In next-gen sequencing, aiming for wider access

Fiona, as you look at the customer base, do you see movement between in-house solutions for NGS versus the use of send-out laboratories? Or is that at a steady state?
Fiona Nohilly, staff product marketing manager, Americas regional marketing, Illumina: That can vary a lot depending on the physicians or oncologists at the particular hospitals. For example, there may be a sponsor, someone who’s advocating for bringing in new technology because they’ve had exposure to it in their fellowship or residency. So when they go into a community-based hospital, they have the knowledge to advocate for it.

Whether they opt to send out or bring it in-house, there are many variables, and we want to support all options and situations. We have developed analysis tools for all the way from the sequencing FASTQ file to aligning against a reference, variant calling, and interpretation. We’re investing in the interpretation side to help those who may not have the exposure or experience to be able to get the results in a report they can read within seconds.

Jeremy, can you comment on the movement to set up a center internally within a system or hospital versus the send-out alternative, which so many must rely on?
Dr. Segal (University of Chicago): I may be biased, but I think the best thing is to have it done in-house. By doing it in-house we make it a multidisciplinary team effort to diagnose the patient’s tumor, look at the genetics, and correlate it together. Many times questions arise about what we should test or prioritize, and we’ll loop in the oncologists and get feedback from them in real time.

Everyone thinks of NGS as a technology for determining treatment for the patient, but just as important, it’s useful for figuring out the diagnosis. There are a lot of cases in which we don’t know the diagnosis or are surprised by the next-gen sequencing results. Being integrated with our anatomic pathology group means we discuss as a group. If they have a case they don’t know what to do with, they’ll order immunohistochemistry testing, but they may also order our test and we can talk about the results. We make better diagnoses for patients that way, and having our in-house service is a positive for everybody.

Maybe you could get the equivalent if you sent out the testing and had everybody reviewing it and integrating the results together, but that’s not what tends to happen. It tends to be that the oncologist sends the test out, the result goes back to the oncologist, and the pathologists may not have an opportunity to see it and correlate their findings. It needs to be a team-based effort.

We have talked about the importance of a democratization of NGS on the vendors’ side and doing work on the analytics to help make that happen. Are you satisfied, Fiona, with the progress you’re making in simplifying and democratizing NGS?
Fiona Nohilly (Illumina): We are looking at two types of customers. We have folks who are new to NGS or to Illumina technology who need simple solutions. So we have efforts to develop technologies as end-to-end solutions.

We also have customers who are more familiar with our sequencing technology, and they are looking for data analysis platforms that enable them to analyze massive amounts of data, and we’re working on developing solutions for that too. We’re supporting those customers with our newest informatics platform, Illumina Connected Analytics [ICA]. We launched ICA this year fully at scale to customers. It’s been several years in the making, and we acquired companies that helped make the user interface more user-friendly and integrated data compression into our DRAGEN [Dynamic Read Analysis for GENomics] secondary analysis platform. ICA is a scalable and secure cloud-based platform that health care systems like SickKids in Canada have implemented for their COVID-related sequencing projects.

ICA will be the future of our analysis platforms in the cloud, and we’re working on new cloud partners. It’s on AWS now, but it will be scaled out to other cloud partners.

Let us hear the Thermo perspective.
Dr. Qureshi (Thermo Fisher): Since we last spoke, we have launched several products to further our mission of making NGS more accessible to labs with all levels of expertise. We launched the Genexus Purification system, which, combined with the Genexus Integrated Sequencer, allows labs to go from preprocessed samples to NGS report in as little as 24 hours, with two touch points and 10 minutes of hands-on time. Both instruments have built-in technology to minimize errors, which is crucial in clinical settings. We also launched Oncomine Reporter, which enables labs to create reports linking biomarkers to relevant evidence, greatly simplifying bioinformatic analysis.

The last piece is integration into electronic health records. I speak to many in the field who say their EHR is out of date. So there’s work to do on the solution provider’s end, but also in hospitals and other care settings.

Jeremy, do you want to speak to the EHR at the University of Chicago?
Dr. Segal (University of Chicago): We’re in the process of upgrading to Epic Beaker for our whole laboratory. They claim it can do everything but we’ll have to see what the reality is when we install it.

That’s an important question because in no one’s wildest imagination was the EHR supposed to be able to display complex results like NGS.
Dr. Segal (University of Chicago): Our internally developed NGS lab information system covers our upfront workload, what sample needs what, the pooling, et cetera, and it also includes our variant interpretation platform. The variant interpretation part of the system will have to be retained because I don’t believe Beaker can replace it. Hopefully Beaker will do a good job with the upfront accessioning and laboratory process tracking.

Pierre, how is artificial intelligence affecting your life and work now?
Dr. Del Moral (Illumina): AI is a bigger component of everyday life. When we’re thinking beyond traditional care in oncology, multiomics, and of additional signatures that could be involved, artificial intelligence is taking a greater part in data aggregation and insight, which our Connected Analytics platform provides. Pharmaceutical companies, within their clinical trial design, are looking and asking for more of that capability.

Sohaib, can you speak to that from the Thermo point of view?
Dr. Qureshi: I don’t think you can live a day without some interaction with AI, whether you know it or you don’t. That mindset will need to be applied to NGS. There’s a lot of development at Thermo Fisher Scientific in terms of leveraging the power of NGS. It’s not just for data aggregation; it helps improve the overall sequencing quality. There are a lot of new entrants in the field and they are going to be applying AI technology to improve their platform from an aggregation perspective and with overall data quality.

Jeremy, can you give us your take on artificial intelligence today?
Dr. Segal (University of Chicago): It has a ton of potential for the future. It’s already being used to some degree; for example, many of the algorithms for refinement of basic sequencing quality are AI-based. But you could also imagine AI-based detection systems for a variety of genomic applications, such as assessments for homologous recombination or other biomarkers like methylation pattern recognition.

I do have concerns about how we should go about validating something like that. With our normal pipeline, we like to think we know what it’s doing under the hood. However, I must say I don’t really understand all the steps of the Burrows-Wheeler Alignment algorithm. But at least it’s a program that someone sat down and wrote. The AI algorithms, on the other hand, they weren’t written but instead evolved, and you don’t really know what they are looking at or doing. This raises concerns about how we get comfortable to put our medical seal of approval on the results of AI-based algorithms.

I think we’ll ultimately find that AI-based algorithms are better for many aspects of sequencing analytics than our current methods, but we will have to figure out how to develop the highest level of confidence in them, and that may only come through exhaustive testing.

You worry a lot in pathology, and pharmaceutical companies do too, about getting adequate samples to work up cancer patients from the beginning through NGS analysis, which includes large gene panels, because of the limitations of materials. Where are we in terms of understanding best practices for procuring tissues, sampling, doling it out? Jeremy, what are you seeing at the University of Chicago?
Dr. Segal (University of Chicago): We do a lot to maximize what we can get from any sample. Most of the samples from lung cancer patients are small, so we do a few things. We do a lot of endobronchial-guided ultrasound procedures, and we’ve found value in focusing on fine-needle aspirate material rather than biopsy material.

In the early days of our laboratory, our rapid on-site evaluation team would review smears for basic diagnostic adequacy. Then the focus of the procedure would shift to biopsies or additional FNA passes for cell blocks, and we would focus on these specimen types for our testing. But we found these specimens to be quite variable, and if they were inadequate we would be back to square one. So we shifted our focus to doing repeated FNA passes and having the rapid on-site evaluation team assess them for molecular adequacy. The smears work great, and now we use those smears for more than half of all our lung cancer testing. The great thing is we get a determination of molecular adequacy up front, so we know we’ll be able to do the testing successfully before they finish the procedure with the patient.

We’ve changed the way some of our other procedures are done to try to maximize tissue. We’ve set up reflex systems, especially for lung, so as soon as the tissue hits cytology or thoracic pathology, they can put the orders through for our tests. That means when they order recuts for immunohistochemistry, they also order recuts for us at the same time. So we only have the specimen going on the microtome once instead of multiple times, and that saves a lot of tissue.

On the laboratory side, if we have a tiny FFPE specimen, we might stage the extraction. For example, we might take half the slides and do a DNA extraction to see if it’s enough to run our DNA base panel. If it is, we’ll spend the rest of the slides and get RNA for our fusion panel. If it’s not, we have to decide and maybe get the oncologist to weigh in—how much do you prefer this test over another test? This helps maximize our testing throughput and helps focus our testing on the biomarkers most important to our oncology team.

Pierre, what are your thoughts on this? In the early days of NGS there was a lot of concern about adequacy of sample.
Dr. Del Moral (Illumina): There always is. What Jeremy said is right, and it showcases the need for in-house testing, because the oncologist-pathologist relationship is crucial to perform sample shepherding and maximize the sample. We’ve seen with some of our customers that reflex comprehensive genomic profiling had significantly lowered the rate of nonbiomarker informed care. That was enabled by the oncologist and the pathologist discussing it, by automating some of those processes, and by maximizing the tissue and having the ability to test for all variants and signatures using a single panel.

The quality of the NGS reads is also important. We’ve implemented DRAGEN pipelines for tissue and liquid biopsy. We participated in an FDA challenge assessing bioinformatics pipeline accuracy. When NGS reads from Illumina were combined with the DRAGEN pipeline, it yielded the industry’s most accurate results. So one side is the oncologist-pathologist relationship and the other side is how accurately the technology can translate tissue information into genetic information.

Sohaib, please speak to that same question.
Dr. Qureshi (Thermo Fisher): As Jeremy noted, biopsy samples can be quite small, particularly for lung cancer. This can manifest in limited surface area or tumor content, both of which can limit access to sufficient nucleic acid for analysis. This creates a challenge not only for sequential or reflex testing but also for parallel testing.

Sample stewardship has become an important consideration for clinical labs as the demand for biomarker testing and genomic profiling has increased with approved targeted therapies. We have been able to help labs with sample stewardship by providing technology that requires minimal sample input. A key feature of the AmpliSeq chemistry, which is at the core of our NGS technology, allows for minimal DNA and/or RNA to be interrogated using short targeted sequences, which affords greater depth of coverage and high accuracy. With as little as 10 ng of nucleic acid, labs can generate genomic profiling results for both DNA and RNA analysis. This significantly increases success rate, hence reducing quantity not sufficient, or QNS, results and the need for rebiopsy.

Andy, in today’s discussion we have a health care provider and platform and software people. Does it help clarify what the dilemmas are for Janssen, which offers targeted therapies?
Dr. Johnson (Janssen): It does. In the end, it all comes down to one patient’s tumor. Regardless of whether that patient is treated at an academic center or a small community practice, our goal is that they receive comprehensive biomarker testing and the appropriate therapy. I like Jeremy’s approach, which I frame as: Begin with the end in mind. With reflex testing in place, the diagnostic team knows what “perfect” looks like but can pivot, for instance, to liquid biopsy because there was insufficient tissue.

It’s not just the technology though; it’s the overall operational approach, the thinking from everybody involved to go from beginning to end. We want to get patients the appropriate therapy. We have good data from the real world and clinical trials to show that getting targeted therapy often provides the best outcome for patients when they have a driver mutation in lung cancer. That’s the goal.

Jeremy, as our health care provider here, do you have a final thought about our conversation?
Dr. Segal (University of Chicago): Keep the focus on the patients and try to figure out at every step what they need and what is the right thing to do for them. That will help us make the right decisions, whether it’s about setting up tests in the lab, or how to interpret tests, or how to regulate tests. We have to think about their needs first.