Scandinavian Journal of Clinical & Laboratory Investigation, 2014; 74: 44–52
ORIGINAL ARTICLE
Automation of diagnostic genetic testing: Mutation detection by cyclic minisequencing
Scand J Clin Lab Invest Downloaded from informahealthcare.com by National Taiwan University on 01/12/15 For personal use only.
KATARIINA ALAGRUND & ARTO K. ORPANA Helsinki University Central Hospital, HUSLAB, Laboratory of Genetics, Helsinki, and Department of Clinical Chemistry, University of Helsinki, Finland Abstract Background. The rising role of nucleic acid testing in clinical decision making is creating a need for efficient and automated diagnostic nucleic acid test platforms. Clinical use of nucleic acid testing sets demands for shorter turnaround times (TATs), lower production costs and robust, reliable methods that can easily adopt new test panels and is able to run rare tests in random access principle. Here we present a novel home-brew laboratory automation platform for diagnostic mutation testing. Method. This platform is based on the cyclic minisequecing (cMS) and two color near-infrared (NIR) detection. Pipetting is automated using Tecan Freedom EVO pipetting robots and all assays are performed in 384-well micro plate format. The automation platform includes a data processing system, controlling all procedures, and automated patient result reporting to the hospital information system. Conclusions. We have found automated cMS a reliable, inexpensive and robust method for nucleic acid testing for a wide variety of diagnostic tests. The platform is currently in clinical use for over 80 mutations or polymorphisms. Additionally to tests performed from blood samples, the system performs also epigenetic test for the methylation of the MGMT gene promoter, and companion diagnostic tests for analysis of KRAS and BRAF gene mutations from formalin fixed and paraffin embedded tumor samples. Automation of genetic test reporting is found reliable and efficient decreasing the work load of academic personnel. Key Words: Genetic diseases, inborn, genetic testing, genetic variation, genetics, medical, routine diagnostic tests
Introduction The role of nucleic acid testing is rising in health care and the demand for automated high-throughput diagnostic genetic testing is increasing. A need has grown for simple and inexpensive, yet robust and flexible, systems for analyzing a limited number of mutations and polymorphisms, which is typically the case in clinical questions. Our laboratory, like many other diagnostic laboratories, is facing bipolar demand for genetic tests. Sample numbers of the frequently requested common tests are high and rapidly rising, while sample numbers for many of the rarely requested tests in the test selection are low, because they are tests for the rare severe inherited diseases. The population in Finland carry numerous founder mutations, seen only locally and often only in a restricted area of Finland [1]. Aspartylglucosaminuria (AGU) is one example of these: One mutation is responsible for over 98% of disease cases
in Finnish population [2,3]. Besides AGU, over 30 recessively inherited diseases, together known as the Finnish Disease Heritage with local founder mutations exist [1,4–7]. There is no commercial interest towards these tests, but for the patient they are very important and thus diagnostic, carrier detection and predictive testing must be available for these mutations. Pharmacogenetic tests and companion diagnostics are rapidly developing fields in genetics, as personalized medicine becomes more common [8–10]. One example, serving large patient group, is the s ingle-nucleotide polymorphism rs4363657 located within SLCO1B1 gene. It is used as a pharmacogenetic test defining suitability of Simvastatin as cholesterol level lowering drug [11]. Companion testing is required when a cancer drug is designed to attack tumor cells carrying (or not carrying) a certain type of a genetic variation. Characteristically these are fusion proteins expressed in uncontrolled way, or
Correspondence: Arto K. Orpana, PhD, HUSLAB Laboratory of Genetics, POB 140, FI-00290 Helsinki, Finland. Tel: ⫹ 358 50 4270647. Fax: ⫹ 358 9 47174001. E-mail: [email protected] (Received 8 May 2013 ; accepted 12 October 2013) ISSN 0036-5513 print/ISSN 1502-7686 online © 2014 Informa Healthcare DOI: 10.3109/00365513.2013.857040
Scand J Clin Lab Invest Downloaded from informahealthcare.com by National Taiwan University on 01/12/15 For personal use only.
Automation of diagnostic genetic testing modified functional proteins, e.g. tyrosine kinases carrying mutations keeping them uncontrollable active. Typical examples are Kirsten rat sarcoma 2 viral oncogene homolog (KRAS) and v-raf murine sarcoma viral oncogene homolog B1 (BRAF) gene mutations found in numerous cancers including colon, lung and breast cancer and melanoma [12–15]. Epigenetic modification of DNA is involved in drug selection for treatment of gliomas. Methylation of the promoter of O(6)-methylguanine-DNA methyltransferase (MGMT) gene associates with lower activity of this DNA repair enzyme and better response to nucleotide analog chemotherapy and overall prognosis [16]. There are currently numerous analytical technologies, most of them relying on quantitative PCR or hybridization of probes. The hybridization assays may also be enzyme-assisted using a nucleic acid modifying enzyme as part of the system. One such type of assay is the single nucleotide primer extension, or solid phase minisequencing, in which the nucleotide recognition specificity of the DNA polymerase is used for identification of the nucleotide sequence in the template strand [17]. However, both high-throughput tests and on-the-need-based combinations of the rare mutation tests require novel laboratory automation in order to respond to increasing throughput and economic demands. The sensitivity of single nucleotide primer extension tests depends on the detection method. Radioactive detection, which was previously widely used in many nucleic acid tests, has been replaced by fluorescent detection technologies. The fluorescent detection has major advantages: No radiation risk, no decay and a possibility to detect multiple reactions simultaneously using different fluorophores. However, for many applications conventional fluorescence is not sensitive enough [18]. In this paper we present our analytical platform and the use of near-infrared (NIR)-labeled reagents for nucleic acid testing. These reagents are available in multiple sources, they have very high quantum yields, and as the tissue background fluorescence in the NIR wave lengths is negligible, orders of magnitude higher sensitivity can be reached than by using conventional fluorophores.
Materials and methods Samples The new platform was validated with duplicate diagnostic samples and controls used in our laboratory routinely. Materials The 348-well micro plates Thermo-Fast®384 (Thermo Scientific, Finland), custom designed
45
PCR primers (Sigma-Aldrich, Finland), DNA Polymerase DyNAZyme (Finnzymes, Thermo Scientific, Finland), Exonuclease I (Exo I) and Shrimp alkaline phosphatase (SAP) enzymes (Fermentas, Thermo Scientific, Finland), custom designed, biotinylated detection primers (SigmaAldrich, Finland). Cyanine-5- dNTP and Fluorescein-12-dNTP (PerkinElmer, Finland), Monoclonal Anti-Cy5 antibody (Sigma-Aldrich, Finland), Anti-Fluorescein/Oregon Green IgG (Invitrogen, Thermo Scientific, Finland), IRDye680CW and IRDye800CW (LI-COR, Thermo Scientific, Finland), Odyssey Blocking Buffer (LI-COR, Thermo Scientific, Finland), custom made clearbottom black well streptavidin-coated ViewPlate384 detection plates (PerkinElmer, Finland).
Instruments PCR amplifications were performed with GeneAmp® PCR System 9700 (Applied Biosystems®, Finland) cyclers carrying dual 384-well blocks. The robotic platform utilized one Tecan Freedom EVO 100 and three EVO 150 pipetting robots (PLD, Finland). NIR-detection was performed with LI-COR Aerius Automated Imaging System (LI-COR, Thermo Scientific, Finland).
Software and databases The Autoreport IT solution for automated generation of genotyping results and reports to the patients was programmed using Visual Basic for Applications (VBA) scripts and multiple databases. As a whole they formed a laboratory automation system (LAS). The cyclic minisequencing (cMS) assay work flow initiated by technician producing a work list in the hospital laboratory information management system. Based on tests selected into this work list the program created sample tables and pipetting maps, and calculated reagent mixes. Every test run could contain 1–8 different tests consisting of 1–24 mutations or variations in one or multiple genes ad hoc. All assay parameters were stored in a database, and no programming skills were required from the technicians. Three files were stored. The first contained the original patient data and was used for final reporting. Assay parameters were stored as CSV file on USB memory stick which was carried along the samples and assay plates between multiple Tecan robots during the procedure. The reagent mixture pipetting commands were stored on the same USB stick in the third file and used for robotic reagent mixing. The parameter files were read by Tecan Freedom Evoware software, and Tecan VBScripts were called for handling these parameters. No interventions were required by the technician.
Scand J Clin Lab Invest Downloaded from informahealthcare.com by National Taiwan University on 01/12/15 For personal use only.
46
K. Alagrund & A. K. Orpana
At the end of the assay run the fluorescence results from the IR fluorometer were fetched into the original file containing the patient data, and data analysis and genotype calling was performed for each mutation using automated VBA scripts. The control value database selected results from control samples of 20 previous test runs and calculated cut-off values for automated genotype calling. These cut-off values could be changed by the person performing genotyping. Called genotypes were coded and after analyzing all variations of the test, the result code strings (RCS) were formed for each patient. In addition to genotype results also information about the age and gender from the patient’s test request were used for creation of age and gender group codes. For creation of patient reports this RCS was used for lookup from report component database. Six report components, the test specific background chapter, the obtained genotype result, the test specific interpretation, the genotype specific interpretation, the genotype specific closing sentence and finally, the test specific closing sentence, were combined. If age group or gender specific components existed, they were used as they were coded into the RCS and in that case also the database contained specific components for them. Combined patient reports were sent to the work list, visually inspected and accepted individually before transferring to the hospital patient records.
bisulfite conversion that converted unmethylated cytosine residues to uracils but leaved methylated residues unchanged. EpiTect Bisulfite Kit (Qiagen, Finland) was used in conversion.
PCR When solid phase minisequencing (MS) tests were converted to cMS tests their PCR conditions were harmonized. If needed, the PCR primers were redesigned to fulfill strict amplification condition requirements. Practically all tests could be performed using a single PCR program, which was a prerequisite for random access testing principle and possibility of free mixing of tests on 384-well PCR plate. All samples and positive controls were loaded to cooled IC22 Two Position Chilling Dry Bath blocks (Torrey Pines Scientific, Inc, CA, USA) on the deck of a Tecan robotic workstation. Pipetting protocol was programmed to minimize the use of the valuable positive control material. In the high-throughput single mutation assays samples and controls were pre-ordered on a 96-well micro plate during automated preanalytical processing. In both single and multi-mutation PCR systems Tecan Freedom EVO 150 robot pipetted reagents, controls and samples for PCR reactions without any intervention by the technician, based on the parameter file on an USB stick.
Preanalytics The high-throughput cMS tests were performed using DNA released from EDTA blood dried on paper disks. It was inexpensive, fast and suitable if the number of individual PCR reactions was low, such as in the single polymorphism (-13190 C⬎T, rs 4988234) test for lactose malabsorption [19]. Because of a relatively large number of samples, this preanalytical process was automated. The system was able to process 192 samples (2 ⫻ 96 sample plates) in a single run. Tube order was verified by bar code reader before pipetting. The Tecan Freedom EVO 100 robot transferred an aliquot of blood from the original EDTA blood tubes to the blotting paper spot on the bottom of a 96-well micro plate. The plates were allowed to dry. This was an automated version of the original ‘Stixing’ method [20]. For tests requiring higher amounts of DNA, extraction was partly automated with Maxwell® 16 Instrument (Promega, Biofellows, Finland) following manufacturer’s instructions. DNA was quantified spectrophotometrically using NanoDrop ND-1000 Spectrophotometer (Thermo Scientific, Finland)
Bisulfite conversion in epigenetic tests Methylation of the promoter of the MGMT gene cytosines of the CpG islands was detectable after
cMS PCR product was purified by incubation with exonuclease I (ExoI) and shrimp alkaline phosphatase (SAP) enzymes, which were automatically pipetted on a micro plate by Tecan Freedom EVO 150 robot. Reagent preparation for cMS reaction was partly automated. Detection primers required in current analysis were loaded to a reagent block and labeled nucleotides set in fixed order in another block. The Tecan Freedom EVO 150 robot pipetted primers and nucleotides to reagent tubes and just prior to use the master mixture containing aqua, PCR buffer and DNA polymerase was pipetted manually to these reagent tubes. The wild type and mutant cyclic single nucleotide primer extension reactions of cMS were performed in duplicates. Biotinylated detection primers and Fluoresceine- and Cy5-labeled nucleotides were used for wild type and mutant reactions, respectively. Separate reactions allowed us to use inexpensive dNTP instead of ddNTP and conventional DNA polymerase. During the reaction the DNA polymerase incorporated only the right nucleotide matching to the template sequence of the PCR product. As described above, one PCR reaction, containing one mutation site was divided into four reaction wells during cMS, as the wild type reactions and the
Scand J Clin Lab Invest Downloaded from informahealthcare.com by National Taiwan University on 01/12/15 For personal use only.
Automation of diagnostic genetic testing mutant reactions were prepared separately and in duplicates. In the detection stage one wild type and one mutant reaction were combined together, ending up with two detection wells for each sample. The wild type and the mutant reaction products were combined in a streptavidin coated detection plate and incubated at 37°C, during which the biotin-labeled detection primers were trapped to the bottom of the well. Automated procedure was continued using primary and NIR-labeled secondary antibodies. Although being fluorophores themselves, Fluorescein- and Cy5-labeled nucleotides were not found sensitive enough in this type of an assay and were used here as antigens, because no NIR-labeled dNTPs were yet available. Fluorophore-binding primary antibodies and NIR-labeled secondary antibodies were used for signal amplification. After the final washing step the plate was automatically transferred by Tecan Freedom EVO 150 robot to the LI-COR Aerius Automated Imaging System where the signal from the wild type reaction was detected at the 700 nm channel and the mutant reaction signal was detected at the 800 nm channel.
47
Relative fluorescence units (RFU) were stored in a file for genotyping. The initial file created for sample pipetting was re-opened and results were fetched into this file via laboratory intranet. The results of the control samples were saved in a control value database. After all the results in the test run were analyzed and approved, the program automatically created complete patient reports based on specific codes created by information in the electronic request and results of the genotyping assays. These statements were exported to customers electrically or in paper. All information was archived as files. The cMS process is presented in Figure 1.
Results Validation We developed and validated a robust analytical platform based on an automation friendly modification of the traditional minisequencing method. This automated genotyping system included sample preprocessing, PCR-amplification, cMS with NIR-detection, and automated data processing and result reporting
Figure 1. A simplified diagram of the stages in the cMS process: data processing, preprocessing, PCR, purification, cMS and detection. Tecan 1: Tecan Freedom EVO 100 used in preprocessing, Tecan 2: Tecan Freedom EVO 150 used in PCR-pipetting, Tecan 3 and Tecan 4: Tecan Freedom EVO 150 robots used in cMS steps.
48
K. Alagrund & A. K. Orpana
Scand J Clin Lab Invest Downloaded from informahealthcare.com by National Taiwan University on 01/12/15 For personal use only.
system. The system was gradually created and validated during several years and enables us now to add new tests and mutations to the test panels after a short specific validation period. All clinical validations were made according to the diagnostic laboratory protocol based on collaboration with the clinicians. The tests were validated for multiple types of DNA samples: Paraffin embedded tissue blocks, fresh tissue, EDTA blood, dried on paper spots, and cell lysates depending on expected sample type. Tissue sample types tested and found suitable for cMS originated from liver, muscle and placenta. Also
samples of amniotic fluid cells and buccal swabs were successfully analyzed. Genotyping The genotyping was based on the ratio (R value) between wild type and mutant well signals, calculated by the LAS automatically. The R values optimally were ⬍ 0.1 in wild-type samples, around 1 in heterozygous samples and ⬎ 10 in homozygous samples. However, there were great variations in R values between mutation sites due to different amounts of
Table I. Diseases and epigenetic modifications included in the cMS test panels. Symbol
Name/Phenotype
One-mutation panel LAKT Lactose malabsorption Two-mutation panel FV Thrombotic risk evaluation FII Thrombotic risk evaluation Sub acute test panel HFE Hemochromatosis TPMT Thiopurine S-methyltransferase Varfa Warfarin anticoagulant response SLCO1B1 Statin induced myopathy Random access principle multi-mutation test panel FDH and LQTs AGU Aspartylglucosaminuria APECED Autoimmunepolyendocrine syndrome type 1 CCD Cleidocranialdysplasi CLD Chloride diarrhea, congenital, Finnish type CLN5 Ceroidlipofuscinosis, neuronal, 5 CNF Nephrotic syndrome, type 1 Cohen Cohen syndrome DTD Diastrophic dysplasia Gelso Amyloidosis, Finnish type GRACILE GRACILE syndrome HLS Hydrolethalus syndrome INCL Ceroidlipofuscinosis, neuronal, 1 IOSCA Infantile-onset spinocerebellar ataxia JNCL Ceroidlipofuscinosis, neuronal, 3 MEB Muscle-eye-braindisease MKS Meckel syndrome 1, Meckel syndrome 6 MUL Mulibreynanism SD Salla disease TMD Tibial musculardystrophy, tardive USH3 Usher syndrome, type 3A XLRS Retinoschisis 1, X-linked, Juvenile LQT1 Long QT syndrome 1 LQT2 Long QT syndrome 2 BRCA BRCA1 Breast cancer type 1 BRCA2 Breast cancer type 2 CYP2D6 CYP2D6 Cytochrome P450, Subfamily IID, Polypeptide 6 KRAS-BRAF panel KRAS V-Ki-ras2 Kirsten rat sarcoma viral oncogene homolog BRAF V-raf murine sarcoma viral oncogene homolog B1 Epigenetic panel MGMT O(6)-methylguanine-DNA-methyltransferrase
Gene
Number of mutation sites
LCT
1
F5 F2
1 1
HFE TPMT CYP2C9 VKORC1 SLCO1B1
2 5 2 2 1
AGA AIRE RUNX2 SLC26A3 CLN5 NPHS1 COH1 SLC26A2 GSN BCS1L HYLS1 PPT1 C10ORF2 CLN3 POMGNT1 MKS1, CC2D2A TRIM37 SLC17A5 TTN CLRN1 RS1 KCNQ1 KCNH2
2 3 1 1 1 2 1 1 1 1 1 1 1 2 1 1 1 1 1 1 1 3 2 2
BRCA1 BRCA2
9 8
CYP2D6
8
KRAS
7
BRAF
2
MGMT
5
Scand J Clin Lab Invest Downloaded from informahealthcare.com by National Taiwan University on 01/12/15 For personal use only.
Automation of diagnostic genetic testing nucleotides incorporated and the local sequence environment. Typically a ten-fold R value difference between genotypes was obtained. The autovalidation part of LAS calculated R value ranges based on the results of 20 previous control samples. Diagnosis of inherited mutations and polymorphisms was based on the R value, but when a more quantitative result was needed, as e.g. in epigenetic modification or companion testing (MGMT, KRAS, BRAF) the results were calculated as a relative amount of the mutant signal. This Percentage value was more sensitive when testing acquired mutations samples with low amounts of tumor tissue. Assays for seven mutations in KRAS gene and two in BRAF gene were validated for clinical use. Methylation of the promoter of the MGMT gene cytosines of the CpG islands was detectable after bisulfite conversion. Commonly used methods, such as methyl-specific PCR or pyrosequencing also lean on bisulfate treated DNA [21]. We validated five reported cytosine methylation sites and one cytosine control site in the promoter region of the MGMT gene [22]. The tests analyzing mutation or epigenetic status of a tumor tissue needed to reach high sensitivity. The amount of affected DNA varies between almost none to nearly 100% in tissue samples taken from the tumor area. Using cMS we could detect down to 5% of affected DNA in tumor samples in most mutation sites. Autoreport In result reporting an important factor was presenting the test results to the clinician in an unambiguous, understandable form. Using a strategy in which patient information from the request and the multiple genotyping results formed a result code string, which was then used for creation of the final report allowed us to create very specific reports. The genotyping information was not always sufficient, as clinicians were used to alleles, not genotypes. In the pharmacogenetic tests which explored the inherited forms of metabolic pathways, the metabolism of a therapeutic drug was classified based on alleles which were results of various mutation combinations [23]. The deficiency of Thiopurine-S-Methyltransferase (TPMT) predisposes to severe side effects of thiopurine [24,25]. Several allele combinations cause severe reactions with therapeutic dose of drug. In clinical use 18 allele combinations (alleles *1,*2, *3A, 3B, *3C and *16) are significant and frequently found in Finnish population. The Autoreport system could create automatic allele specific patient reports. This service of the clinicians was taken one step further in VarfaD testing. Warfarin is a K-vitamin antagonist used to prevent excessive blood clotting [26]. The VarfaD test based on two VKORC1- and two CYP2C9- polymorphisms was used to deter-
49
Figure 2. (A) Sample volumes in MS process and cMS process during 2008–2011. Number of the tests in the cMS process on the left Y-axis and number of the tests in the MS process on the rigth Y-axis. (B) Conversion from MS to cMS during 2008–2011 did not affect the cMS process mean TATs.
mine the Warfarin metabolic status and initial Warfarin dose of the patient [23]. The medical statements based on CYP2C9*1-, CYP2C9*2-, CYP2C9*3- and, two VKORC1- alleles were created similarly to TPMT above. Based on the age, weight and the genotyping result of the patient the program calculated an initial Warfarin dosing recommendation helping physicians to estimate the Warfarin dosage. These tests are two examples of the sophisticated automatic reporting system of this assay platform. Use of automatic report creation significantly saved working time in routine as specialist’s intervention was not needed in great majority of cases. Current status To date we have successfully validated over 40 diagnostic DNA tests, altogether over 100 mutations into this platform. Tests currently included in the cMS process are listed in Table I. More than 30,000 autoreports have been created by the system.The throughput could be improved further, as up to four different cMS runs could be combined into a single detection run. In this format 768 patient samples could be analyzed in duplicates in 4 h with very little manual intervention. First results would be available in
50
K. Alagrund & A. K. Orpana
1.5 h in case of urgent samples, but typically series would be automatically processed overnight and analyzed next morning. The genetic tests for inherited diseases based on cMS were accredited by FINAS (Finnish Accreditation Service) according to ISO EN 15189 standard.
sion of manual MS tests to automated cMS format enabled maintenance of stabile TATs despite of significantly larger test selection and rising sample amounts (Figure 2A and B).
Precision Effects on turnaround times (TATs)
Comparing 90 consecutive series of the lactose malabsorption test during one year the R-values of the genotypes remained stable despite of the minor variation between series (Figure 3A). The variation resulted
Scand J Clin Lab Invest Downloaded from informahealthcare.com by National Taiwan University on 01/12/15 For personal use only.
During the four-year period of automation, method development and conversion the growth in analyzed sample amount in our laboratory was 23%. Conver-
Figure 3. (A) Mean R-value of genotypes ⫾ 1 SD of 90 consecutive lactase series; wild type (N/N), heterozygote (M/N), homozygote (M/M). (B) Mean R-value of genotypes normalized as for M/N genotype, 90 lactase series; wild type (N/N), heterozygote (M/N), homozygote (M/M).
Automation of diagnostic genetic testing mostly from changing reagent lots, and assay conditions. This variation usually affected the signal levels of each genotype, shifting all values simultaneously. If this variation was abolished by normalization, remaining technical variation was very small (Figure 3B).
[4]
[5]
[6]
Scand J Clin Lab Invest Downloaded from informahealthcare.com by National Taiwan University on 01/12/15 For personal use only.
Discussion The introduction of nearly total laboratory automation in our diagnostic genetic testing enabled analysis of large amounts of samples without addition of laboratory personnel and improved turnaround times. This was based on two advantages: The ability to perform high-throughput series with large sample volumes and, on the other hand, combine up to multiple infrequent mutations randomly in one test run. Automated cyclic minisequencing combined with antibody-based near-infrared detection was inexpensive and highly sensitive. Ability to use reagents from multiple vendors aided in avoiding single vendor trap and provided security to the process. The IT developments based on electronic data transfer and multiple databases, including automatically updated control value database creating genotype cut offs, automated and simplified processes significantly. The current tendency is to avoid manual data analysis and reporting, and thereby minimize human errors and labor also in genetic laboratories. Automatically generated patient reports dramatically dropped TATs as no manual typing was required. The purpose of this work was to develop a robust and economical automated robotic based method to decrease laborious manual laboratory work and reduce reagent costs. The cMS system could be used as proof-of-principle process for developing a complete walk-a-way analyzer and bring total laboratory automation also into diagnostic mutation testing.
[7] [8]
[9] [10]
[11]
[12] [13]
[14]
[15]
[16]
Acknowledgements We thank the staff at the HUSLAB Laboratory of Genetics for technical assistance. Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.
[17]
[18]
[19]
References [1] Peltonen L. Molecular background of the Finnish disease heritage. Ann Med 1997;29:553–6. [2] Valkonen S, Hietala M, Savontaus ML, Aula P. Origin of Finnish mutations causing aspartylglucosaminuria. Hereditas 1999;131:191–5. [3] Saarela J, Laine M, Oinonen C, von Schantz C, Jalanko A, Rouvinen J, Peltonen, L. Molecular pathogenesis of a
[20]
[21]
51
disease: structural consequences of aspartylglucosaminuria mutations. Hum Mol Genet 2001;10:983–95. Hietala M, Gron K, Syvanen AC, Peltonen L, Aula P. Prospects of carrier screening of aspartylglucosaminuria in Finland. Eur J Hum Genet 1993;1:296–300. Peltonen L, Jalanko A, Varilo T. Molecular genetics of the Finnish disease heritage. Hum Mol Genet 1999;8: 1913–23. Peltonen L, Pekkarinen P, Aaltonen J. Messages from an isolate: lessons from the Finnish gene pool. Biol Chem Hoppe Seyler 1995;376:697–704. Sipila K, Aula P. Database for the mutations of the Finnish disease heritage. Hum Mutat 2002;19:16–22. Cohen J, Wilson A, Manzolillo K. Clinical and economic challenges facing pharmacogenomics. Pharmacogenomics J 2013;4:378–88. Snyderman R. Personalized health care: From theory to practice. Biotechnol J 2012;7:973–9. Ginsburg GS, Willard HF. Genomic and personalized medicine: foundations and applications. Transl Res 2009; 154:277–87. SEARCH Collaborative Group,Link E, Parish S, Armitage J, Bowman L, Heath S, Matsuda F, Gut I, Lathrop M, Collins R. SLCO1B1 variants and statininduced myopathy – a genomewide study. N Engl J Med 2008;359:789–99. Kranenburg O. The KRAS oncogene: past, present, and future. Biochim Biophys Acta 2005;1756:81–2. Haigis KM, Kendall KR, Wang Y, Cheung A, Haigis MC, Glickman JN, Niwa-Kawakita M, Sweet-Cordero A, Sebolt-Leopold J, Shannon KM, Settleman J, Giovannini M, Jacks T. Differential effects of oncogenic K-Ras and N-Ras on proliferation, differentiation and tumor progression in the colon. Nat Genet 2008;40:600–8. Amanuel B, Grieu F, Kular J, Millward M, Iacopetta B. Incidence of BRAF p.Val600Glu and p.Val600Lys mutations in a consecutive series of 183 metastatic melanoma patients from a high incidence region. Pathology 2012; 44:357–9. Davies H, Bignell GR, Cox C, Stephens P, Edkins S, Clegg S, Teague J, Woffendin H, Garnett MJ, Bottomley W, Davis N, Dicks E, Ewing R, Floyd Y, Gray K, Hall S, Hawes R, Hughes J, Kosmidou V, Menzies A, Mould C, Parker A, Stevens C, Watt S, Hooper S, Wilson R, Jayatilake H, Gusterson BA, Cooper C, Shipley J, Hargrave D, Pritchard-Jones K, Maitland N, ChenevixTrench G, Riggins GJ, Bigner DD. Mutations of the BRAF gene in human cancer. Nature 2002; 417:949–54. Esteller M, Garcia-Foncillas J, Andion E, Goodman SN, Hidalgo OF, Vanaclocha V, Baylin SB, Herman JB. Inactivation of the DNA-repair gene MGMT and the clinical response of gliomas to alkylating agents. N Engl J Med 2000;343:1350–4. Jalanko A, Kere J, Savilahti E, Schwartz M, Syvanen AC, Ranki M, Soderlund H. Screening for defined cystic fibrosis mutations by solid-phase minisequencing. Clin Chem 1992;38:39–43. Orpana AK. Fluorescence resonance energy transfer (FRET) using ssDNA binding fluorescent dye. Biomol Eng 2004; 21:45–50. Enattah NS, Sahi T, Savilahti E, Terwilliger JD, Peltonen L, Jarvela I. Identification of a variant associated with adulttype hypolactasia. Nat Genet 2002;30:233–7. Ihalainen J, Siitari H, Laine S, Syvanen AC, Palotie A. Towards automatic detection of point mutations: use of scintillating microplates in solid-phase minisequencing. BioTechniques 1994;16:938–94. Karayan-Tapon L, Quillien V, Guilhot J, Wager M, Fromont G, Saikali S, Etcheverry A, Hamlat A, Loussouarn D, Campion L, Campone M, Vallette FM, Gratas-Rabbia-Re C. Prognostic value of O6-methylguanine-DNA
52
K. Alagrund & A. K. Orpana
Scand J Clin Lab Invest Downloaded from informahealthcare.com by National Taiwan University on 01/12/15 For personal use only.
methyltransferase status in glioblastoma patients, assessed by five different methods. J Neurooncol 2010;97:311–22. [22] Nakagawachi T, Soejima H, Urano T, Zhao W, Higashimoto K, Satoh Y, Matsukura S, Kudo S, Kitajima Y, Harada H, Furukawa K, Matsuzaki H, Emi M, Nakabeppu Y, Miyazaki K, Sekiguchi M, Mukai T. Silencing effect of CpG island hypermethylation and histone modifications on O6-methylguanine-DNA methyltransferase (MGMT) gene expression in human cancer. Oncogene 2003;22:8835–44. [23] Takahashi H, Wilkinson GR, Nutescu EA, Morita T, Ritchie MD, Scordo MG, Pengo V, Barban M, Padrini R, Ieiri I, Otsubo K, Kashima T, Kimura S, Kijima S, Echizen H. Different contributions of polymorphisms in VKORC1 and CYP2C9 to intra- and inter-population differences in
maintenance dose of warfarin in Japanese, Caucasians and African-Americans. Pharmacogenet Genomics 2006;16: 101–10. [24] Coulthard S, Hogarth L. The thiopurines: an update. Invest New Drugs 2005;23:523–32. [25] Haglund S, Lindqvist M, Almer S, Peterson C, Taipalensuu J. Pyrosequencing of TPMT alleles in a general Swedish population and in patients with inflammatory bowel disease. Clin Chem 2004;50:288–95. [26] Ross KA, Bigham AW, Edwards M, Gozdzik A, Suarez-Kurtz G, Parra EJ. Worldwide allele frequency distribution of four polymorphisms associated with warfarin dose requirements. J Hum Genet 2010;55: 582–9.
ORIGINAL ARTICLE
Automation of diagnostic genetic testing: Mutation detection by cyclic minisequencing
Scand J Clin Lab Invest Downloaded from informahealthcare.com by National Taiwan University on 01/12/15 For personal use only.
KATARIINA ALAGRUND & ARTO K. ORPANA Helsinki University Central Hospital, HUSLAB, Laboratory of Genetics, Helsinki, and Department of Clinical Chemistry, University of Helsinki, Finland Abstract Background. The rising role of nucleic acid testing in clinical decision making is creating a need for efficient and automated diagnostic nucleic acid test platforms. Clinical use of nucleic acid testing sets demands for shorter turnaround times (TATs), lower production costs and robust, reliable methods that can easily adopt new test panels and is able to run rare tests in random access principle. Here we present a novel home-brew laboratory automation platform for diagnostic mutation testing. Method. This platform is based on the cyclic minisequecing (cMS) and two color near-infrared (NIR) detection. Pipetting is automated using Tecan Freedom EVO pipetting robots and all assays are performed in 384-well micro plate format. The automation platform includes a data processing system, controlling all procedures, and automated patient result reporting to the hospital information system. Conclusions. We have found automated cMS a reliable, inexpensive and robust method for nucleic acid testing for a wide variety of diagnostic tests. The platform is currently in clinical use for over 80 mutations or polymorphisms. Additionally to tests performed from blood samples, the system performs also epigenetic test for the methylation of the MGMT gene promoter, and companion diagnostic tests for analysis of KRAS and BRAF gene mutations from formalin fixed and paraffin embedded tumor samples. Automation of genetic test reporting is found reliable and efficient decreasing the work load of academic personnel. Key Words: Genetic diseases, inborn, genetic testing, genetic variation, genetics, medical, routine diagnostic tests
Introduction The role of nucleic acid testing is rising in health care and the demand for automated high-throughput diagnostic genetic testing is increasing. A need has grown for simple and inexpensive, yet robust and flexible, systems for analyzing a limited number of mutations and polymorphisms, which is typically the case in clinical questions. Our laboratory, like many other diagnostic laboratories, is facing bipolar demand for genetic tests. Sample numbers of the frequently requested common tests are high and rapidly rising, while sample numbers for many of the rarely requested tests in the test selection are low, because they are tests for the rare severe inherited diseases. The population in Finland carry numerous founder mutations, seen only locally and often only in a restricted area of Finland [1]. Aspartylglucosaminuria (AGU) is one example of these: One mutation is responsible for over 98% of disease cases
in Finnish population [2,3]. Besides AGU, over 30 recessively inherited diseases, together known as the Finnish Disease Heritage with local founder mutations exist [1,4–7]. There is no commercial interest towards these tests, but for the patient they are very important and thus diagnostic, carrier detection and predictive testing must be available for these mutations. Pharmacogenetic tests and companion diagnostics are rapidly developing fields in genetics, as personalized medicine becomes more common [8–10]. One example, serving large patient group, is the s ingle-nucleotide polymorphism rs4363657 located within SLCO1B1 gene. It is used as a pharmacogenetic test defining suitability of Simvastatin as cholesterol level lowering drug [11]. Companion testing is required when a cancer drug is designed to attack tumor cells carrying (or not carrying) a certain type of a genetic variation. Characteristically these are fusion proteins expressed in uncontrolled way, or
Correspondence: Arto K. Orpana, PhD, HUSLAB Laboratory of Genetics, POB 140, FI-00290 Helsinki, Finland. Tel: ⫹ 358 50 4270647. Fax: ⫹ 358 9 47174001. E-mail: [email protected] (Received 8 May 2013 ; accepted 12 October 2013) ISSN 0036-5513 print/ISSN 1502-7686 online © 2014 Informa Healthcare DOI: 10.3109/00365513.2013.857040
Scand J Clin Lab Invest Downloaded from informahealthcare.com by National Taiwan University on 01/12/15 For personal use only.
Automation of diagnostic genetic testing modified functional proteins, e.g. tyrosine kinases carrying mutations keeping them uncontrollable active. Typical examples are Kirsten rat sarcoma 2 viral oncogene homolog (KRAS) and v-raf murine sarcoma viral oncogene homolog B1 (BRAF) gene mutations found in numerous cancers including colon, lung and breast cancer and melanoma [12–15]. Epigenetic modification of DNA is involved in drug selection for treatment of gliomas. Methylation of the promoter of O(6)-methylguanine-DNA methyltransferase (MGMT) gene associates with lower activity of this DNA repair enzyme and better response to nucleotide analog chemotherapy and overall prognosis [16]. There are currently numerous analytical technologies, most of them relying on quantitative PCR or hybridization of probes. The hybridization assays may also be enzyme-assisted using a nucleic acid modifying enzyme as part of the system. One such type of assay is the single nucleotide primer extension, or solid phase minisequencing, in which the nucleotide recognition specificity of the DNA polymerase is used for identification of the nucleotide sequence in the template strand [17]. However, both high-throughput tests and on-the-need-based combinations of the rare mutation tests require novel laboratory automation in order to respond to increasing throughput and economic demands. The sensitivity of single nucleotide primer extension tests depends on the detection method. Radioactive detection, which was previously widely used in many nucleic acid tests, has been replaced by fluorescent detection technologies. The fluorescent detection has major advantages: No radiation risk, no decay and a possibility to detect multiple reactions simultaneously using different fluorophores. However, for many applications conventional fluorescence is not sensitive enough [18]. In this paper we present our analytical platform and the use of near-infrared (NIR)-labeled reagents for nucleic acid testing. These reagents are available in multiple sources, they have very high quantum yields, and as the tissue background fluorescence in the NIR wave lengths is negligible, orders of magnitude higher sensitivity can be reached than by using conventional fluorophores.
Materials and methods Samples The new platform was validated with duplicate diagnostic samples and controls used in our laboratory routinely. Materials The 348-well micro plates Thermo-Fast®384 (Thermo Scientific, Finland), custom designed
45
PCR primers (Sigma-Aldrich, Finland), DNA Polymerase DyNAZyme (Finnzymes, Thermo Scientific, Finland), Exonuclease I (Exo I) and Shrimp alkaline phosphatase (SAP) enzymes (Fermentas, Thermo Scientific, Finland), custom designed, biotinylated detection primers (SigmaAldrich, Finland). Cyanine-5- dNTP and Fluorescein-12-dNTP (PerkinElmer, Finland), Monoclonal Anti-Cy5 antibody (Sigma-Aldrich, Finland), Anti-Fluorescein/Oregon Green IgG (Invitrogen, Thermo Scientific, Finland), IRDye680CW and IRDye800CW (LI-COR, Thermo Scientific, Finland), Odyssey Blocking Buffer (LI-COR, Thermo Scientific, Finland), custom made clearbottom black well streptavidin-coated ViewPlate384 detection plates (PerkinElmer, Finland).
Instruments PCR amplifications were performed with GeneAmp® PCR System 9700 (Applied Biosystems®, Finland) cyclers carrying dual 384-well blocks. The robotic platform utilized one Tecan Freedom EVO 100 and three EVO 150 pipetting robots (PLD, Finland). NIR-detection was performed with LI-COR Aerius Automated Imaging System (LI-COR, Thermo Scientific, Finland).
Software and databases The Autoreport IT solution for automated generation of genotyping results and reports to the patients was programmed using Visual Basic for Applications (VBA) scripts and multiple databases. As a whole they formed a laboratory automation system (LAS). The cyclic minisequencing (cMS) assay work flow initiated by technician producing a work list in the hospital laboratory information management system. Based on tests selected into this work list the program created sample tables and pipetting maps, and calculated reagent mixes. Every test run could contain 1–8 different tests consisting of 1–24 mutations or variations in one or multiple genes ad hoc. All assay parameters were stored in a database, and no programming skills were required from the technicians. Three files were stored. The first contained the original patient data and was used for final reporting. Assay parameters were stored as CSV file on USB memory stick which was carried along the samples and assay plates between multiple Tecan robots during the procedure. The reagent mixture pipetting commands were stored on the same USB stick in the third file and used for robotic reagent mixing. The parameter files were read by Tecan Freedom Evoware software, and Tecan VBScripts were called for handling these parameters. No interventions were required by the technician.
Scand J Clin Lab Invest Downloaded from informahealthcare.com by National Taiwan University on 01/12/15 For personal use only.
46
K. Alagrund & A. K. Orpana
At the end of the assay run the fluorescence results from the IR fluorometer were fetched into the original file containing the patient data, and data analysis and genotype calling was performed for each mutation using automated VBA scripts. The control value database selected results from control samples of 20 previous test runs and calculated cut-off values for automated genotype calling. These cut-off values could be changed by the person performing genotyping. Called genotypes were coded and after analyzing all variations of the test, the result code strings (RCS) were formed for each patient. In addition to genotype results also information about the age and gender from the patient’s test request were used for creation of age and gender group codes. For creation of patient reports this RCS was used for lookup from report component database. Six report components, the test specific background chapter, the obtained genotype result, the test specific interpretation, the genotype specific interpretation, the genotype specific closing sentence and finally, the test specific closing sentence, were combined. If age group or gender specific components existed, they were used as they were coded into the RCS and in that case also the database contained specific components for them. Combined patient reports were sent to the work list, visually inspected and accepted individually before transferring to the hospital patient records.
bisulfite conversion that converted unmethylated cytosine residues to uracils but leaved methylated residues unchanged. EpiTect Bisulfite Kit (Qiagen, Finland) was used in conversion.
PCR When solid phase minisequencing (MS) tests were converted to cMS tests their PCR conditions were harmonized. If needed, the PCR primers were redesigned to fulfill strict amplification condition requirements. Practically all tests could be performed using a single PCR program, which was a prerequisite for random access testing principle and possibility of free mixing of tests on 384-well PCR plate. All samples and positive controls were loaded to cooled IC22 Two Position Chilling Dry Bath blocks (Torrey Pines Scientific, Inc, CA, USA) on the deck of a Tecan robotic workstation. Pipetting protocol was programmed to minimize the use of the valuable positive control material. In the high-throughput single mutation assays samples and controls were pre-ordered on a 96-well micro plate during automated preanalytical processing. In both single and multi-mutation PCR systems Tecan Freedom EVO 150 robot pipetted reagents, controls and samples for PCR reactions without any intervention by the technician, based on the parameter file on an USB stick.
Preanalytics The high-throughput cMS tests were performed using DNA released from EDTA blood dried on paper disks. It was inexpensive, fast and suitable if the number of individual PCR reactions was low, such as in the single polymorphism (-13190 C⬎T, rs 4988234) test for lactose malabsorption [19]. Because of a relatively large number of samples, this preanalytical process was automated. The system was able to process 192 samples (2 ⫻ 96 sample plates) in a single run. Tube order was verified by bar code reader before pipetting. The Tecan Freedom EVO 100 robot transferred an aliquot of blood from the original EDTA blood tubes to the blotting paper spot on the bottom of a 96-well micro plate. The plates were allowed to dry. This was an automated version of the original ‘Stixing’ method [20]. For tests requiring higher amounts of DNA, extraction was partly automated with Maxwell® 16 Instrument (Promega, Biofellows, Finland) following manufacturer’s instructions. DNA was quantified spectrophotometrically using NanoDrop ND-1000 Spectrophotometer (Thermo Scientific, Finland)
Bisulfite conversion in epigenetic tests Methylation of the promoter of the MGMT gene cytosines of the CpG islands was detectable after
cMS PCR product was purified by incubation with exonuclease I (ExoI) and shrimp alkaline phosphatase (SAP) enzymes, which were automatically pipetted on a micro plate by Tecan Freedom EVO 150 robot. Reagent preparation for cMS reaction was partly automated. Detection primers required in current analysis were loaded to a reagent block and labeled nucleotides set in fixed order in another block. The Tecan Freedom EVO 150 robot pipetted primers and nucleotides to reagent tubes and just prior to use the master mixture containing aqua, PCR buffer and DNA polymerase was pipetted manually to these reagent tubes. The wild type and mutant cyclic single nucleotide primer extension reactions of cMS were performed in duplicates. Biotinylated detection primers and Fluoresceine- and Cy5-labeled nucleotides were used for wild type and mutant reactions, respectively. Separate reactions allowed us to use inexpensive dNTP instead of ddNTP and conventional DNA polymerase. During the reaction the DNA polymerase incorporated only the right nucleotide matching to the template sequence of the PCR product. As described above, one PCR reaction, containing one mutation site was divided into four reaction wells during cMS, as the wild type reactions and the
Scand J Clin Lab Invest Downloaded from informahealthcare.com by National Taiwan University on 01/12/15 For personal use only.
Automation of diagnostic genetic testing mutant reactions were prepared separately and in duplicates. In the detection stage one wild type and one mutant reaction were combined together, ending up with two detection wells for each sample. The wild type and the mutant reaction products were combined in a streptavidin coated detection plate and incubated at 37°C, during which the biotin-labeled detection primers were trapped to the bottom of the well. Automated procedure was continued using primary and NIR-labeled secondary antibodies. Although being fluorophores themselves, Fluorescein- and Cy5-labeled nucleotides were not found sensitive enough in this type of an assay and were used here as antigens, because no NIR-labeled dNTPs were yet available. Fluorophore-binding primary antibodies and NIR-labeled secondary antibodies were used for signal amplification. After the final washing step the plate was automatically transferred by Tecan Freedom EVO 150 robot to the LI-COR Aerius Automated Imaging System where the signal from the wild type reaction was detected at the 700 nm channel and the mutant reaction signal was detected at the 800 nm channel.
47
Relative fluorescence units (RFU) were stored in a file for genotyping. The initial file created for sample pipetting was re-opened and results were fetched into this file via laboratory intranet. The results of the control samples were saved in a control value database. After all the results in the test run were analyzed and approved, the program automatically created complete patient reports based on specific codes created by information in the electronic request and results of the genotyping assays. These statements were exported to customers electrically or in paper. All information was archived as files. The cMS process is presented in Figure 1.
Results Validation We developed and validated a robust analytical platform based on an automation friendly modification of the traditional minisequencing method. This automated genotyping system included sample preprocessing, PCR-amplification, cMS with NIR-detection, and automated data processing and result reporting
Figure 1. A simplified diagram of the stages in the cMS process: data processing, preprocessing, PCR, purification, cMS and detection. Tecan 1: Tecan Freedom EVO 100 used in preprocessing, Tecan 2: Tecan Freedom EVO 150 used in PCR-pipetting, Tecan 3 and Tecan 4: Tecan Freedom EVO 150 robots used in cMS steps.
48
K. Alagrund & A. K. Orpana
Scand J Clin Lab Invest Downloaded from informahealthcare.com by National Taiwan University on 01/12/15 For personal use only.
system. The system was gradually created and validated during several years and enables us now to add new tests and mutations to the test panels after a short specific validation period. All clinical validations were made according to the diagnostic laboratory protocol based on collaboration with the clinicians. The tests were validated for multiple types of DNA samples: Paraffin embedded tissue blocks, fresh tissue, EDTA blood, dried on paper spots, and cell lysates depending on expected sample type. Tissue sample types tested and found suitable for cMS originated from liver, muscle and placenta. Also
samples of amniotic fluid cells and buccal swabs were successfully analyzed. Genotyping The genotyping was based on the ratio (R value) between wild type and mutant well signals, calculated by the LAS automatically. The R values optimally were ⬍ 0.1 in wild-type samples, around 1 in heterozygous samples and ⬎ 10 in homozygous samples. However, there were great variations in R values between mutation sites due to different amounts of
Table I. Diseases and epigenetic modifications included in the cMS test panels. Symbol
Name/Phenotype
One-mutation panel LAKT Lactose malabsorption Two-mutation panel FV Thrombotic risk evaluation FII Thrombotic risk evaluation Sub acute test panel HFE Hemochromatosis TPMT Thiopurine S-methyltransferase Varfa Warfarin anticoagulant response SLCO1B1 Statin induced myopathy Random access principle multi-mutation test panel FDH and LQTs AGU Aspartylglucosaminuria APECED Autoimmunepolyendocrine syndrome type 1 CCD Cleidocranialdysplasi CLD Chloride diarrhea, congenital, Finnish type CLN5 Ceroidlipofuscinosis, neuronal, 5 CNF Nephrotic syndrome, type 1 Cohen Cohen syndrome DTD Diastrophic dysplasia Gelso Amyloidosis, Finnish type GRACILE GRACILE syndrome HLS Hydrolethalus syndrome INCL Ceroidlipofuscinosis, neuronal, 1 IOSCA Infantile-onset spinocerebellar ataxia JNCL Ceroidlipofuscinosis, neuronal, 3 MEB Muscle-eye-braindisease MKS Meckel syndrome 1, Meckel syndrome 6 MUL Mulibreynanism SD Salla disease TMD Tibial musculardystrophy, tardive USH3 Usher syndrome, type 3A XLRS Retinoschisis 1, X-linked, Juvenile LQT1 Long QT syndrome 1 LQT2 Long QT syndrome 2 BRCA BRCA1 Breast cancer type 1 BRCA2 Breast cancer type 2 CYP2D6 CYP2D6 Cytochrome P450, Subfamily IID, Polypeptide 6 KRAS-BRAF panel KRAS V-Ki-ras2 Kirsten rat sarcoma viral oncogene homolog BRAF V-raf murine sarcoma viral oncogene homolog B1 Epigenetic panel MGMT O(6)-methylguanine-DNA-methyltransferrase
Gene
Number of mutation sites
LCT
1
F5 F2
1 1
HFE TPMT CYP2C9 VKORC1 SLCO1B1
2 5 2 2 1
AGA AIRE RUNX2 SLC26A3 CLN5 NPHS1 COH1 SLC26A2 GSN BCS1L HYLS1 PPT1 C10ORF2 CLN3 POMGNT1 MKS1, CC2D2A TRIM37 SLC17A5 TTN CLRN1 RS1 KCNQ1 KCNH2
2 3 1 1 1 2 1 1 1 1 1 1 1 2 1 1 1 1 1 1 1 3 2 2
BRCA1 BRCA2
9 8
CYP2D6
8
KRAS
7
BRAF
2
MGMT
5
Scand J Clin Lab Invest Downloaded from informahealthcare.com by National Taiwan University on 01/12/15 For personal use only.
Automation of diagnostic genetic testing nucleotides incorporated and the local sequence environment. Typically a ten-fold R value difference between genotypes was obtained. The autovalidation part of LAS calculated R value ranges based on the results of 20 previous control samples. Diagnosis of inherited mutations and polymorphisms was based on the R value, but when a more quantitative result was needed, as e.g. in epigenetic modification or companion testing (MGMT, KRAS, BRAF) the results were calculated as a relative amount of the mutant signal. This Percentage value was more sensitive when testing acquired mutations samples with low amounts of tumor tissue. Assays for seven mutations in KRAS gene and two in BRAF gene were validated for clinical use. Methylation of the promoter of the MGMT gene cytosines of the CpG islands was detectable after bisulfite conversion. Commonly used methods, such as methyl-specific PCR or pyrosequencing also lean on bisulfate treated DNA [21]. We validated five reported cytosine methylation sites and one cytosine control site in the promoter region of the MGMT gene [22]. The tests analyzing mutation or epigenetic status of a tumor tissue needed to reach high sensitivity. The amount of affected DNA varies between almost none to nearly 100% in tissue samples taken from the tumor area. Using cMS we could detect down to 5% of affected DNA in tumor samples in most mutation sites. Autoreport In result reporting an important factor was presenting the test results to the clinician in an unambiguous, understandable form. Using a strategy in which patient information from the request and the multiple genotyping results formed a result code string, which was then used for creation of the final report allowed us to create very specific reports. The genotyping information was not always sufficient, as clinicians were used to alleles, not genotypes. In the pharmacogenetic tests which explored the inherited forms of metabolic pathways, the metabolism of a therapeutic drug was classified based on alleles which were results of various mutation combinations [23]. The deficiency of Thiopurine-S-Methyltransferase (TPMT) predisposes to severe side effects of thiopurine [24,25]. Several allele combinations cause severe reactions with therapeutic dose of drug. In clinical use 18 allele combinations (alleles *1,*2, *3A, 3B, *3C and *16) are significant and frequently found in Finnish population. The Autoreport system could create automatic allele specific patient reports. This service of the clinicians was taken one step further in VarfaD testing. Warfarin is a K-vitamin antagonist used to prevent excessive blood clotting [26]. The VarfaD test based on two VKORC1- and two CYP2C9- polymorphisms was used to deter-
49
Figure 2. (A) Sample volumes in MS process and cMS process during 2008–2011. Number of the tests in the cMS process on the left Y-axis and number of the tests in the MS process on the rigth Y-axis. (B) Conversion from MS to cMS during 2008–2011 did not affect the cMS process mean TATs.
mine the Warfarin metabolic status and initial Warfarin dose of the patient [23]. The medical statements based on CYP2C9*1-, CYP2C9*2-, CYP2C9*3- and, two VKORC1- alleles were created similarly to TPMT above. Based on the age, weight and the genotyping result of the patient the program calculated an initial Warfarin dosing recommendation helping physicians to estimate the Warfarin dosage. These tests are two examples of the sophisticated automatic reporting system of this assay platform. Use of automatic report creation significantly saved working time in routine as specialist’s intervention was not needed in great majority of cases. Current status To date we have successfully validated over 40 diagnostic DNA tests, altogether over 100 mutations into this platform. Tests currently included in the cMS process are listed in Table I. More than 30,000 autoreports have been created by the system.The throughput could be improved further, as up to four different cMS runs could be combined into a single detection run. In this format 768 patient samples could be analyzed in duplicates in 4 h with very little manual intervention. First results would be available in
50
K. Alagrund & A. K. Orpana
1.5 h in case of urgent samples, but typically series would be automatically processed overnight and analyzed next morning. The genetic tests for inherited diseases based on cMS were accredited by FINAS (Finnish Accreditation Service) according to ISO EN 15189 standard.
sion of manual MS tests to automated cMS format enabled maintenance of stabile TATs despite of significantly larger test selection and rising sample amounts (Figure 2A and B).
Precision Effects on turnaround times (TATs)
Comparing 90 consecutive series of the lactose malabsorption test during one year the R-values of the genotypes remained stable despite of the minor variation between series (Figure 3A). The variation resulted
Scand J Clin Lab Invest Downloaded from informahealthcare.com by National Taiwan University on 01/12/15 For personal use only.
During the four-year period of automation, method development and conversion the growth in analyzed sample amount in our laboratory was 23%. Conver-
Figure 3. (A) Mean R-value of genotypes ⫾ 1 SD of 90 consecutive lactase series; wild type (N/N), heterozygote (M/N), homozygote (M/M). (B) Mean R-value of genotypes normalized as for M/N genotype, 90 lactase series; wild type (N/N), heterozygote (M/N), homozygote (M/M).
Automation of diagnostic genetic testing mostly from changing reagent lots, and assay conditions. This variation usually affected the signal levels of each genotype, shifting all values simultaneously. If this variation was abolished by normalization, remaining technical variation was very small (Figure 3B).
[4]
[5]
[6]
Scand J Clin Lab Invest Downloaded from informahealthcare.com by National Taiwan University on 01/12/15 For personal use only.
Discussion The introduction of nearly total laboratory automation in our diagnostic genetic testing enabled analysis of large amounts of samples without addition of laboratory personnel and improved turnaround times. This was based on two advantages: The ability to perform high-throughput series with large sample volumes and, on the other hand, combine up to multiple infrequent mutations randomly in one test run. Automated cyclic minisequencing combined with antibody-based near-infrared detection was inexpensive and highly sensitive. Ability to use reagents from multiple vendors aided in avoiding single vendor trap and provided security to the process. The IT developments based on electronic data transfer and multiple databases, including automatically updated control value database creating genotype cut offs, automated and simplified processes significantly. The current tendency is to avoid manual data analysis and reporting, and thereby minimize human errors and labor also in genetic laboratories. Automatically generated patient reports dramatically dropped TATs as no manual typing was required. The purpose of this work was to develop a robust and economical automated robotic based method to decrease laborious manual laboratory work and reduce reagent costs. The cMS system could be used as proof-of-principle process for developing a complete walk-a-way analyzer and bring total laboratory automation also into diagnostic mutation testing.
[7] [8]
[9] [10]
[11]
[12] [13]
[14]
[15]
[16]
Acknowledgements We thank the staff at the HUSLAB Laboratory of Genetics for technical assistance. Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.
[17]
[18]
[19]
References [1] Peltonen L. Molecular background of the Finnish disease heritage. Ann Med 1997;29:553–6. [2] Valkonen S, Hietala M, Savontaus ML, Aula P. Origin of Finnish mutations causing aspartylglucosaminuria. Hereditas 1999;131:191–5. [3] Saarela J, Laine M, Oinonen C, von Schantz C, Jalanko A, Rouvinen J, Peltonen, L. Molecular pathogenesis of a
[20]
[21]
51
disease: structural consequences of aspartylglucosaminuria mutations. Hum Mol Genet 2001;10:983–95. Hietala M, Gron K, Syvanen AC, Peltonen L, Aula P. Prospects of carrier screening of aspartylglucosaminuria in Finland. Eur J Hum Genet 1993;1:296–300. Peltonen L, Jalanko A, Varilo T. Molecular genetics of the Finnish disease heritage. Hum Mol Genet 1999;8: 1913–23. Peltonen L, Pekkarinen P, Aaltonen J. Messages from an isolate: lessons from the Finnish gene pool. Biol Chem Hoppe Seyler 1995;376:697–704. Sipila K, Aula P. Database for the mutations of the Finnish disease heritage. Hum Mutat 2002;19:16–22. Cohen J, Wilson A, Manzolillo K. Clinical and economic challenges facing pharmacogenomics. Pharmacogenomics J 2013;4:378–88. Snyderman R. Personalized health care: From theory to practice. Biotechnol J 2012;7:973–9. Ginsburg GS, Willard HF. Genomic and personalized medicine: foundations and applications. Transl Res 2009; 154:277–87. SEARCH Collaborative Group,Link E, Parish S, Armitage J, Bowman L, Heath S, Matsuda F, Gut I, Lathrop M, Collins R. SLCO1B1 variants and statininduced myopathy – a genomewide study. N Engl J Med 2008;359:789–99. Kranenburg O. The KRAS oncogene: past, present, and future. Biochim Biophys Acta 2005;1756:81–2. Haigis KM, Kendall KR, Wang Y, Cheung A, Haigis MC, Glickman JN, Niwa-Kawakita M, Sweet-Cordero A, Sebolt-Leopold J, Shannon KM, Settleman J, Giovannini M, Jacks T. Differential effects of oncogenic K-Ras and N-Ras on proliferation, differentiation and tumor progression in the colon. Nat Genet 2008;40:600–8. Amanuel B, Grieu F, Kular J, Millward M, Iacopetta B. Incidence of BRAF p.Val600Glu and p.Val600Lys mutations in a consecutive series of 183 metastatic melanoma patients from a high incidence region. Pathology 2012; 44:357–9. Davies H, Bignell GR, Cox C, Stephens P, Edkins S, Clegg S, Teague J, Woffendin H, Garnett MJ, Bottomley W, Davis N, Dicks E, Ewing R, Floyd Y, Gray K, Hall S, Hawes R, Hughes J, Kosmidou V, Menzies A, Mould C, Parker A, Stevens C, Watt S, Hooper S, Wilson R, Jayatilake H, Gusterson BA, Cooper C, Shipley J, Hargrave D, Pritchard-Jones K, Maitland N, ChenevixTrench G, Riggins GJ, Bigner DD. Mutations of the BRAF gene in human cancer. Nature 2002; 417:949–54. Esteller M, Garcia-Foncillas J, Andion E, Goodman SN, Hidalgo OF, Vanaclocha V, Baylin SB, Herman JB. Inactivation of the DNA-repair gene MGMT and the clinical response of gliomas to alkylating agents. N Engl J Med 2000;343:1350–4. Jalanko A, Kere J, Savilahti E, Schwartz M, Syvanen AC, Ranki M, Soderlund H. Screening for defined cystic fibrosis mutations by solid-phase minisequencing. Clin Chem 1992;38:39–43. Orpana AK. Fluorescence resonance energy transfer (FRET) using ssDNA binding fluorescent dye. Biomol Eng 2004; 21:45–50. Enattah NS, Sahi T, Savilahti E, Terwilliger JD, Peltonen L, Jarvela I. Identification of a variant associated with adulttype hypolactasia. Nat Genet 2002;30:233–7. Ihalainen J, Siitari H, Laine S, Syvanen AC, Palotie A. Towards automatic detection of point mutations: use of scintillating microplates in solid-phase minisequencing. BioTechniques 1994;16:938–94. Karayan-Tapon L, Quillien V, Guilhot J, Wager M, Fromont G, Saikali S, Etcheverry A, Hamlat A, Loussouarn D, Campion L, Campone M, Vallette FM, Gratas-Rabbia-Re C. Prognostic value of O6-methylguanine-DNA
52
K. Alagrund & A. K. Orpana
Scand J Clin Lab Invest Downloaded from informahealthcare.com by National Taiwan University on 01/12/15 For personal use only.
methyltransferase status in glioblastoma patients, assessed by five different methods. J Neurooncol 2010;97:311–22. [22] Nakagawachi T, Soejima H, Urano T, Zhao W, Higashimoto K, Satoh Y, Matsukura S, Kudo S, Kitajima Y, Harada H, Furukawa K, Matsuzaki H, Emi M, Nakabeppu Y, Miyazaki K, Sekiguchi M, Mukai T. Silencing effect of CpG island hypermethylation and histone modifications on O6-methylguanine-DNA methyltransferase (MGMT) gene expression in human cancer. Oncogene 2003;22:8835–44. [23] Takahashi H, Wilkinson GR, Nutescu EA, Morita T, Ritchie MD, Scordo MG, Pengo V, Barban M, Padrini R, Ieiri I, Otsubo K, Kashima T, Kimura S, Kijima S, Echizen H. Different contributions of polymorphisms in VKORC1 and CYP2C9 to intra- and inter-population differences in
maintenance dose of warfarin in Japanese, Caucasians and African-Americans. Pharmacogenet Genomics 2006;16: 101–10. [24] Coulthard S, Hogarth L. The thiopurines: an update. Invest New Drugs 2005;23:523–32. [25] Haglund S, Lindqvist M, Almer S, Peterson C, Taipalensuu J. Pyrosequencing of TPMT alleles in a general Swedish population and in patients with inflammatory bowel disease. Clin Chem 2004;50:288–95. [26] Ross KA, Bigham AW, Edwards M, Gozdzik A, Suarez-Kurtz G, Parra EJ. Worldwide allele frequency distribution of four polymorphisms associated with warfarin dose requirements. J Hum Genet 2010;55: 582–9.
