Almost five years after first exploring next-gen sequencing for clinical HLA testing, the Red Cross Blood Transfusion Service of Upper Austria has won accreditation for an HLA typing method using Roche's 454 GS Junior, and is the first lab to begin routinely using the technology to test clinical samples for that application.
Collaborating with Roche, the team spent several years working to automate pre-sequencing steps and optimize a bioinformatics strategy that would make the 454 method both accurate and cost-effective enough to replace Sanger sequencing, the blood service's medical director, Christian Gabriel, told Clinical Sequencing News.
The work paid off, with accreditation at the end of last year from the European Federation for Immunogenetics — an approval necessary for testing reimbursement within the Austrian health system, Gabriel said.
"At the beginning of last year we started with our final validation on patient samples," said Gabriel. "We validated on 172 patients who had all had their Sanger sequencing done and had additional Sanger sequencing of siblings and parents, so we had a full construction on every patient of what HLA type he or she had."
"We went through with the 454 and got really very nice results, so we put this together, and got the accreditation process going in June, and received the accreditation in December last year … It is a must in Austria if you want to get reimbursement, which is a must if you want to do transplantation. If you are not EFI accredited, no one would touch you," he said.
The Upper Austria blood service is one of several groups that have been exploring the potential of 454 sequencing for use in clinical HLA testing. Roche has also developed its own HLA kits for research purposes (CSN 4/5/2011).
At the beginning of these efforts, several US investigators expressed skepticism about how the method could become cost-competitive with Sanger sequencing, especially considering its pre-sequencing sample preparation requirements. They also cited read lengths that were not long enough to phase polymorphisms across exons, and the need to pool many samples in a run, as hindrances to clinical use (CSN 3/17/2009).
Since then, 454 read lengths have improved with the introduction of Titanium chemistry to around 400 bases. And with the introduction of the GS Junior, smaller numbers of samples can be processed in one run. Researchers working with the system, like the Upper Austrian blood service, have also worked with Roche to adopt various automation instruments for sample preparation.
Nezih Cereb, president of HLA-typing firm HistoGenetics, who expressed doubt several years ago that 454 could replace his lab's capillary sequencing approach, told CSN this week that NGS methods still can't scale up to the throughput necessary for the volume of samples his lab goes through.
"Yes, you can do it, and get accredited, he said. But [no next-gen sequencing instrument] is in a position to replace what we can do – to do 5000 HLA types a day. The scalability question is still there," he said.
According to Gabriel, the biggest challenges for the Austrian group's effort were in building a robust bioinformatics strategy and in solving the problem of the complexity of manual sample preparation.
On the bioinformatics side, soon after the group began working with the 454, Roche partnered with Conexio Genomics, an Australian software company, to adapt its HLA-genotyping software for Sanger sequencing to analyze 454 data, which Gabriel said gave the group a huge boost.
The blood center team also collaborated with scientists at the University of Applied Sciences Upper Austria in Hagenberg to establish a high-throughput computer cluster to provide the hardware to support its Conexio-based software solution, Gabriel said.
The next challenge was sample preparation. "We realized that if we wanted to do high-resolution HLA typing, which is a requirement for stem cell transplantation, we came across this huge manual workload," he said. "We calculated at the time that if we wanted to type 30 patients at once at high resolution on 17 exons, then it would require 14,000 different pipetting steps as a preload for that."
"So we went to Roche," he said, " and we told them, we know that you are putting your nose into the diagnostics market, and there is no way to use [454 for diagnostics] if there is no automation," Gabriel said.
"Roche came up with a very good idea and connected us with Hamilton. Now instead of five days with two persons we have one person working for two hours. It's really reduced [pre-sequencing work] to nil," he said.
"We have also scaled down the whole procedure to the GS Junior," Gabriel added, "which gives us a little more flexibility to move samples around – to speed up some samples from the transplantation unit [if we have to,] for example."
According to Gabriel, without the advantages conferred by the Conexio software and the Hamilton hardware, the 454 next-gen sequencing approach would not be usable for routine clinical testing.
"It only got cost-effective as we kicked in the bioinformatics and the automation," he said. "Now we can test 10 samples at once in a very good, very cost-effective manner: much cheaper than Sanger, much more exactly, and at a very high resolution."
Gabriel said his group does not use the HLA assay kits Roche has commercialized. Like other groups who have worked with Roche in testing the 454 for potential clinical use — such as Matthew Anderson, assistant director of the Histocompatibility, Immunogenetics, and Disease Profiling Laboratory at Stanford University, who has said his group is also planning to validate a method using the GS Junior for clinical HLA typing (CSN 4/18/2012) — the Austrian lab has designed its own primers.
"Our first approach was just to do a copy-paste of Sanger, so just exon 2 and 3 of HLA class 1 and 2," Gabriel said. "But we then expanded it – so we are sequencing exon 7 on C, or we have also expanded the region we are sequencing to also get the attached intron sequences."
Roche did not comment on whether it is planning to pursue releasing its own HLA kits for clinical use. The company told CSN in an email that researchers are currently working on an "improved, simplified workflow for the GS Gtype HLA MR/HR Primer Sets that significantly cuts down the manual hands-on time … a third primer plate that will complement the GS Gtype HLA MR/HR Primer Sets … and an automation solution for the HLA high-throughput workflow using Hamilton instrumentation."
While 454 sequencing seems to have garnered a lion's share of attention for potential clinical HLA work, research groups have also explored competing sequencing platforms in the space.
Anderson's Stanford team is also testing Ion Torrent's PGM in parallel with its work with the GS Junior. Other Stanford researchers have also published a method, still in early research, for HLA genotyping using the Illumina HiSeq and MiSeq platforms (CSN 5/23/2012).
Gabriel said his team also considered platforms like Illumina and Ion Torrent.
"What we saw was [that] misreading and misinterpretation was so high you really couldn’t use [them] in routine use, especially for HLA," he said. "HLA is highly polymorphic, so if there is a misread in one spot, it could come out as a different HLA type," he explained.
"For our strategy, we need to run through long reads and have a clear discrimination between high quality reads and non-quality reads."
However, he said, the Upper Austria service also has other work, in hematology and oncology for example, for which other sequencing technologies might be a better fit than 454. "What I foresee is that labs will have to have different machines running for different purposes," he said.