Performance Characteristics and Validation of Next-Generation Sequencing for Human Leucocyte Antigen Typing
2016; Elsevier BV; Volume: 18; Issue: 5 Linguagem: Inglês
10.1016/j.jmoldx.2016.03.009
ISSN1943-7811
AutoresEric T. Weimer, Maureen C. Montgomery, Rosanne Petraroia, John Crawford, John L. Schmitz,
Tópico(s)T-cell and B-cell Immunology
ResumoHigh-resolution human leukocyte antigen (HLA) matching reduces graft-versus-host disease and improves overall patient survival after hematopoietic stem cell transplant. Sanger sequencing has been the gold standard for HLA typing since 1996. However, given the increasing number of new HLA alleles identified and the complexity of the HLA genes, clinical HLA typing by Sanger sequencing requires several rounds of additional testing to provide allele-level resolution. Although next-generation sequencing (NGS) is routinely used in molecular genetics, few clinical HLA laboratories use the technology. The performance characteristics of NGS HLA typing using TruSight HLA were determined using Sanger sequencing as the reference method. In total, 211 samples were analyzed with an overall accuracy of 99.8% (2954/2961) and 46 samples were analyzed for precision with 100% (368/368) reproducibility. Most discordant alleles were because of technical error rather than assay performance. More important, the ambiguity rate was 3.5% (103/2961). Seventy-four percentage of the ambiguities were within the DRB1 and DRB4 loci. HLA typing by NGS saves approximately $6000 per run when compared to Sanger sequencing. Thus, TruSight HLA assay enables high-throughput HLA typing with an accuracy, precision, ambiguity rate, and cost savings that should facilitate adoption of NGS technology in clinical HLA laboratories. High-resolution human leukocyte antigen (HLA) matching reduces graft-versus-host disease and improves overall patient survival after hematopoietic stem cell transplant. Sanger sequencing has been the gold standard for HLA typing since 1996. However, given the increasing number of new HLA alleles identified and the complexity of the HLA genes, clinical HLA typing by Sanger sequencing requires several rounds of additional testing to provide allele-level resolution. Although next-generation sequencing (NGS) is routinely used in molecular genetics, few clinical HLA laboratories use the technology. The performance characteristics of NGS HLA typing using TruSight HLA were determined using Sanger sequencing as the reference method. In total, 211 samples were analyzed with an overall accuracy of 99.8% (2954/2961) and 46 samples were analyzed for precision with 100% (368/368) reproducibility. Most discordant alleles were because of technical error rather than assay performance. More important, the ambiguity rate was 3.5% (103/2961). Seventy-four percentage of the ambiguities were within the DRB1 and DRB4 loci. HLA typing by NGS saves approximately $6000 per run when compared to Sanger sequencing. Thus, TruSight HLA assay enables high-throughput HLA typing with an accuracy, precision, ambiguity rate, and cost savings that should facilitate adoption of NGS technology in clinical HLA laboratories. It is widely accepted that human leukocyte antigen (HLA) matching reduces patient morbidity and mortality after hematopoietic stem cell transplantation (HSCT).1Lee S.J. Klein J. Haagenson M. Baxter-Lowe L.A. Confer D.L. Eapen M. Fernandez-Vina M. Flomenberg N. Horowitz M. Hurley C.K. Noreen H. Oudshoorn M. Petersdorf E. Setterholm M. Spellman S. Weisdorf D. Williams T.M. Anasetti C. High-resolution donor-recipient HLA matching contributes to the success of unrelated donor marrow transplantation.Blood. 2007; 110: 4576-4583Crossref PubMed Scopus (986) Google Scholar The current standard of care is high-resolution HLA typing by Sanger sequencing or sequence-based typing (SBT). High-resolution HLA typing is defined as a set of alleles that specifies and encodes the same protein sequence for the peptide-binding region of an HLA molecule. However, SBT cannot accurately phase heterozygous alleles and provides limited sequencing information. Traditionally, HLA typing by SBT typically involves sequencing only exons 2, 3, and 4 of HLA class I genes and exons 2 and 3 of HLA class II genes. Since the regulatory requirement was established for high-resolution HLA typing for HLA-A/B/C/DRB1 in 2005, many clinical laboratories are putting significant resources toward ambiguity resolution.2National Marrow Donor Program: NMDP Policy for Confirmatory Typing. Minneapolis, MN: National Marrow Donor Program, 2005.Google Scholar Even as several restrictions were removed, such as the requirement to exclude rare alleles, there is still a growing list of ambiguities that require additional testing and delay patient results. The issue of ambiguities is a testament to the complexity of the HLA region in the human genome with >13,000 alleles identified to date.3Robinson J. Halliwell J.A. Hayhurst J.D. Flicek P. Parham P. Marsh S.G. The IPD and IMGT/HLA database: allele variant databases.Nucleic Acids Res. 2015; 43: D423-D431Crossref PubMed Scopus (1449) Google Scholar The combination of the inability to phase heterozygous alleles and the growing number of HLA alleles has led to a significant number of ambiguities in HLA typing that require time-consuming and costly additional tests to be performed. In 2004, Adams et al4Adams S.D. Barracchini K.C. Chen D. Robbins F. Wang L. Larsen P. Luhm R. Stroncek D.F. Ambiguous allele combinations in HLA Class I and Class II sequence-based typing: when precise nucleotide sequencing leads to imprecise allele identification.J Transl Med. 2004; 2: 30Crossref PubMed Scopus (59) Google Scholar reported the ambiguity rate for HLA-A, HLA-B, and HLA-C of 24% to 41%. Three years later, Voorter et al5Voorter C.E. Mulkers E. Liebelt P. Sleyster E. van den Berg-Loonen E.M. Reanalysis of sequence-based HLA-A, -B and -Cw typings: how ambiguous is today's SBT typing tomorrow.Tissue Antigens. 2007; 70: 383-389Crossref PubMed Scopus (26) Google Scholar reported that ambiguities for HLA-A, HLA-B, and HLA-C had increased to approximately 50%. Currently, our experience is that ambiguity resolution is required in 53% of patient specimens to determine a specific allele (E.T. Weimer and J.L. Schmitz, unpublished observation). This high level of additional testing delays patient HLA typing results and increases the cost of HLA typing. With no sign that the number of HLA alleles identified will decline and the increasing application of typing of additional loci (DPB1 and DRB3/4/5), there is a great need for a technology that allows for accurate high-resolution HLA typing without the requirement of additional testing.3Robinson J. Halliwell J.A. Hayhurst J.D. Flicek P. Parham P. Marsh S.G. The IPD and IMGT/HLA database: allele variant databases.Nucleic Acids Res. 2015; 43: D423-D431Crossref PubMed Scopus (1449) Google Scholar, 6Crivello P. Zito L. Sizzano F. Zino E. Maiers M. Mulder A. Toffalori C. Naldini L. Ciceri F. Vago L. Fleischhauer K. The impact of amino acid variability on alloreactivity defines a functional distance predictive of permissive HLA-DPB1 mismatches in hematopoietic stem cell transplantation.Biol Blood Marrow Transplant. 2015; 21: 233-241Abstract Full Text Full Text PDF PubMed Scopus (86) Google Scholar, 7Fleischhauer K. Shaw B.E. Gooley T. Malkki M. Bardy P. Bignon J.-D. Dubois V. Horowitz M.M. Madrigal J.A. Morishima Y. Oudshoorn M. Ringden O. Spellman S. Velardi A. Zino E. Petersdorf E.W. Effect of T-cell-epitope matching at HLA-DPB1 in recipients of unrelated-donor haemopoietic-cell transplantation: a retrospective study.Lancet Oncol. 2012; 13: 366-374Abstract Full Text Full Text PDF PubMed Scopus (250) Google Scholar, 8Fernandez-Vina M.A. Klein J.P. Haagenson M. Spellman S.R. Anasetti C. Noreen H. Baxter-Lowe L.A. Cano P. Flomenberg N. Confer D.L. Horowitz M.M. Oudshoorn M. Petersdorf E.W. Setterholm M. Champlin R. Lee S.J. de Lima M. Multiple mismatches at the low expression HLA loci DP, DQ, and DRB3/4/5 associate with adverse outcomes in hematopoietic stem cell transplantation.Blood. 2013; 121: 4603-4610Crossref PubMed Scopus (111) Google Scholar Next-generation sequencing (NGS) technology is the massive parallel sequencing of clonal DNA molecules. By using unique molecular signatures or barcodes, many samples can be pooled together and sequenced simultaneously. A key feature of the NGS method is the ability to generate massive amounts of genetic data from many DNA molecules simultaneously.9Metzker M.L. Sequencing technologies: the next generation.Nat Rev Genet. 2010; 11: 31-46Crossref PubMed Scopus (5015) Google Scholar, 10Gabriel C. Furst D. Fae I. Wenda S. Zollikofer C. Mytilineos J. Fischer G.F. HLA typing by next-generation sequencing: getting closer to reality.Tissue Antigens. 2014; 83: 65-75Crossref PubMed Scopus (68) Google Scholar The combination of clonal DNA sequencing and application of long-range PCR techniques to increase HLA genomic information aids in the reduction in HLA allele ambiguities. For example, Danzer et al11Danzer M. Niklas N. Stabentheiner S. Hofer K. Proll J. Stuckler C. Raml E. Polin H. Gabriel C. Rapid, scalable and highly automated HLA genotyping using next-generation sequencing: a transition from research to diagnostics.BMC Genomics. 2013; 14: 221Crossref PubMed Scopus (55) Google Scholar used long-range PCR of HLA genes and NGS to demonstrate an average ambiguity reduction of 93.5% for HLA-A, HLA-B, HLA-C, DRB1, DQB1, and DPB1. Although the ambiguity reduction varied by locus [ie, the reduction of DRB1 was less pronounced (46.1%)], this was still considered significant given the amount of additional testing required by SBT. HLA typing by NGS using long-range PCR techniques to increase genomic coverage of HLA genes has proven effective at improving ambiguity resolution.11Danzer M. Niklas N. Stabentheiner S. Hofer K. Proll J. Stuckler C. Raml E. Polin H. Gabriel C. Rapid, scalable and highly automated HLA genotyping using next-generation sequencing: a transition from research to diagnostics.BMC Genomics. 2013; 14: 221Crossref PubMed Scopus (55) Google Scholar, 12Wang C. Krishnakumar S. Wilhelmy J. Babrzadeh F. Stepanyan L. Su L.F. Levinson D. Fernandez-Vina M.A. Davis R.W. Davis M.M. Mindrinos M. High-throughput, high-fidelity HLA genotyping with deep sequencing.Proc Natl Acad Sci U S A. 2012; 109: 8676-8681Crossref PubMed Scopus (142) Google Scholar, 13Shiina T. Suzuki S. Ozaki Y. Taira H. Kikkawa E. Shigenari A. Oka A. Umemura T. Joshita S. Takahashi O. Hayashi Y. Paumen M. Katsuyama Y. Mitsunaga S. Ota M. Kulski J.K. Inoko H. Super high resolution for single molecule-sequence-based typing of classical HLA loci at the 8-digit level using next generation sequencers.Tissue Antigens. 2012; 80: 305-316Crossref PubMed Scopus (114) Google Scholar, 14Lind C. Ferriola D. Mackiewicz K. Heron S. Rogers M. Slavich L. Walker R. Hsiao T. McLaughlin L. D'Arcy M. Gai X. Goodridge D. Sayer D. Monos D. Next-generation sequencing: the solution for high-resolution, unambiguous human leukocyte antigen typing.Hum Immunol. 2010; 71: 1033-1042Crossref PubMed Scopus (118) Google Scholar The increased genomic coverage is partly what enables the accurate phasing of heterozygous bp positions often observed in SBT.13Shiina T. Suzuki S. Ozaki Y. Taira H. Kikkawa E. Shigenari A. Oka A. Umemura T. Joshita S. Takahashi O. Hayashi Y. Paumen M. Katsuyama Y. Mitsunaga S. Ota M. Kulski J.K. Inoko H. Super high resolution for single molecule-sequence-based typing of classical HLA loci at the 8-digit level using next generation sequencers.Tissue Antigens. 2012; 80: 305-316Crossref PubMed Scopus (114) Google Scholar, 15Hosomichi K. Jinam T.A. Mitsunaga S. Nakaoka H. Inoue I. Phase-defined complete sequencing of the HLA genes by next-generation sequencing.BMC Genomics. 2013; 14: 355Crossref PubMed Scopus (88) Google Scholar Several recent reports have demonstrated the feasibility of using NGS technology to provide >97% concordance with SBT.11Danzer M. Niklas N. Stabentheiner S. Hofer K. Proll J. Stuckler C. Raml E. Polin H. Gabriel C. Rapid, scalable and highly automated HLA genotyping using next-generation sequencing: a transition from research to diagnostics.BMC Genomics. 2013; 14: 221Crossref PubMed Scopus (55) Google Scholar, 13Shiina T. Suzuki S. Ozaki Y. Taira H. Kikkawa E. Shigenari A. Oka A. Umemura T. Joshita S. Takahashi O. Hayashi Y. Paumen M. Katsuyama Y. Mitsunaga S. Ota M. Kulski J.K. Inoko H. Super high resolution for single molecule-sequence-based typing of classical HLA loci at the 8-digit level using next generation sequencers.Tissue Antigens. 2012; 80: 305-316Crossref PubMed Scopus (114) Google Scholar, 15Hosomichi K. Jinam T.A. Mitsunaga S. Nakaoka H. Inoue I. Phase-defined complete sequencing of the HLA genes by next-generation sequencing.BMC Genomics. 2013; 14: 355Crossref PubMed Scopus (88) Google Scholar, 16Smith A.G. Pyo C.W. Nelson W. Gow E. Wang R. Shen S. Sprague M. Pereira S.E. Geraghty D.E. Hansen J.A. Next generation sequencing to determine HLA class II genotypes in a cohort of hematopoietic cell transplant patients and donors.Hum Immunol. 2014; 75: 1040-1046Crossref PubMed Scopus (29) Google Scholar, 17Holcomb C.L. Hoglund B. Anderson M.W. Blake L.A. Bohme I. Egholm M. Ferriola D. Gabriel C. Gelber S.E. Goodridge D. Hawbecker S. Klein R. Ladner M. Lind C. Monos D. Pando M.J. Proll J. Sayer D.C. Schmitz-Agheguian G. Simen B.B. Thiele B. Trachtenberg E.A. Tyan D.B. Wassmuth R. White S. Erlich H.A. A multi-site study using high-resolution HLA genotyping by next generation sequencing.Tissue Antigens. 2011; 77: 206-217Crossref PubMed Scopus (99) Google Scholar Although the high-throughput nature of NGS is thought to be cost-effective for HLA typing, this has yet to be shown. More important, previous reports focused on laboratory-developed assays rather than commercially available reagents.11Danzer M. Niklas N. Stabentheiner S. Hofer K. Proll J. Stuckler C. Raml E. Polin H. Gabriel C. Rapid, scalable and highly automated HLA genotyping using next-generation sequencing: a transition from research to diagnostics.BMC Genomics. 2013; 14: 221Crossref PubMed Scopus (55) Google Scholar, 12Wang C. Krishnakumar S. Wilhelmy J. Babrzadeh F. Stepanyan L. Su L.F. Levinson D. Fernandez-Vina M.A. Davis R.W. Davis M.M. Mindrinos M. High-throughput, high-fidelity HLA genotyping with deep sequencing.Proc Natl Acad Sci U S A. 2012; 109: 8676-8681Crossref PubMed Scopus (142) Google Scholar, 13Shiina T. Suzuki S. Ozaki Y. Taira H. Kikkawa E. Shigenari A. Oka A. Umemura T. Joshita S. Takahashi O. Hayashi Y. Paumen M. Katsuyama Y. Mitsunaga S. Ota M. Kulski J.K. Inoko H. Super high resolution for single molecule-sequence-based typing of classical HLA loci at the 8-digit level using next generation sequencers.Tissue Antigens. 2012; 80: 305-316Crossref PubMed Scopus (114) Google Scholar, 14Lind C. Ferriola D. Mackiewicz K. Heron S. Rogers M. Slavich L. Walker R. Hsiao T. McLaughlin L. D'Arcy M. Gai X. Goodridge D. Sayer D. Monos D. Next-generation sequencing: the solution for high-resolution, unambiguous human leukocyte antigen typing.Hum Immunol. 2010; 71: 1033-1042Crossref PubMed Scopus (118) Google Scholar, 17Holcomb C.L. Hoglund B. Anderson M.W. Blake L.A. Bohme I. Egholm M. Ferriola D. Gabriel C. Gelber S.E. Goodridge D. Hawbecker S. Klein R. Ladner M. Lind C. Monos D. Pando M.J. Proll J. Sayer D.C. Schmitz-Agheguian G. Simen B.B. Thiele B. Trachtenberg E.A. Tyan D.B. Wassmuth R. White S. Erlich H.A. A multi-site study using high-resolution HLA genotyping by next generation sequencing.Tissue Antigens. 2011; 77: 206-217Crossref PubMed Scopus (99) Google Scholar, 18Lange V. Bohme I. Hofmann J. Lang K. Sauter J. Schone B. Paul P. Albrecht V. Andreas J.M. Baier D.M. Nething J. Ehninger U. Schwarzelt C. Pingel J. Ehninger G. Schmidt A.H. Cost-efficient high-throughput HLA typing by MiSeq amplicon sequencing.BMC Genomics. 2014; 15: 63Crossref PubMed Scopus (180) Google Scholar With the increasing availability of commercial NGS HLA reagents, there is a need to better understand each assay's characteristics and their utility to solve the limitations of SBT. In this study, we evaluated the performance of the TruSight HLA assay, a commercially available NGS assay for HLA typing, to not only accurately type HLA alleles but also identify a set of quality control criteria required to ensure accurate HLA allele determination. More important, we also performed a cost analysis between NGS and SBT for HLA typing that provides the first evidence that NGS is a cost-effective alternative to SBT. Two-hundred and eleven samples that were already high-resolution HLA typed were used for comparison and clinical validation. An additional 79 samples of genomic DNA extracted from buccal swabs from patients and donors under evaluation for HSCT were also used. Genomic DNA was extracted from each sample using Qiagen DNA Tissue Extraction kits (Qiagen, Valencia, CA). The study was approved by the Institutional Review Board of the University of North Carolina at Chapel Hill. For HLA gene amplification, primers specific for each HLA gene were used in a long-range PCR. DNA was quantified using a QuBit fluorometer (Life Technologies, Carlsbad, CA). After quantification, sample DNA was diluted to 10 ng/μL. For buccal swabs, 40 of the 79 samples received additional purification according to the manufacturer's instructions. Fifty nanogram of genomic DNA was used for each HLA locus and amplified according to the manufacturer's instructions. PCR amplicons were visualized using 2% agarose gel electrophoresis before preparing NGS libraries. Twenty-four samples (192 HLA loci) were run in a single NGS experiment. A sample with a known HLA typing was run on each NGS run to ensure library preparation, data quality, and analysis were of sufficient quality to ensure accurate HLA typing. Amplified DNA for NGS sequencing was prepared according to the supplied instructions (Illumina, Inc., San Diego, CA). Briefly, amplified DNA was purified using magnetic AMPure XP beads, fragmented, and Illumina-specific adaptors were applied using a tagmentation enzyme supplied by Illumina. After tagmentation, fragments were purified using AMPure XP beads and patient-specific indices were added to individual HLA loci by a short PCR, followed by magnetic bead purification. Post-barcoding samples were pooled and the libraries quantified using QuBit fluorimeter. The size of HLA libraries was determined using TapeStation Bioanalyzer 2200. To assess sequencer-based errors, a 1% to 5% concentration of 12.5 pmol/L PhiX control (Illumina, San Diego, CA) was spiked into pooled HLA libraries. Pooled HLA libraries and PhiX control were loaded onto the cartridge and 2 × 250 bp sequencing was performed using a regular flow cell on an Illumina MiSeq. Demultiplexing and generation of FASTQ files was performed on the MiSeq system. A total of 34 MiSeq runs were performed and used for quantification of sequencing metrics. Sample analysis was performed using Conexio Assign for TruSight HLA software version 1.0.0.729 supplied by Illumina. Sample consensus sequences were compared to the IMGT/HLA database (version 3.15 to 3.20). A complete HLA genotype was determined by loading FASTQ files into Conexio Assign software. Samples were analyzed for mismatches throughout the entire amplified region of the HLA genes. Mismatches that represented potential novel HLA alleles were noted. Ambiguity resolution was performed by Sanger sequencing with analysis on uTYPE 6.0 (SeCore HLA kit; Life Technologies). The cost analysis between SBT and NGS was performed using the available list price for each reagent necessary for the respective technologies. For NGS, reagents included HLA locus–specific primers, library preparation, and sequencing reagents. For Sanger sequencing, reagents included HLA locus–specific primers, ExoSAP, sequencing reagents (cathode, anode, polymer), sequence-specific oligonucleotides, and group-specific sequencing primers for ambiguity resolution. Ambiguity resolution was added to each technology using laboratory-specific 53% for Sanger sequencing and 3% for NGS. Instrumentation cost was excluded for the analysis. An estimation of time required for each method was determined by averaging the hands-on time from amplification to completed analysis (including ambiguity resolution) for each technologist (n = 5) performing the assay. The average labor time was multiplied by the average hourly rate for HLA technologists at UNC Hospitals. Assay time was defined as the time from PCR to completion of sequencing analysis. Turnaround-time (TAT) was determined for all HSCT-related HLA typings from receiving in laboratory to verification of results in calendar days. Date range for NGS (n = 323) was August to December 2015 and the same time period 1 year prior for Sanger sequencing (n = 324). NGS runs were between 10 and 23 patients per run, with HLA-A, HLA-B, HLA-C, HLA-DRB1, HLA-DQB1, HLA-DQA1, HLA-DPB1, HLA-DPA1 typed. Only HLA-A, HLA-B, HLA-C, HLA-DRB1, HLA-DQB1, HLA-DPB1 were reported clinically. HLA-DQA1 and HLA-DPA1 typing were used for academic purposes. For comparison, a maximum of five patients were run using Sanger sequencing. Significant differences between two proportions were determined by two-tailed probability test from the calculated z-ratio. Significant differences between NGS and Sanger for TAT were calculated using unpaired t-test with Welch's correction. P ≤ 0.05 was considered significant. To enrich HLA genes, genomic DNA was amplified by long-range PCR. A known HLA-typed sample and negative control were used with every PCR to ensure proper HLA amplification and to assess for potential contamination. There were no cases of PCR-based contamination throughout the validation. The PCR fragment length for each HLA gene is shown in Table 1, and a representative gel electrophoresis of HLA amplicons is shown in Figure 1A. Overall, the amplification success rate of HLA genes was 95.0% (2399/2526) (Figure 1B). Only 37 HLA class I genes failed to amplify of 945 amplifications (3.9%) and 31 of the 37 were from buccal swabs. There were a total of 127 incidences of no detectable PCR band on electrophoresis that were subsequently sequenced and HLA typed (Figure 1, C and D). DPA1 and DPB1 were the most difficult HLA genes to amplify, most likely because of their length (Figure 1, B–D). In addition, DPA1 and DPB1 had the lowest HLA typing success overall. Amplification success was significantly lower for DPA1 (P < 0.001), DPB1 (P = 0.001), and DQB1 (P = 0.006) when using DNA isolated from buccal swabs (BSs) compared to DNA from peripheral blood19Duke J.L. Lind C. Mackiewicz K. Ferriola D. Papazoglou A. Derbeneva O. Wallace D. Monos D.S. Towards allele-level human leucocyte antigens genotyping - assessing two next-generation sequencing platforms: Ion Torrent Personal Genome Machine and Illumina MiSeq.Int J Immunogenet. 2015; 42: 346-358Crossref PubMed Scopus (18) Google Scholar (Figure 1B). To determine the reason for decreased HLA amplification rate from BSs, the effect of concentration and quality of DNA on HLA typing was evaluated. Additional purification of BSs significantly increased HLA typing rates for DQB1 (P = 0.048) and DPB1 (P = 0.002) compared to nonpurified samples (Figure 1E). Similarly, BSs with higher concentrations were significantly more often successfully HLA typed for DPA1 (P < 0.001) and DPB1 (P = 0.001) compared to lower concentrations (Figure 1F).Table 1Amplification Size for HLA Genes Using TruSight HLALociSequencer region (kb)A4.1B2.6C4.2DPA110.3DPB19.7DQA17.3DQB17.1DRB1/3/4/54.1HLA, human leucocyte antigen. Open table in a new tab HLA, human leucocyte antigen. NGS library fragments vary in size and DNA fragments >1000 bp are less efficient at cluster generation on the MiSeq than smaller fragments.19Duke J.L. Lind C. Mackiewicz K. Ferriola D. Papazoglou A. Derbeneva O. Wallace D. Monos D.S. Towards allele-level human leucocyte antigens genotyping - assessing two next-generation sequencing platforms: Ion Torrent Personal Genome Machine and Illumina MiSeq.Int J Immunogenet. 2015; 42: 346-358Crossref PubMed Scopus (18) Google Scholar To determine the DNA fragment size generated using TruSight HLA, three (576 loci) pooled HLA libraries were analyzed using a TapeStation Bioanalyzer. The average library fragment size was 1268 bp (95% CI, 856–1680 bp) (Figure 2A). Larger fragment sizes aid in correct phasing of HLA alleles.15Hosomichi K. Jinam T.A. Mitsunaga S. Nakaoka H. Inoue I. Phase-defined complete sequencing of the HLA genes by next-generation sequencing.BMC Genomics. 2013; 14: 355Crossref PubMed Scopus (88) Google Scholar Next, overall read quality and depth of coverage were determined for each HLA locus. The average percentage of reads ≥ Q30 was 96.1% (95% CI, 95.4%–96.7%) for all HLA loci (Figure 2B). The lowest quality reads were consistently observed with HLA-B and HLA-DQB1. The average depth of coverage was 280 (95% CI, 270–289) for all HLA loci (Figure 2C). These results indicate that NGS data generated by TruSight HLA are high quality and core exons within each HLA are covered beyond 250×. To determine the quality of each library preparation and sequencing run, cluster density, percentage of clusters passing filter, percentage of reads ≥ Q30, and error rate were monitored using Illumina's sequencing analysis viewer. The acceptable range for each parameter was determined to be the average plus or minus 2 SDs or using the MiSeq performance specifications provided by the manufacturer. There were two runs in which PhiX was not added to the pooled libraries before sequencing and thus no error rate could be determined. The shaded areas in Figure 3 show the acceptable range for each parameter. For a run to be considered high quality, it must fall within specific ranges for each category. There were two runs that required repeat sequencing because of reagent issues (Figure 3, B and D). The average cluster density and proportion of clusters passing filter were 1071 ± 35 K/mm2 and 88% ± 1%, respectively (Figure 3, A and B). The average percentage of read ≥ Q30 was 80% ± 1% (Figure 3C). The average error rate was 1.65% ± 0.1% (Figure 3D). After run 7, there was a 54.5% reduction in error rate variability compared to the first six runs (Figure 3D). Establishing these criteria is crucial for ensuring library preparation consistency and sequencing performance over time. Clinical validation of HLA typing by NGS involves determination of the assay's accuracy and precision compared to the gold standard, Sanger sequencing. To determine TruSight HLA assay accuracy, 211 samples for which existing genomic DNA and high-resolution HLA typing was known for each locus were used. Fifty of the 211 samples were blinded (all authors were blinded). Forty-six samples were used to assess assay reproducibility. HLA alleles were considered equivalent on the basis of the National Marrow Donor Program HLA reporting criteria. The National Marrow Donor Program requires identification of eight null (nonexpressed) HLA alleles within specific HLA G groups.20National Marrow Donor Program. NMDP policy for HLA confirmatory typing requirements for adult donor and patient HLA confirmatory typing. Number P00079. 2015. Available at https://bioinformatics.bethematchclinical.org/workarea/downloadasset.aspx?id=10528 (accessed March 10, 2016).Google Scholar All eight null HLA alleles could be identified or ruled out as potential HLA alleles using TruSight HLA. Overall, accuracy for the assay was (2954/2961) 99.8% with a precision of 100% (Table 2). Two-hundred and sixty-five unique HLA alleles were accurately identified by TruSight HLA (Supplemental Table S1). Only ambiguities outside of HLA G groups are considered significant because those alter the peptide-binding region of the HLA protein. There were 103 (3.5%) ambiguities of 2961 alleles identified. Of 103 ambiguities, 91 (88.4%) were from DRB1 (26.3% of all DRB1 typings) and 11 (10.7%) were from DRB4 (47.8% of all DRB4 typings) loci (Table 2). Because DRB4 is not commonly part of HLA matching for HSCT, the impact on routine clinical use is minimal. There was one ambiguity in the A locus (0.3%). All remaining HLA loci had no ambiguities. HLA typing by TruSight HLA demonstrated a 93.4% reduction in ambiguities compared to Sanger sequencing.Table 2Accuracy and Ambiguities for TruSight HLA AssayGeneNAllele level mismatch% correctAmbiguitiesAmbiguity rate (%)HLA-A353199.71∗Ambiguity exists between 03:01:01, 11:01:01 pair, and 03:63/11:12.0.3HLA-B353199.700.0HLA-C3530100.000.0HLA-DPA1345199.700.0HLA-DPB13540100.000.0HLA-DQA1346199.700.0HLA-DQB13450100.000.0HLA-DRB1346199.791†Ambiguities exist for the following HLA alleles: 03:01; 03:50, 08:04; 08:59, 15:01; 15:110, 13:01; 13:112, 16:02; 16:22, 14:54; 14:113; 14:125; 14:157, 15:02; 15:19, 13:02; 13:128.26.3HLA-DRB31000100.000.0HLA-DRB4230100.011‡Ambiguities because of inability to accurately identify 01:03N.47.8HLA-DRB543197.700.0Total2961699.81033.5∗ Ambiguity exists between 03:01:01, 11:01:01 pair, and 03:63/11:12.† Ambiguities exist for the following HLA alleles: 03:01; 03:50, 08:04; 08:59, 15:01; 15:110, 13:01; 13:112, 16:02; 16:22, 14:54; 14:113; 14:125; 14:157, 15:02; 15:19, 13:02; 13:128.‡ Ambiguities because of inability to accurately identify 01:03N. Open table in a new tab The seven cases of incorrect HLA typing emphasized the need for HLA-specific data criteria. Three of the seven instances were because of allele dropout. Given those results, HLA-specific data quality criteria were established. A depth of coverage (DOC) of at least 100 and at least 81% of the reads ≥ Q30 were determined to be necessary for HLA-A, HLA-B, HLA-C, HLA-DRB1, HLA-DQB1, HLA-DQA1, HLA-DPB1, HLA-DPA1 typing. Higher quality data (88% of the reads are ≥ Q30) was necessary for HLA-DRB3/4/5 to prevent false identification of those alleles (Figure 2, B and C). The sequencing and HLA-specific data quality metrics were validated using the 50 blinded samples.
Referência(s)