In 2024, University Medical Center Utrecht introduced a 523-gene next-generation sequencing (NGS) panel into its molecular pathology workflow, opening the door to comprehensive genomic profiling (CGP). The experience proved successful enough for the laboratory to take the next step: adopting CGP across its oncology testing.
Following her presentation at Illumina’s precision oncology symposium at the European Congress of Pathology (ECP) 2026, The Pathologist caught up with clinical molecular biologist Wendy de Leng to explore what drove the transition. Here, she shares how the laboratory made CGP work in routine practice, and reflects on what it will take to make comprehensive testing accessible to every patient who could benefit.
Take us back to the era when molecular diagnostics largely meant testing one gene or biomarker at a time. What did targeted NGS enable, and where did its limitations become apparent?
We started with single-gene testing around the early 2000s. That allowed us to identify specific therapeutic targets for individual patients. In lung cancer, for example, we could test for an EGFR mutation and, if present, determine whether the patient might benefit from an EGFR inhibitor.
In my laboratory, we moved to targeted next-generation sequencing in 2013. That was a major step forward because, instead of looking only at EGFR, we could examine multiple relevant genes at the same time. We could also identify mutations in genes such as KRAS, which at that time was not targetable but could still inform treatment decisions. If a KRAS mutation was present, for example, we knew an EGFR mutation was unlikely to be present, which helped guide the clinical pathway.
However, targeted NGS still had limitations. We were looking mainly at known hotspots across about 50 genes. So, we had effectively moved from hotspots in a single gene to hotspots in 50 genes.
In 2024, we transitioned to CGP, covering 523 genes and, importantly, analyzing the full genes rather than only known hotspots. That means we can identify clinically relevant alterations that more targeted approaches may miss. For a small proportion of patients, this can reveal additional treatment options.
So, for me, the evolution from one gene, to 50 genes, to more than 500 genes has ultimately been about providing a more complete picture of the tumor and creating more opportunities to guide patient care.
How has that switch to CGP changed the information your laboratory provides to clinicians for a single patient sample?
We generate a lot more data now, so the challenge is making sure that data is analyzed correctly and reliably, but also reported in a way that is clear and useful to the clinician. When we were testing only EGFR, the answer was essentially yes or no. That was easy. When you are sequencing more than 500 genes, you obviously cannot give the clinician 500 yes-or-no answers. So we discussed with our clinicians what they actually wanted us to report.
Although our lung cancer assay covers more than 500 genes, we currently analyze around 100 that are clinically relevant. That number will evolve as the evidence changes; perhaps next year it will be 120. Within those genes, we report the pathogenic and likely pathogenic variants we identify. In other words, we focus on alterations that change protein function and for which there is evidence of clinical relevance. Usually, that produces a relatively short list, which makes the report manageable for the clinician.
Of course, when you sequence this broadly, you also find many variants of uncertain significance. We generally choose not to report those because, by definition, we do not yet know what they mean, and a long list of uncertain findings risks confusing rather than helping the clinician.
There are exceptions. If we find a variant of uncertain significance in a particularly important gene – EGFR in lung cancer, for example – we may report it while making its uncertain significance very clear. We might not know what that variant means today, but in one or two years the evidence could have changed. In that situation, we think there can be value in having that information available for the future.
On a practical level, how has the laboratory workflow had to evolve to support the shift toward CGP?
For our laboratory, it meant completely changing the workflow. Our targeted NGS and CGP used different library preparation methods and different sequencing platforms. In 2024, we were still using targeted NGS for roughly 90 percent of patients and CGP for perhaps 5–10 percent, so we effectively had two separate workflows running in parallel.
That was interesting for the technicians because they enjoyed the variety, but operationally it was very inefficient. Every time we received a request, we had to decide: should this patient have targeted sequencing or comprehensive profiling? We spent a surprising amount of time having those discussions.
That was one of the reasons we decided to move toward a single CGP workflow. Now there is one way to request the test and one laboratory process because, essentially, every patient goes through the same assay. It also simplifies quality management. Every test we offer has to meet our ISO 15189 requirements and undergo the appropriate external quality assessment. Maintaining all of that for two workflows requires considerably more work than maintaining one.
What were the economic effects of switching to a CGP-only workflow?
Of course, we also had to consider the finances. If you look at a single patient, CGP is more expensive than targeted NGS. But once you start testing larger numbers of patients, the economics change because the workflow becomes much more efficient. Previously, we might run CGP for five patients at a time; now we can run around 30. From the technician’s perspective, processing 30 samples does not take six times as long as processing five, so the cost per patient falls considerably.
Sequencing costs have also decreased, and collaboration has helped us use the technology more efficiently. We work with our clinical genetics department, which operates high-capacity NovaSeq instruments. We use one lane of a flow cell for our samples while they use the remaining lanes for their work.
All those factors came together to make CGP for all feasible. Ultimately, however good the technology is, if you cannot make it financially sustainable, you cannot implement it in routine practice.
How has the switch to CGP affected turnaround times?
It has actually improved turnaround times for CGP requests considerably. In 2024, when we were running targeted NGS and CGP in parallel, the turnaround time for CGP was around six to ten days. Ten days is quite a long time when a patient is waiting for a result.
Now that we perform CGP for all patients, we can run the assay twice a week. That higher testing frequency has reduced the turnaround time. So, although we are generating much more genomic information for each patient, we are actually able to provide the results faster.
Looking beyond your own institution, which patients are still most likely to miss out on comprehensive genomic testing today?
From a technical perspective, CGP can be more challenging when you have older tissue, a very small amount of tissue, or a sample with a low percentage of tumor cells. But challenging does not necessarily mean impossible. In our laboratory, fewer than 1 percent of patients cannot be tested with our CGP assay, so technical failure is actually very uncommon.
Access is a different question. In the Netherlands, we have agreed nationally that every patient should have access to the molecular diagnostics they need. A laboratory does not necessarily have to perform CGP itself; the sample can be sent to another laboratory that can. Because the Netherlands is relatively small and laboratories work closely together, that referral pathway works quite well.
But somebody still has to recognize that comprehensive testing is needed. The clinician has to consider whether CGP could provide useful information for that patient and request it or refer the sample appropriately.
Then there is reimbursement. Even when the testing is available and technically possible, the question becomes: who is going to pay for it? I think inadequate reimbursement remains one of the important reasons why patients can miss out on CGP.
What advice can you offer on reporting CGP results?
I would always recommend using a virtual gene panel. Just because you sequence more than 500 genes does not mean you have to analyze all of them. Focus on the genes that are clinically relevant for that particular tumor type. It saves a lot of time and resources.
I know not everybody agrees with me on that. From a research perspective, of course it is fascinating to analyze all 500 genes. But in a hospital, we have limited resources, and our time is also healthcare money. If we spend time analyzing findings that have no clinical relevance, are we really using those resources in the best way? My view is: analyze what matters for the patient.
The other important thing is to talk to your clinicians. Don't simply assume they are happy with the report you give them. Ask them: Does this tell you what you need to know?
For example, if we identify an EGFR mutation in lung cancer, does the clinician just want us to report the mutation, or do they want us to add that it may indicate eligibility for treatment with an EGFR tyrosine kinase inhibitor? There is still debate about how much clinical interpretation a molecular pathology report should provide, and different hospitals take different approaches.
So, my advice is very simple: talk to each other. Have a coffee or a tea and discuss what clinicians actually need from the laboratory. That communication is really important, and I think we sometimes forget to do it.
Looking ahead to a world where CGP becomes a standard part of the diagnostic pathway, how might that change the role of the clinical molecular laboratory?
I think automation and AI will change a lot. If I look at my own job, for example, I currently spend a significant amount of time on variant interpretation. AI is not good enough to take that over at the moment, but I can imagine it playing a much bigger role within the next few years.
The laboratory itself will probably become more automated too. In my lab, we currently perform around 15 different tests. I expect that number to decrease as we move toward broader assays that can answer more questions at once. We will do more of the same thing – and hopefully become very good at doing it. There may also be opportunities for laboratories to collaborate more and consolidate some testing, although that raises its own practical questions.
For clinical molecular biologists, I hope these changes will free us to spend more time thinking alongside pathologists and other clinicians. Sometimes my colleagues and I will look at the genomic data from a case and see a pattern that suggests there is something else going on. We might contact the pathologist, discuss the findings together, and realize that another test could help us reach the diagnosis.
For me, that would be a really exciting development – moving away from spending so much time getting everything sorted out in the laboratory and toward working more closely with doctors to help solve the difficult cases.
