx 96 sequencing purification kit (QIAGEN) AccuPrep Gel purification kit, SeqScape 2.0 software Typing assay, preceded by HRM analysis as screening assay (method 20a), reference Solassol et al. (2016) 3500 series Genetic Analyzer (Applied Biosystems), reference Yamane et al. (2014) Applied Biosystems Avant 310 Genetic Analyser, quantification by reference to NGS-quantified KRAS mutation samples Preceded by HRM analysis, using initial primer sequences from van Dongan et al. (2003), then KRAS HRM primers based on the 92bp amplicon described in Krypuy et al. (2006) modified with M13 universal sequencing primers, using Roche Lightcycler LC480 and/or COBAS z480, positive HRM samples are sequenced with BigDye Terminator v3.1 chemistry (Applied Biosystems), data analysed with Mutation Surveyor v4.0.5 Preceded by wild-type allele clamped PCR, with sequencing of the amplified mutant allele, not fully validated for quantitative reporting Preceded by peptide nucleic acid (PNA)-clamping of the wild-type allele, with sequencing of the amplified mutant allele MultiScreen Filter Plates PCR 96 purification kit (Millipore), sequenced with common primers from PCR and BigDye Terminator v3.1 Cycle Sequencing kit (Applied Biosystems), Montage SEO96 Sequencing Reaction Cleanup kit (Millipore), Applied Biosystems 3130 Genetic Analyzer, analysis with Sequencing Analysis Software 6 v6.0 Build id:FC3

48

50

54a

54b 54c 58a 58b 8 16 18 20b 27b 28

30a

31 32

40b

WHO/BS/2017.2317 Page 56 42 Two analyses with altered annealing temperature and reverse primer, BigDye Terminator v3.1 Cycle Sequencing Kit (ThermoFisher scientific) Primers from reference Na et al. (2007), JETquick PCR Product Purification Spin kit (Genomed), BigDye Terminator v1.1 cycle sequencing kit (Applied Biosystems), purification with ilustra AutoSeq G-50 Dye Terminator Removal kit (SATIS Amersham GE Healthcare), 3130 Genetic Analyzer (Applied Biosystems), Data Collection v3.0, Sequencinganalysis 5.2 (Applied Biosystems) Independent amplification of codons 12 and 13, selective cleavage of wild-type amplicons with restriction endonucleases, extraction of the mutant-containing amplicons from agarose gel, BigDye cycle sequencing (Life Technologies) Real-time PCR Easy KRAS kit (Diatech Pharmacogenetics) to selectively amplify mutated DNA, each assay includes endogenous control gene amplification assay, using Stratagene MX3005P with MxPro v4.10 build 389 software (Agilent Technologies) In-house peptide clamped real-time PCR technology, reference Trung et al. KRAS/BRAF Mutation Analysis kit (EntroGen), using Roche Lightcycler 480 and software v1.5 Idylla KRAS mutation test on the Idylla platform (Biocartis) which includes an integrated sample preparation method Cobas KRAS mutation test in Cobas 4800 system (Lightcycler 480) with Cobas 4800 software 2.2.0 (Roche) In-house KRAS TaqMan assay using ABI7900 KRAS Mutation Test v2 (LSR; Roche) according to manufacturer’s instructions Cobas KRAS mutation test in Cobas Z 480 Analyzer (Roche) Idylla KRAS mutation test (KRAS_IVD/2.0) on the Idylla platform (Biocartis), references Solassol et al. (2016), Weyn et al. (2017) KRAS Diagnostic Kit (Lucence Diagnostics), SYBR Green-based real-time ARMS-PCR, using QuantStudio 3 (Applied Biosystems) and normalisation to plasmid standards Idylla KRAS mutation test on the Idylla platform (Biocartis), references Solassol et al. (2016), Weyn et al. (2017) Pyrosequencing PyroMark PCR kit with PyroMark Q24 and associated software 2.0 (Qiagen) PyroMark Gold Q96 reagents with PyroMark Q96 workstation and pyrosequencer, and 2.5.8 software (Qiagen), not validated for quantitative reporting PyroMark PCR kit with custom primers, with PyroMark Q96 ID and software with allele quantification (AQ) analysis (Qiagen) KRAS Gene Mutation Detection kit (GeneTech (Shanghai) Company) with PyroMark Gold Q96 kit, PyroMark Q24 workstation, PyroMark Q24 pyrosequencer and 2.0.6 software (Qiagen) Therascreen KRAS Pyro kit (Qiagen), with PyroMark Q24 workstation and pyrosequencer and 2.0 software (Qiagen) PyroMark Q24 as in reference Shigaki et al. (2013)

44

53

2b 3 9 17 19 24b 26 27a 40a 46 51 5 6 10 23 37 49

WHO/BS/2017.2317 Page 57 59 60 21 52b 55b Therascreen KRAS RGQ (Qiagen) digital PCR ddPCR with PrimePCR ddPCR Mutation Assays for single KRAS mutations, QX200 and QuantaSoft 1.7.4 analysis (BioRad) ddPCR with in-house designed assays, initially multiplexed followed with specific individual assays, QX100 and QuantaSoft 1.7.4 analysis (BioRad) ddPCR with KRAS G12/G13 Screening kit as quantification assay, with QX100 and QuantaSoft analysis (BioRad), preceded by PCR-rSSO for initial genotyping (method 52a) TaqMan SNP genotyping assays with QuantStudio3D as quantification assay, preceded by ARMS-PCR for initial genotyping (method 55a) MassARRAY Myriapod Colon Status kit (Diatech Pharmacogenetics) with direct detection of the SNV based upon molecular mass i.e. not by fluorescent probes (or other indirect methods), allelic frequency calculated by height and relative area of the peak, considered to be semi-quantitative Myriapod Colon Status kit (Diatech Pharmacogenetics), with results validated by Sanger sequencing, considered to be semi-quantitative Myriapod Colon Status kit (Diatech Pharmacogenetics), Analyzer 4 system, Typer 4.0 software (Agena Bioscience), reference Fumagalli et al. (2010) KRAS StripAssay KRAS XL StripAssay according to manufacturer’s instructions According to manufacturer’s instructions KRAS XL StripAssay, with StripAssay Evaluator software (ViennaLab Diagnostics) High Resolution Melt Analysis Laboratory-developed test with primers and PCR based on reference Krypuy et al. (2006), as in reference Levi et al. (2017) HRM analysis as screening assay, reference Solassol et al. (2016), followed by Sanger sequencing for typing (method 20b) ARMS-PCR In-house real-time PCR assay utilizing Ct and delta Ct values KRAS RGQ PCR kit (Qiagen) for initial genotyping, followed by dPCR for quantification (method 55b) PCR-rSSO MEBGEN RASKET RAS mutation detection kit (Medical & Biological Laboratories), Luminex system, UniMAG2.0 software analysis (Luminex) PCR-Luminex for initial genotyping with MEBGEN RASKET RAS mutation detection kit (Medical & Biological Laboratories), Luminex 200 (Luminex), and UniMAG2.0 software analysis (Medical & Biological Laboratories), followed by dPCR for quantification (method 52b) RFLP Laboratory-developed test with primers, PCR, and digestion based on reference Nollau et al. (1996), as in reference Levi et al. (2017) Minisequencing

2a

11 22 24a 41 47

15a 20a 35 55a

39

52a

15b

WHO/BS/2017.2317 Page 58 36 56 In-house-developed assay using SNaPshot multiplex kit (Applied Biosystems) In-house-developed assay using SNaPshot multiplex kit (Applied Biosystems)

WHO/BS/2017.2317 Page 59

Appendix V. Collaborative Study Results for Quantitative Methods 16/252 KRAS p.G12A Method / Lab 1 7 4 25 1 crude 65 65 16 crude 65 65 21 crude 65 65 1 1:2 16 1:2 21 1:2 1 1:2.5 22 16 1:2.5 25 21 1:2.5 24 1 1:5 10 14 16 1:5 11 12 21 1:5 11 13 8 crude 9 crude 20 crude 8 1:2 9 1:2 20 1:2 16/258 KRAS p.G12C 8 1:2.5 46 9 1:2.5 45 20 1:2.5 48 8 1:5 33 100 25.8 26.7 27.2 100 100 34.8 37.7 40.1 9 1:5 22 20 1:5 31

31

34

29

29 30b

13.58 28.5 26.7 32.2

10.12

12.66 52.2 55.3 52

27.48

30.43

29.63

33 34 38 43 48 50 54a 54b 54c 58a 58b 16 28 31 3 46 5 6 10 23 37 49 60 21 52b 55b 2a 11 22

67 66.3

65 66.1

65 65.7 14.72 29.62 31.18 30.23 12 28 28 29 13 13 14.42 12.77

100 99.7

100 99.7

100 99.8 24.01 55.86 66.73 55.69 47 50 49 43 28 39 29 44 30 24.58 25.55

67.5 65.3 63.9

66.6 68.2 63.6

66.7 64.3 64.2 14 10 12 10 12 12

100 99.5 100

99.9 99.3 100

99.9 99.5 100

59.2 54.9 55.4

57.8 52.9 55.7

59 53.3 56.5 23 27 19 20 20 19

70

60

60

35 50 20.35 33.8

30 40 29.39 33.7

30 40 25 27.61 33.4 >75% 27.6 49.11 25.4 73.6 26 12 12 14 13 21 25

95

95

95 80 14.69 60.4 80 12.62 60.6 80 44 10.18 54.85 58.8 G12C (no %) G12C (no %) 42 G12C (no %) 41 74.64 >50% 46 51

63.7

66.3

62.9 16 26.8 29.2 24.10 28.3 25.60 13.60 17.80 13.50 14

95.8

100

95.5 44 55.9 56.3 49.50 46.39 54.2 46.70 47.92 31.69 57 42 60.54 64.1 59.2 27.82 46 46 27.38 42 46 27.70 35.30 25.20

68 67.20 52.86

66 67.10 54.49

67 66.70 52.78

33.50 31.79

30.50 32.17

30.00 31.7

26.70

94 99.95 99.99 Not done

95 99.98 99.96 Not done

96 99.99 100 Not done

74 60.45 55.87

58 58.10 57.37

60 58.80 58.75

51.40 34.49

58

57

56 28.3 29.5 24.25

12.17 16 17.5

12.84 17 18

12.12 15 17.5

WHO/BS/2017.2317 Page 60 16/260 KRAS p.G12D Method / Lab 1 7 4 25 5 crude 12 crude 17 crude 5 1:2 12 1:2 17 1:2 5 1:2.5 39 12 1:2.5 41 17 1:2.5 41 5 1:5 29 70 71 72 34.8 35.9 38 12 1:5 27 17 1:5 24 83 86 86 75.1 72.2 73.2 56.75 (STK11 P281L 13.33%, F354L 14.36%) 72.3 70.5 71.9 57.28 (STK11 P281L 14.68%, F354L 15.68%) 60.68 (STK11 P281L 15.76%, F354L 16.95%) 2 crude 15 crude 22 crude 2 1:2 15 1:2 22 1:2 16/254 KRAS p.G12R 2 1:2.5 G12A (10%) 15 1:2.5 72 22 1:2.5 72 2 1:5 64 15 1:5 64 22 1:5 58

29 30b

24.62 43.9 45.8 45.8

26.68

27.2

33 34 38 43 48 50 54a 54b 54c 58a 58b 16 28 31 3 46 5 6 10 23 37 49 60 21 52b 55b 2a 11 22

73 71.5

72 71.5

72 71.6

48 50.31

45.1 50.91

46.6 22.42 47.97 37 41 42 43 37 35 26.59 23.99

87 85.5 85.05

86 85.7 84.26

87 85.7 85.71 78.97 77.16 78.15

54.61 54 70 70 69

61.81 53

56.31 56

73.2 71.9 70.1

70.8 73.1 71.4

70.4 70.2 71 29 23 24.4 17 G13D (22%) 18

82.7 86.4 85.8

79.3 85.7 86.3

80.8 87.4 85.9 63 54 58 47 57 54

80

80

80 60 7.00 54.3 60 15.42 54.8 65 48 5.30 42.34 54.2 32 38 33.7 26.40 33 42 32.8 26.50 30 19.42 30.2 44 46 26 20 31

85

85

85 80 61.27 76 80 36.69 75.4 77 75 34.95 75 G12R (no %) G12R (no %) 58 G12R (no %) 56 78.3 60 70.3 72.6 72.99 67.9 69.20 58.50 60.90 55.80 63 94.17 73.46 52 93.52 96.59 71 48.07 80.83

72 70.1

73 80.6

68 73.7 51.9 51 41.40 49.3 41.60

82.5 42 32.2 24.10 84 86.27 72.41

90.7

93.8

78.3

75 71.80 69.93

73 71.20 70.07

75 71.60 69.46

59.30 47.18

47.90 45.69

47.50 47.71

42.50

82 85.33 71.89

83 85.79 72.18

77.26 75.77

74.97 75.08

74.65 74.56

73.43

24.32 51 42.5 54.9 53.8 52.6

24.57 43 41.5

23.8 42 41

65

62

63 47.36 51.71 48.6

Not done 39 46.5

Not done 40 42

Not done 42 40.5

WHO/BS/2017.2317 Page 61 16/256 KRAS p.G12S Method / Lab 1 7 4 25 3 crude 14 crude 23 crude 3 1:2 14 1:2 23 1:2 3 1:2.5 42 14 1:2.5 43 23 1:2.5 43 3 1:5 20 99 100 100 45.3 47.2 45.8 20.77 (STK11 Q37* 11.16%) 41.9 40.5 41.9 20.45 (STK11 Q37* 12.65%) 22.83 (STK11 Q37* 12.10%) 50.35 (EGFR R836R 20.33%) 51.89 (EGFR R836R 19.95%) 50.8 (EGFR R836R 20.39%) 24.9 24.2 25.7 14 1:5 23 23 1:5 45 49 45 45 20.3 24.4 24.6 9 9.82 (EGFR R836R 5.19%) 14.6 11.4 13.02 (EGFR R836R 5.95%) 13.9 12.5 13.2 (EGFR R836R 6.29%) 14.3 4 crude 13 crude 24 crude 4 1:2 13 1:2 24 1:2 16/264 KRAS p.G12V 13 1:2.5 21 24 1:2.5 20 4 1:2.5 21 4 1:5 16 13 1:5 14 24 1:5 15

29 30b

33 34 38 43 48 50 54a 54b 54c 58a 58b 16 28 31 3 46 5 6 10 23 37 49 60 21 52b 55b 2a 11 22

100 99.8

100 99.7

100 99.8

47

15

46 21.2 20.2 15 21.53

49 50.3

50 49.4

50 50.1 10.39 25.74 31.1 21.18 18 27 26 27 17 19 15.08 10.4

52.65

53.93

51.21 13 44 42 44 48.2 44.2 47.7 26 23 24 20 16 15 40 40 40 49.7 48.4 45.4 50 48.6 49.2 17

99.8 99.7 100 99 99 100

100 99.5 100 99 97 100

99.9 99.7 100 99.9 98 100 40 19.66 53.2 50 26.12 55.5 40 36 16.19 G12S (no %) 53.9 27.96 71.55 37 35

13 12 40 4.40 26.8 40 8.74 28.2 40 6.44 33.96 27.4 8.5 46.3 57.4 47.5 20 27.2 25 24.40 22.7 22.70 15.60 45 51.20 50.41 42 50.90 50.63 44 52.30 49.88 26.38 24.44 30 31 30 28 30 28

11 9

11 11

G12S (no %) 100 97 97 99.98 99.53 100 100 86 99.97 99.59 100 100 45.2 98 99.89 99.51 49.90 50.72 50.50 50.13 49.50 55.79 Not done 32 24 21.7 40.90 44.3 43.70 41.9 40.60 21.00 18

G12S (no %) 17

G12S (no %) 18

11 16

7 17

25.70

19.50

32.30 28.5 22.6

29.80 26.1 27.52

27.70 28.14 26.07

24.30

13.50

12.10

38.63

41.31

40.63

Not done 26 22 27.4

Not done 26 25 23

12.07 23 27 38.12 39.75 34.6

12.05 25 25.5

13.31 26 27

WHO/BS/2017.2317 Page 62 16/262 KRAS p.G13D Method / Lab 1 7 4 25 6 crude 11 crude 18 crude 6 1:2 11 1:2 18 1:2 6 1:2.5 22 11 1:2.5 36 18 1:2.5 33 6 1:5 24 68 65 67 30.2 32.7 36.8 11 1:5 19 18 1:5 19

29 30b 68 (IDH2 c.435delG (p.T146fs) 6%) 67 68 (IDH2 c.435delG (p.T146fs) 6%) 66.6 67 (IDH2 c.435delG (p.T146fs) 6%) 67.1

20.62 37.1 36.9 38.4

19.07

21.08

33 34 38 43 48 50 54a 54b 54c 58a 58b 16 28 31 3 46 5 6 10 23 37 49 60 21 52b 55b 2a 11 22

16.01 38.2 68 66.2 65.4 69 68 67 66.1 65.4 67 36 65 67.1 68.8 20 16 80 80 80 50 8.86 42.3 66.2 65.2 68.2 30 39.6 64 68.50 53.89 64 67.60 54.06 63 67.60 52.94 44.80 38.85 42.90 38.47 42.10 44.64 Not done 37 32 39.79 41 40.5 37.00 38.7 34.30 39.2 35.80 20.00 50 18.22 43.3 50 41 12.02 40.96 44.6 27 81.2 54.43 17.51 31 39 37 37 43.7 37.78 31.8 not done not done 26

18.65 26

17.35 28

20 17

21 18

40.61 27 31

32.52 27 30

20.80

19.30

Not done 33 32

Not done 30 35

Data from methods 3 and 46 were excluded from further analysis as they were clearly distinct from those of other quantitative methods (red); method 1 reported KRAS p.G12A in sample 2 of material 16/254 (at 1:2.5 dilution; red); method 7 reported 45% KRAS p.G12S for sample 23 of material 16/256 (at 1:5 dilution), which was considered an outlier and excluded from further analysis (yellow); method 29 additionally reported STK11 p. P281L and p.F354L in all three samples of material 16/254 (at 1:5 dilution), STK11 p.Q37* in all three samples of material 16/256 (at 1:5 dilution), and EGFR p.R836R in all three samples of material 16/264 (at 1:5 dilution and crude; orange); method 33 reported 15% KRAS p.G12S for sample 14 of material 16/256 (at 1:2 dilution), which was considered an outlier and excluded from further analysis (yellow), and additionally reported IDH2 c.435delG (p.T146fs) in all three samples of material 16/262 (crude; orange); method 43 was unable to test samples 11 and 18 of material 16/262 at 1:2.5 dilution (orange); method 58a reported KRAS p.G13D in sample 17 of material 16/260 (at 1:5 dilution; red); method 46 reported >75% KRAS p.G12A for sample 1 of material 16/252 (at 1:2.5 dilution), >50% KRAS p.G12C for sample 20 of material 16/258 (at 1:2.5 dilution), and unknown percentage KRAS p.G12S for sample 3 of material 16/256 (at 1:2.5 dilution), all likely due to the absence of high percentage quantification controls in these assays (orange); method 6 was unable to quantify KRAS c.G34 mutations, and thus the three samples of material 16/258 were reported as KRAS p.G12C of unknown percentage (at 1:5 dilution), the three samples of material 16/254 were reported as KRAS p.G12R of unknown percentage (at 1:5 dilution), and the three samples of material 16/256 were reported as KRAS p.G12S of unknown percentage (at 1:5 dilution); the three samples of material 16/264 were each tested twice with the mean for the two results reported, except sample 11 which failed on one of replicates (orange); method 55b did not perform dPCR analysis (and therefore quantification) of materials 16/258 (crude), 16/254 (at 1:5 dilution), 16/256 (at 1:5 dilution), or 16/262 (at 1:5 dilution), although genotypes were identified in the laborator y’s other method, 55a (orange). Results are reported to a maximum of two decimal places.

WHO/BS/2017.2317 Page 63

Appendix VI. Agreement of the collaborative study consensus values, mathematical modelling, and in-house ddPCR

The accuracy of the mathematically-derived model for the dilution response of each mutant material was confirmed by the evaluation of additional dilutions with in-house ddPCR. The resulting mutation percentages were consistent with those predicated by the model. The line shown is the best fit for the collaborative study consensus mutation percentages for each of the materials and their dilutions (blue circles). Mean in-house ddPCR data align closely to the best-fit model (n=3-6 for each dilution; green circles). N/A, not applicable, as lower dilutions were not performed in the collaborative study.

WHO/BS/2017.2317 Page 64 KRAS codon 12 Material or 13 mutation Collaborative study consensus Dilution mutation percentage (%) crude 1:2 1:2.5 1:5 1:6.25 1:12.50 1:31.25 crude 1:2 1:2.5 16/258 p.G12C 1:5 1:13.6 1:27.6 crude 1:2 1:2.5 1:5 1:14.8 1:30.85 crude 1:2 1:2.5 1:5 1:22 1:67 1:142 crude 1:2 1:2.5 1:5 1:10 1:20 crude 1:2 1:2.5 65.7 31.3 26.6 12.7 N/A N/A N/A 99.98 58.0 47.5 29.0 N/A N/A 71.5 48.3 41.8 25.9 N/A N/A 85.6 75.6 71.9 58.2 N/A N/A N/A 99.7 51.1 42.7 21.4 N/A N/A 49.7 26.8 23.8 Predicted mutation percentage from model fitting (%) 65.6 32.2 25.6 12.7 10.2 5.1 2.0 99.58 58.8 48.8 26.4 10.2 5.1 71.2 48.8 42.2 25.1 9.7 4.8 85.1 76.1 72.3 57.8 24.4 9.7 4.8 99.6 51.9 41.9 21.3 10.7 5.4 49.5 27.9 22.9 In- house ddPCR mutation percentage (%) 66.4 N/A N/A N/A 11.2 5.3 2.1 99.99 N/A N/A N/A 10.4 5.3 72.0 N/A N/A N/A 9.5 4.7 85.9 N/A N/A N/A 23.6 9.3 4.3 100.0 N/A N/A N/A 11.0 5.2 50.6 N/A N/A

16/252

p.G12A

16/260

p.G12D

16/254

p.G12R

16/256

p.G12S

16/264

p.G12V

WHO/BS/2017.2317 Page 65 1:5 1:6.6 1:13.6 crude 1:2 1:2.5 1:5 1:10.82 1:22.47 12.2 N/A N/A 66.9 40.7 34.5 20.3 N/A N/A 12.1 9.3 4.6 66.7 41.3 34.7 19.3 9.5 4.7 N/A 9.5 4.6 67.7 N/A N/A N/A 10.4 5.1

16/262

p.G13D

WHO/BS/2017.2317 Page 66

Appendix VII. Instructions for Use

WHO/BS/2017.2317 Page 67

WHO/BS/2017.2317 Page 68

WHO/BS/2017.2317 Page 69

WHO/BS/2017.2317 Page 70

WHO/BS/2017.2317 Page 71

WHO/BS/2017.2317 Page 72

Appendix VIII. Collaborative Study Results for Qualitative Methods 16/252 KRAS p.G12A Method / Lab 8 18 20b 27b 30a 32 40b 42 44 53 2b 9 17 19 24b 26 27a 40a 51 59 24a 41 47 15a 20a 35 55a 39 52a 15b 36 56 1 crude detected 16 crude detected 21 crude detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected mutated mutated mutated mutated G12X mutated detected detected detected mutated detected detected detected mutated detected detected detected detected detected detected mutated G12X mutated G12X mutated detected detected detected detected detected mutated mutated wild-type mutated mutated wild-type mutated mutated wild-type detected detected detected mutated detected detected mutated detected detected mutated detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected mutated detected detected mutated detected detected mutated detected detected detected detected detected detected detected detected detected mutated mutated detected mutated mutated G12C or G12V mutated mutated detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected mutated detected detected detected mutated detected detected detected detected detected detected detected detected detected detected detected detected detected detected mutated mutated mutated mutated G12X mutated G12X mutated G12X detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected 1 1:2 16 1:2 21 1:2 1 1:2.5 detected 16 1:2.5 detected 21 1:2.5 detected 1 1:5 detected 16 1:5 detected 21 1:5 detected 8 crude detected 9 crude detected 20 crude detected detected detected detected detected detected detected detected detected detected 8 1:2 9 1:2 20 1:2 16/258 KRAS p.G12C 8 1:2.5 detected 9 1:2.5 detected 20 1:2.5 detected 8 1:5 9 1:5 20 1:5

detected

detected

detected detected detected detected

detected

detected

detected

WHO/BS/2017.2317 Page 73 16/260 KRAS p.G12D Method / Lab 8 18 20b 27b 30a 32 40b 42 44 53 2b 9 17 19 24b 26 27a 40a 51 59 24a 41 47 15a 20a 35 55a 39 52a 15b 36 56 5 crude detected 12 crude detected 17 crude detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected mutated mutated mutated mutated G12X mutated detected detected detected mutated detected detected detected mutated detected detected detected mutated G12X mutated G12X mutated detected detected detected detected detected mutated mutated detected detected detected detected detected mutated detected detected mutated detected detected mutated detected mutated mutated detected mutated mutated detected detected detected detected detected detected mutated mutated wild-type detected detected detected detected detected detected detected mutated detected detected detected detected detected mutated detected detected mutated detected mutated mutated mutated detected detected detected mutated mutated wild-type mutated mutated wild-type mutated mutated mutated detected detected detected detected detected detected detected detected mutated detected detected detected mutated detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected mutated mutated mutated mutated mutated G12X mutated mutated G12X, A59X mutated mutated G12X, A59X detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected 5 1:2 12 1:2 17 1:2 5 1:2.5 detected 12 1:2.5 detected 17 1:2.5 detected 5 1:5 12 1:5 17 1:5 2 crude detected 15 crude detected 22 crude detected detected detected detected detected detected detected detected detected detected 2 1:2 15 1:2 22 1:2 16/254 KRAS p.G12R 2 1:2.5 detected 15 1:2.5 detected 22 1:2.5 detected 2 1:5 15 1:5 22 1:5

detected

detected

detected

detected

detected

detected

detected detected

detected detected

detected detected

detected

detected

detected

WHO/BS/2017.2317 Page 74 16/256 KRAS p.G12S Method / Lab 8 18 20b 27b 30a 32 40b 42 44 53 2b 9 17 19 24b 26 27a 40a 51 59 24a 41 47 15a 20a 35 55a 39 52a 15b 36 56 3 crude detected 14 crude detected 23 crude detected 3 1:2 detected detected detected 14 1:2 detected detected detected 23 1:2 detected detected detected 3 1:2.5 detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected mutated detected mutated detected mutated mutated mutated detected detected detected detected detected detected detected mutated mutated detected mutated mutated detected mutated mutated detected detected detected detected mutated mutated mutated detected detected detected mutated detected detected mutated detected detected mutated detected detected detected detected detected detected detected detected detected mutated detected detected mutated detected detected mutated detected detected detected detected detected detected mutated mutated detected mutated mutated detected mutated mutated detected detected detected detected detected detected detected mutated detected detected detected mutated detected detected detected detected detected detected detected detected detected detected detected detected mutated mutated mutated G12X mutated G12X mutated G12X G12X G12X, A59X G12X mutated detected detected detected mutated detected detected detected mutated detected detected detected mutated mutated mutated detected detected detected detected detected mutated mutated mutated mutated G12X mutated G12X, A59X mutated G12X detected detected detected detected detected detected detected detected detected detected detected detected detected detected 14 1:2.5 detected detected detected 23 1:2.5 detected detected detected detected detected detected detected detected detected detected detected detected 3 1:5 14 1:5 23 1:5 4 crude detected 13 crude detected 24 crude detected detected detected detected detected detected detected detected detected detected detected detected detected 4 1:2 13 1:2 24 1:2 16/264 KRAS p.G12V 13 1:2.5 detected 24 1:2.5 detected 4 1:2.5 detected 4 1:5 13 1:5 24 1:5

detected

detected

detected

detected

detected

detected

WHO/BS/2017.2317 Page 75 16/262 KRAS p.G13D Method / Lab 8 18 20b 27b 30a 32 40b 42 44 53 2b 9 17 19 24b 26 27a 40a 51 59 24a 41 47 15a 20a 35 55a 39 52a 15b 36 56 6 crude detected 11 crude detected 18 crude detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected 6 1:2 11 1:2 18 1:2 6 1:2.5 detected 11 1:2.5 detected 18 1:2.5 detected 6 1:5 11 1:5 18 1:5 detected

detected detected

detected detected

detected detected detected detected detected

detected

detected

detected mutated mutated mutated mutated G13X mutated detected detected detected mutated detected detected detected mutated detected detected detected detected detected detected detected detected detected detected detected detected detected detected detected mutated G13X mutated G13X

detected

detected

detected mutated mutated detected mutated mutated detected mutated mutated detected

detected detected detected mutated detected detected mutated detected detected mutated detected detected

detected detected

detected detected

detected

detected

Results from laboratories reporting qualitative data were summarized as either ‘detected’ (where the actual genotype was reported, for example KRAS p.G12A), or ‘mutated’ where the specific genotype was not reported, but established to not be wild-type (for KRAS codons 12 and 13), although method 26 was able to distinguish between mutations in KRAS codons 12 and 13, such that results were reported as KRAS p.G12X or p.G13X. Method 9 was only able to test the three samples of material 16/254 (at crude and 1:2 dilution) due to resource availability; method 26 additionally reported KRAS p.A59X in samples 15 and 22 of material 16/254 (at 1:5 dilution; orange), and in sample 13 of material 16/264 (crude and 1:5 dilution; orange); method 35 reported ‘wild-type’ for the three samples of material 16/252 (at 1:2 dilution) with the results of samples 16 and 21 based on a doubled delta Ct value (red), KRAS p.G12C or p.G12V for sample 9 of material 16/258 (at 1:2 dilution; red), with the results of all three samples of material 16/258 (at 1:2 dilution) based on a doubled delta Ct value (red/orange), the result for sample 5 of material 16/260 (at 1:2 dilution) based on a doubled delta Ct value (orange), and ‘wild-type’ for all three samples of material 16/254 (at 1:2 dilution; red); method 56 reported the failure of the forward primer in SNaPshot analysis to amplify KRAS c.G35, c.G37, and c.G38 for the three samples of material 16/258 (crude only; although KRAS c.G34T was amplified, and the reverse primer did amplify these locations, with KRAS p.G12C concluded; orange).

WHO/BS/2017.2317 Page 76

Appendix IX. Material 16/266 (nominal wild-type KRAS codons 12 and 13) Data Method / 7 10 19 Lab crude crude crude 1 7 4 wild-type wild-type wild-type 25 29 30b 33 no mutations detected no mutations detected no mutations detected 34 wild-type wild-type wild-type 38 43 48 50 54a no mutations detected no mutations detected no mutations detected 54b no mutations detected no mutations detected no mutations detected 54c no mutations detected no mutations detected no mutations identified 58a 58b 8 16 no mutations detected no mutations detected no mutations detected 18 wild-type wild-type wild-type 20b 27b 28 30a 31 32 no mutations detected no mutations detected no mutations detected 40b 42 no mutations detected no mutations detected no mutations detected 44 53 2b 3 9 testing not performed testing not performed testing not performed 17 A59X no mutations detected no mutations detected 19 24b 26 27a 40a 46 51 5 6 10 wild-type (0% mutation) wild-type (0% mutation) wild-type (0% mutation) 23 37 49 no mutations detected no mutations detected no mutations detected 59 60 no mutations detected (0.00%) no mutations detected (0.00%) no mutations detected (0.00%) 21 wild-type (100%) wild-type (100%) wild-type (100%) 52b 55b 2a 11 22 24a 41 47 15a 20a 35 55a 39 52a 15b 36 56 7 1:2 10 1:2 19 1:2 7 1:2.5 no mutations detected 10 1:2.5 no mutations detected 19 1:2.5 no mutations detected 7 1:5 no mutations detected no mutations detected no mutations detected no mutations detected no mutations detected no mutations detected wild-type no mutations detected wild-type wild-type 10 1:5 no mutations detected 19 1:5 no mutations detected

3’UTR variant no mutations detected no mutations detected no mutations detected no mutations detected no mutations detected no mutations detected no mutations detected

3’UTR variant no mutations detected

3’UTR variant no mutations detected

wild-type wild-type wild-type wild-type wild-type

wild-type wild-type

wild-type wild-type

wild-type no mutations detected wild-type

wild-type no mutations detected wild-type

wild-type no mutations detected wild-type no mutations detected wild-type no mutations detected wild-type no mutations detected wild-type

no mutations detected wild-type

no mutations detected wild-type

no mutations detected wild-type wild-type wild-type wild-type wild-type wild-type wild-type

no mutations detected

no mutations detected

no mutations detected

no mutations detected

no mutations detected

no mutations detected no mutations detected no mutations detected no mutations detected no mutations detected no mutations detected no mutations detected

no mutations detected no mutations detected no mutations detected wild-type

no mutations detected no mutations detected no mutations detected wild-type

no mutations detected no mutations detected wild-type no mutations detected wild-type wild-type (99%) wild-type wild-type wild-type wild-type wild-type (98%) wild-type wild-type (100%) wild-type wild-type wild-type

wild-type

wild-type

wild-type

no mutations detected

no mutations detected

no mutations detected wild-type wild-type wild-type no mutations detected no mutations detected no mutations detected no mutations detected no mutations detected no mutations detected no mutations detected no mutations detected no mutations detected no mutations detected no mutations detected no mutations detected wild-type wild-type wild-type wild-type wild-type wild-type

wild-type no mutations detected wild-type

wild-type no mutations detected wild-type

wild-type no mutations detected wild-type wild-type no mutations detected wild-type no mutations detected wild-type no mutations detected

no mutations detected wild-type no mutations detected

no mutations detected wild-type no mutations detected

no mutations detected wild-type no mutations detected wild-type wild-type wild-type

All 67 methods reported results for 16/266, the nominal wild-type KRAS codons 12 and 13 material. All methods used similar terminology in reporting, broadly ‘wild-type’ or ‘no mutations detected’. Method 38 reported a 3’UTR variant (KRAS c.*2505T>G) at 49.71% (sample 7), 51.26% (sample 10), and 49.86% (sample 19; orange); method 17 reported a mutation in KRAS codon 59 (KRAS c.175G>A, or c.176C>A, or c.176C>G, resulting in KRAS p.A59E, or p.A59G, or p.A59T) in sample 7 only (green; since retested as wild-type for codons 12, 13, 59, 61, 117, and 146).